API tools faq
paste
Guest User

Untitled

a guest
Aug 21st, 2018
225
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 382.09 KB | None | 0 0
raw download clone embed print report
  1. #!/usr/bin/env perl
  2. # -*-mode:cperl; indent-tabs-mode: nil-*-
  3.  
  4. ## Perform many different checks against Postgres databases.
  5. ## Designed primarily as a Nagios script.
  6. ## Run with --help for a summary.
  7. ##
  8. ## Greg Sabino Mullane <greg@endpoint.com>
  9. ## End Point Corporation http://www.endpoint.com/
  10. ## BSD licensed, see complete license at bottom of this script
  11. ## The latest version can be found at:
  12. ## http://www.bucardo.org/check_postgres/
  13. ##
  14. ## See the HISTORY section for other contributors
  15.  
  16. package check_postgres;
  17.  
  18. use 5.006001;
  19. use strict;
  20. use warnings;
  21. use utf8;
  22. use Getopt::Long qw/GetOptions/;
  23. Getopt::Long::Configure(qw/ no_ignore_case pass_through /);
  24. use File::Basename qw/basename/;
  25. use File::Temp qw/tempfile tempdir/;
  26. File::Temp->safe_level( File::Temp::MEDIUM );
  27. use Cwd;
  28. use Data::Dumper qw/Dumper/;
  29. $Data::Dumper::Varname = 'POSTGRES';
  30. $Data::Dumper::Indent = 2;
  31. $Data::Dumper::Useqq = 1;
  32.  
  33. binmode STDOUT, ':utf8';
  34.  
  35. our $VERSION = '2.20.0';
  36.  
  37. use vars qw/ %opt $PGBINDIR $PSQL $res $COM $SQL $db /;
  38.  
  39. ## Which user to connect as if --dbuser is not given
  40. $opt{defaultuser} = 'postgres';
  41.  
  42. ## Which port to connect to if --dbport is not given
  43. $opt{defaultport} = 5432;
  44.  
  45. ## What type of output to use by default
  46. our $DEFAULT_OUTPUT = 'nagios';
  47.  
  48. ## If psql binaries are not in your path, it is recommended to hardcode it here,
  49. ## as an alternative to the --PGBINDIR option
  50. $PGBINDIR = '';
  51.  
  52. ## If this is true, $opt{PSQL} and $opt{PGBINDIR} are disabled for security reasons
  53. our $NO_PSQL_OPTION = 1;
  54.  
  55. ## If true, we show how long each query took by default. Requires Time::HiRes to be installed.
  56. $opt{showtime} = 1;
  57.  
  58. ## If true, we show "after the pipe" statistics
  59. $opt{showperf} = 1;
  60.  
  61. ## Default time display format, used for last_vacuum and last_analyze
  62. our $SHOWTIME = 'HH24:MI FMMonth DD, YYYY';
  63.  
  64. ## Always prepend 'postgres_' to the name of the service in the output string
  65. our $FANCYNAME = 1;
  66.  
  67. ## Change the service name to uppercase
  68. our $YELLNAME = 1;
  69.  
  70. ## Preferred order of ways to fetch pages for new_version checks
  71. our $get_method_timeout = 30;
  72. our @get_methods = (
  73. "GET -t $get_method_timeout -H 'Pragma: no-cache'",
  74. "wget --quiet --timeout=$get_method_timeout --no-cache -O -",
  75. "curl --silent --max-time=$get_method_timeout -H 'Pragma: no-cache'",
  76. "fetch -q -T $get_method_timeout -o -",
  77. "lynx --connect-timeout=$get_method_timeout --dump",
  78. 'links -dump',
  79. );
  80.  
  81. ## Nothing below this line should need to be changed for normal usage.
  82. ## If you do find yourself needing to change something,
  83. ## please email the author as it probably indicates something
  84. ## that could be made into a command-line option or moved above.
  85.  
  86. ## Messages. Translations always welcome
  87. ## Items without a leading tab still need translating
  88. ## no critic (RequireInterpolationOfMetachars)
  89. our %msg = (
  90. 'en' => {
  91. 'address' => q{address},
  92. 'age' => q{age},
  93. 'backends-fatal' => q{Could not connect: too many connections},
  94. 'backends-mrtg' => q{DB=$1 Max connections=$2},
  95. 'backends-msg' => q{$1 of $2 connections ($3%)},
  96. 'backends-nomax' => q{Could not determine max_connections},
  97. 'backends-oknone' => q{No connections},
  98. 'backends-po' => q{sorry, too many clients already},
  99. 'backends-users' => q{$1 for number of users must be a number or percentage},
  100. 'bloat-index' => q{(db $1) index $2 rows:$3 pages:$4 shouldbe:$5 ($6X) wasted bytes:$7 ($8)},
  101. 'bloat-nomin' => q{no relations meet the minimum bloat criteria},
  102. 'bloat-table' => q{(db $1) table $2.$3 rows:$4 pages:$5 shouldbe:$6 ($7X) wasted size:$8 ($9)},
  103. 'bug-report' => q{Please report these details to check_postgres@bucardo.org:},
  104. 'checkcluster-id' => q{Database system identifier:},
  105. 'checkcluster-msg' => q{cluster_id: $1},
  106. 'checkcluster-nomrtg'=> q{Must provide a number via the --mrtg option},
  107. 'checkmode-prod' => q{in production},
  108. 'checkmode-recovery' => q{in archive recovery},
  109. 'checkmode-state' => q{Database cluster state:},
  110. 'checkpoint-baddir' => q{Invalid data_directory: "$1"},
  111. 'checkpoint-baddir2' => q{pg_controldata could not read the given data directory: "$1"},
  112. 'checkpoint-badver' => q{Failed to run pg_controldata - probably the wrong version ($1)},
  113. 'checkpoint-badver2' => q{Failed to run pg_controldata - is it the correct version?},
  114. 'checkpoint-nodir' => q{Must supply a --datadir argument or set the PGDATA environment variable},
  115. 'checkpoint-nodp' => q{Must install the Perl module Date::Parse to use the checkpoint action},
  116. 'checkpoint-noparse' => q{Unable to parse pg_controldata output: "$1"},
  117. 'checkpoint-noregex' => q{Unable to find the regex for this check},
  118. 'checkpoint-nosys' => q{Could not call pg_controldata: $1},
  119. 'checkpoint-ok' => q{Last checkpoint was 1 second ago},
  120. 'checkpoint-ok2' => q{Last checkpoint was $1 seconds ago},
  121. 'checkpoint-po' => q{Time of latest checkpoint:},
  122. 'checksum-msg' => q{checksum: $1},
  123. 'checksum-nomd' => q{Must install the Perl module Digest::MD5 to use the checksum action},
  124. 'checksum-nomrtg' => q{Must provide a checksum via the --mrtg option},
  125. 'custom-invalid' => q{Invalid format returned by custom query},
  126. 'custom-norows' => q{No rows returned},
  127. 'custom-nostring' => q{Must provide a query string},
  128. 'database' => q{database},
  129. 'dbsize-version' => q{Target database must be version 8.1 or higher to run the database_size action},
  130. 'depr-pgcontroldata' => q{PGCONTROLDATA is deprecated, use PGBINDIR instead.},
  131. 'die-action-version' => q{Cannot run "$1": server version must be >= $2, but is $3},
  132. 'die-badtime' => q{Value for '$1' must be a valid time. Examples: -$2 1s -$2 "10 minutes"},
  133. 'die-badversion' => q{Invalid version string: $1},
  134. 'die-noset' => q{Cannot run "$1" $2 is not set to on},
  135. 'die-nosetting' => q{Could not fetch setting '$1'},
  136. 'diskspace-fail' => q{Invalid result from command "$1": $2},
  137. 'diskspace-msg' => q{FS $1 mounted on $2 is using $3 of $4 ($5%)},
  138. 'diskspace-nodata' => q{Could not determine data_directory: are you connecting as a superuser?},
  139. 'diskspace-nodf' => q{Could not find required executable /bin/df},
  140. 'diskspace-nodir' => q{Could not find data directory "$1"},
  141. 'file-noclose' => q{Could not close $1: $2},
  142. 'files' => q{files},
  143. 'fsm-page-highver' => q{Cannot check fsm_pages on servers version 8.4 or greater},
  144. 'fsm-page-msg' => q{fsm page slots used: $1 of $2 ($3%)},
  145. 'fsm-rel-highver' => q{Cannot check fsm_relations on servers version 8.4 or greater},
  146. 'fsm-rel-msg' => q{fsm relations used: $1 of $2 ($3%)},
  147. 'hs-no-role' => q{Not a master/slave couple},
  148. 'hs-no-location' => q{Could not get current xlog location on $1},
  149. 'hs-receive-delay' => q{receive-delay},
  150. 'hs-replay-delay' => q{replay_delay},
  151. 'index' => q{Index},
  152. 'invalid-option' => q{Invalid option},
  153. 'invalid-query' => q{Invalid query returned: $1},
  154. 'language' => q{Language},
  155. 'listener-msg' => q{listeners found: $1},
  156. 'listening' => q{listening},
  157. 'locks-msg' => q{total "$1" locks: $2},
  158. 'locks-msg2' => q{total locks: $1},
  159. 'logfile-bad' => q{Invalid logfile "$1"},
  160. 'logfile-debug' => q{Final logfile: $1},
  161. 'logfile-dne' => q{logfile $1 does not exist!},
  162. 'logfile-fail' => q{fails logging to: $1},
  163. 'logfile-ok' => q{logs to: $1},
  164. 'logfile-openfail' => q{logfile "$1" failed to open: $2},
  165. 'logfile-opt-bad' => q{Invalid logfile option},
  166. 'logfile-seekfail' => q{Seek on $1 failed: $2},
  167. 'logfile-stderr' => q{Logfile output has been redirected to stderr: please provide a filename},
  168. 'logfile-syslog' => q{Database is using syslog, please specify path with --logfile option (fac=$1)},
  169. 'mode-standby' => q{Server in standby mode},
  170. 'mode' => q{mode},
  171. 'mrtg-fail' => q{Action $1 failed: $2},
  172. 'new-ver-nocver' => q{Could not download version information for $1},
  173. 'new-ver-badver' => q{Could not parse version information for $1},
  174. 'new-ver-dev' => q{Cannot compare versions on development versions: you have $1 version $2},
  175. 'new-ver-nolver' => q{Could not determine local version information for $1},
  176. 'new-ver-ok' => q{Version $1 is the latest for $2},
  177. 'new-ver-warn' => q{Please upgrade to version $1 of $2. You are running $3},
  178. 'new-ver-tt' => q{Your version of $1 ($2) appears to be ahead of the current release! ($3)},
  179. 'no-db' => q{No databases},
  180. 'no-match-db' => q{No matching databases found due to exclusion/inclusion options},
  181. 'no-match-fs' => q{No matching file systems found due to exclusion/inclusion options},
  182. 'no-match-rel' => q{No matching relations found due to exclusion/inclusion options},
  183. 'no-match-set' => q{No matching settings found due to exclusion/inclusion options},
  184. 'no-match-table' => q{No matching tables found due to exclusion/inclusion options},
  185. 'no-match-user' => q{No matching entries found due to user exclusion/inclusion options},
  186. 'no-parse-psql' => q{Could not parse psql output!},
  187. 'no-time-hires' => q{Cannot find Time::HiRes, needed if 'showtime' is true},
  188. 'opt-output-invalid' => q{Invalid output: must be 'nagios' or 'mrtg' or 'simple' or 'cacti'},
  189. 'opt-psql-badpath' => q{Invalid psql argument: must be full path to a file named psql},
  190. 'opt-psql-noexec' => q{The file "$1" does not appear to be executable},
  191. 'opt-psql-noexist' => q{Cannot find given psql executable: $1},
  192. 'opt-psql-nofind' => q{Could not find a suitable psql executable},
  193. 'opt-psql-nover' => q{Could not determine psql version},
  194. 'opt-psql-restrict' => q{Cannot use the --PGBINDIR or --PSQL option when NO_PSQL_OPTION is on},
  195. 'pgagent-jobs-ok' => q{No failed jobs},
  196. 'pgbouncer-pool' => q{Pool=$1 $2=$3},
  197. 'pgb-backends-mrtg' => q{DB=$1 Max connections=$2},
  198. 'pgb-backends-msg' => q{$1 of $2 connections ($3%)},
  199. 'pgb-backends-none' => q{No connections},
  200. 'pgb-backends-users' => q{$1 for number of users must be a number or percentage},
  201. 'PID' => q{PID},
  202. 'port' => q{port},
  203. 'preptxn-none' => q{No prepared transactions found},
  204. 'psa-disabled' => q{No queries - is stats_command_string or track_activities off?},
  205. 'psa-noexact' => q{Unknown error},
  206. 'psa-nosuper' => q{No matches - please run as a superuser},
  207. 'qtime-count-msg' => q{Total queries: $1},
  208. 'qtime-count-none' => q{not more than $1 queries},
  209. 'qtime-for-msg' => q{$1 queries longer than $2s, longest: $3s$4 $5},
  210. 'qtime-msg' => q{longest query: $1s$2 $3},
  211. 'qtime-none' => q{no queries},
  212. 'queries' => q{queries},
  213. 'query-time' => q{query_time},
  214. 'range-badcs' => q{Invalid '$1' option: must be a checksum},
  215. 'range-badlock' => q{Invalid '$1' option: must be number of locks, or "type1=#:type2=#"},
  216. 'range-badpercent' => q{Invalid '$1' option: must be a percentage},
  217. 'range-badpercsize' => q{Invalid '$1' option: must be a size or a percentage},
  218. 'range-badsize' => q{Invalid size for '$1' option},
  219. 'range-badtype' => q{validate_range called with unknown type '$1'},
  220. 'range-badversion' => q{Invalid string for '$1' option: $2},
  221. 'range-cactionly' => q{This action is for cacti use only and takes no warning or critical arguments},
  222. 'range-int' => q{Invalid argument for '$1' option: must be an integer},
  223. 'range-int-pos' => q{Invalid argument for '$1' option: must be a positive integer},
  224. 'range-neg-percent' => q{Cannot specify a negative percentage!},
  225. 'range-none' => q{No warning or critical options are needed},
  226. 'range-noopt-both' => q{Must provide both 'warning' and 'critical' options},
  227. 'range-noopt-one' => q{Must provide a 'warning' or 'critical' option},
  228. 'range-noopt-only' => q{Can only provide 'warning' OR 'critical' option},
  229. 'range-noopt-orboth' => q{Must provide a 'warning' option, a 'critical' option, or both},
  230. 'range-noopt-size' => q{Must provide a warning and/or critical size},
  231. 'range-nosize' => q{Must provide a warning and/or critical size},
  232. 'range-notime' => q{Must provide a warning and/or critical time},
  233. 'range-seconds' => q{Invalid argument to '$1' option: must be number of seconds},
  234. 'range-version' => q{must be in the format X.Y or X.Y.Z, where X is the major version number, },
  235. 'range-warnbig' => q{The 'warning' option cannot be greater than the 'critical' option},
  236. 'range-warnbigsize' => q{The 'warning' option ($1 bytes) cannot be larger than the 'critical' option ($2 bytes)},
  237. 'range-warnbigtime' => q{The 'warning' option ($1 s) cannot be larger than the 'critical' option ($2 s)},
  238. 'range-warnsmall' => q{The 'warning' option cannot be less than the 'critical' option},
  239. 'range-nointfortime' => q{Invalid argument for '$1' options: must be an integer, time or integer for time},
  240. 'relsize-msg-ind' => q{largest index is "$1": $2},
  241. 'relsize-msg-reli' => q{largest relation is index "$1": $2},
  242. 'relsize-msg-relt' => q{largest relation is table "$1": $2},
  243. 'relsize-msg-tab' => q{largest table is "$1": $2},
  244. 'rep-badarg' => q{Invalid repinfo argument: expected 6 comma-separated values},
  245. 'rep-duh' => q{Makes no sense to test replication with same values},
  246. 'rep-fail' => q{Row not replicated to slave $1},
  247. 'rep-noarg' => q{Need a repinfo argument},
  248. 'rep-norow' => q{Replication source row not found: $1},
  249. 'rep-noslaves' => q{No slaves found},
  250. 'rep-notsame' => q{Cannot test replication: values are not the same},
  251. 'rep-ok' => q{Row was replicated},
  252. 'rep-sourcefail' => q{Source update failed},
  253. 'rep-timeout' => q{Row was not replicated. Timeout: $1},
  254. 'rep-unknown' => q{Replication check failed},
  255. 'rep-wrongvals' => q{Cannot test replication: values are not the right ones ($1 not $2 nor $3)},
  256. 'runcommand-err' => q{Unknown error inside of the "run_command" function},
  257. 'runcommand-nodb' => q{No target databases could be found},
  258. 'runcommand-nodupe' => q{Could not dupe STDERR},
  259. 'runcommand-noerr' => q{Could not open STDERR?!},
  260. 'runcommand-nosys' => q{System call failed with a $1},
  261. 'runcommand-pgpass' => q{Created temporary pgpass file $1},
  262. 'runcommand-timeout' => q{Command timed out! Consider boosting --timeout higher than $1},
  263. 'runtime-badmrtg' => q{invalid queryname?},
  264. 'runtime-badname' => q{Invalid queryname option: must be a simple view name},
  265. 'runtime-msg' => q{query runtime: $1 seconds},
  266. 'schema' => q{Schema},
  267. 'ss-createfile' => q{Created file $1},
  268. 'ss-different' => q{"$1" is different:},
  269. 'ss-existson' => q{Exists on:},
  270. 'ss-failed' => q{Databases were different. Items not matched: $1},
  271. 'ss-matched' => q{All databases have identical items},
  272. 'ss-missingon' => q{Missing on:},
  273. 'ss-noexist' => q{$1 "$2" does not exist on all databases:},
  274. 'ss-notset' => q{"$1" is not set on all databases:},
  275. 'ss-suffix' => q{Error: cannot use suffix unless looking at time-based schemas},
  276. 'seq-die' => q{Could not determine information about sequence $1},
  277. 'seq-msg' => q{$1=$2% (calls left=$3)},
  278. 'seq-none' => q{No sequences found},
  279. 'size' => q{size},
  280. 'slony-noschema' => q{Could not determine the schema for Slony},
  281. 'slony-nonumber' => q{Call to sl_status did not return a number},
  282. 'slony-lagtime' => q{Slony lag time: $1},
  283. 'symlink-create' => q{Created "$1"},
  284. 'symlink-done' => q{Not creating "$1": $2 already linked to "$3"},
  285. 'symlink-exists' => q{Not creating "$1": $2 file already exists},
  286. 'symlink-fail1' => q{Failed to unlink "$1": $2},
  287. 'symlink-fail2' => q{Could not symlink $1 to $2: $3},
  288. 'symlink-name' => q{This command will not work unless the program has the word "postgres" in it},
  289. 'symlink-unlink' => q{Unlinking "$1":$2 },
  290. 'table' => q{Table},
  291. 'testmode-end' => q{END OF TEST MODE},
  292. 'testmode-fail' => q{Connection failed: $1 $2},
  293. 'testmode-norun' => q{Cannot run "$1" on $2: version must be >= $3, but is $4},
  294. 'testmode-noset' => q{Cannot run "$1" on $2: $3 is not set to on},
  295. 'testmode-nover' => q{Could not find version for $1},
  296. 'testmode-ok' => q{Connection ok: $1},
  297. 'testmode-start' => q{BEGIN TEST MODE},
  298. 'time-day' => q{day},
  299. 'time-days' => q{days},
  300. 'time-hour' => q{hour},
  301. 'time-hours' => q{hours},
  302. 'time-minute' => q{minute},
  303. 'time-minutes' => q{minutes},
  304. 'time-month' => q{month},
  305. 'time-months' => q{months},
  306. 'time-second' => q{second},
  307. 'time-seconds' => q{seconds},
  308. 'time-week' => q{week},
  309. 'time-weeks' => q{weeks},
  310. 'time-year' => q{year},
  311. 'time-years' => q{years},
  312. 'timesync-diff' => q{diff},
  313. 'timesync-msg' => q{timediff=$1 DB=$2 Local=$3},
  314. 'transactions' => q{transactions},
  315. 'trigger-msg' => q{Disabled triggers: $1},
  316. 'txn-time' => q{transaction_time},
  317. 'txnidle-count-msg' => q{Total idle in transaction: $1},
  318. 'txnidle-count-none' => q{not more than $1 idle in transaction},
  319. 'txnidle-for-msg' => q{$1 idle transactions longer than $2s, longest: $3s$4 $5},
  320. 'txnidle-msg' => q{longest idle in txn: $1s$2 $3},
  321. 'txnidle-none' => q{no idle in transaction},
  322. 'txntime-count-msg' => q{Total transactions: $1},
  323. 'txntime-count-none' => q{not more than $1 transactions},
  324. 'txntime-for-msg' => q{$1 transactions longer than $2s, longest: $3s$4 $5},
  325. 'txntime-msg' => q{longest txn: $1s$2 $3},
  326. 'txntime-none' => q{No transactions},
  327. 'txnwrap-cbig' => q{The 'critical' value must be less than 2 billion},
  328. 'txnwrap-wbig' => q{The 'warning' value must be less than 2 billion},
  329. 'unknown-error' => q{Unknown error},
  330. 'usage' => qq{\nUsage: \$1 <options>\n Try "\$1 --help" for a complete list of options\n Try "\$1 --man" for the full manual\n},
  331. 'user' => q{User},
  332. 'username' => q{username},
  333. 'vac-nomatch-a' => q{No matching tables have ever been analyzed},
  334. 'vac-nomatch-v' => q{No matching tables have ever been vacuumed},
  335. 'version' => q{version $1},
  336. 'version-badmrtg' => q{Invalid mrtg version argument},
  337. 'version-fail' => q{version $1, but expected $2},
  338. 'version-ok' => q{version $1},
  339. 'wal-numfound' => q{WAL files found: $1},
  340. 'wal-numfound2' => q{WAL "$2" files found: $1},
  341. },
  342. 'fr' => {
  343. 'address' => q{adresse},
  344. 'age' => q{âge},
  345. 'backends-fatal' => q{N'a pas pu se connecter : trop de connexions},
  346. 'backends-mrtg' => q{DB=$1 Connexions maximum=$2},
  347. 'backends-msg' => q{$1 connexions sur $2 ($3%)},
  348. 'backends-nomax' => q{N'a pas pu déterminer max_connections},
  349. 'backends-oknone' => q{Aucune connexion},
  350. 'backends-po' => q{désolé, trop de clients sont déjà connectés},
  351. 'backends-users' => q{$1 pour le nombre d'utilisateurs doit être un nombre ou un pourcentage},
  352. 'bloat-index' => q{(db $1) index $2 lignes:$3 pages:$4 devrait être:$5 ($6X) octets perdus:$7 ($8)},
  353. 'bloat-nomin' => q{aucune relation n'atteint le critère minimum de fragmentation},
  354. 'bloat-table' => q{(db $1) table $2.$3 lignes:$4 pages:$5 devrait être:$6 ($7X) place perdue:$8 ($9)},
  355. 'bug-report' => q{Merci de rapporter ces d??tails ?? check_postgres@bucardo.org:},
  356. 'checkcluster-id' => q{Identifiant système de la base de données :},
  357. 'checkcluster-msg' => q{cluster_id : $1},
  358. 'checkcluster-nomrtg'=> q{Doit fournir un numéro via l'option --mrtg},
  359. 'checkmode-prod' => q{en production},
  360. 'checkmode-recovery' => q{en restauration d'archives},
  361. 'checkmode-state' => q{État de l'instance :},
  362. 'checkpoint-baddir' => q{data_directory invalide : "$1"},
  363. 'checkpoint-baddir2' => q{pg_controldata n'a pas pu lire le répertoire des données indiqué : « $1 »},
  364. 'checkpoint-badver' => q{Échec lors de l'exécution de pg_controldata - probablement la mauvaise version ($1)},
  365. 'checkpoint-badver2' => q{Échec lors de l'exécution de pg_controldata - est-ce la bonne version ?},
  366. 'checkpoint-nodir' => q{Vous devez fournir un argument --datadir ou configurer la variable d'environnement PGDATA},
  367. 'checkpoint-nodp' => q{Vous devez installer le module Perl Date::Parse pour utiliser l'action checkpoint},
  368. 'checkpoint-noparse' => q{Incapable d'analyser le résultat de la commande pg_controldata : "$1"},
  369. 'checkpoint-noregex' => q{La regex pour ce test n'a pas été trouvée},
  370. 'checkpoint-nosys' => q{N'a pas pu appeler pg_controldata : $1},
  371. 'checkpoint-ok' => q{Le dernier CHECKPOINT est survenu il y a une seconde},
  372. 'checkpoint-ok2' => q{Le dernier CHECKPOINT est survenu il y a $1 secondes},
  373. 'checkpoint-po' => q{Heure du dernier point de contrôle :},
  374. 'checksum-msg' => q{somme de contrôle : $1},
  375. 'checksum-nomd' => q{Vous devez installer le module Perl Digest::MD5 pour utiliser l'action checksum},
  376. 'checksum-nomrtg' => q{Vous devez fournir une somme de contrôle avec l'option --mrtg},
  377. 'custom-invalid' => q{Format invalide renvoyé par la requête personnalisée},
  378. 'custom-norows' => q{Aucune ligne renvoyée},
  379. 'custom-nostring' => q{Vous devez fournir une requête},
  380. 'database' => q{base de données},
  381. 'dbsize-version' => q{La base de données cible doit être une version 8.1 ou ultérieure pour exécuter l'action database_size},
  382. 'die-action-version' => q{Ne peut pas exécuter « $1 » : la version du serveur doit être supérieure ou égale à $2, alors qu'elle est $3},
  383. 'die-badtime' => q{La valeur de « $1 » doit être une heure valide. Par exemple, -$2 1s -$2 « 10 minutes »},
  384. 'die-badversion' => q{Version invalide : $1},
  385. 'die-noset' => q{Ne peut pas exécuter « $1 » $2 n'est pas activé},
  386. 'die-nosetting' => q{N'a pas pu récupérer le paramètre « $1 »},
  387. 'diskspace-fail' => q{Résultat invalide pour la commande « $1 » : $2},
  388. 'diskspace-msg' => q{Le système de fichiers $1 monté sur $2 utilise $3 sur $4 ($5%)},
  389. 'diskspace-nodata' => q{N'a pas pu déterminer data_directory : êtes-vous connecté en tant que super-utilisateur ?},
  390. 'diskspace-nodf' => q{N'a pas pu trouver l'exécutable /bin/df},
  391. 'diskspace-nodir' => q{N'a pas pu trouver le répertoire des données « $1 »},
  392. 'files' => q{fichiers},
  393. 'file-noclose' => q{N'a pas pu fermer $1 : $2},
  394. 'fsm-page-highver' => q{Ne peut pas vérifier fsm_pages sur des serveurs en version 8.4 ou ultérieure},
  395. 'fsm-page-msg' => q{emplacements de pages utilisés par la FSM : $1 sur $2 ($3%)},
  396. 'fsm-rel-highver' => q{Ne peut pas vérifier fsm_relations sur des serveurs en version 8.4 ou ultérieure},
  397. 'fsm-rel-msg' => q{relations tracées par la FSM : $1 sur $2 ($3%)},
  398. 'hs-no-role' => q{Pas de couple ma??tre/esclave},
  399. 'hs-no-location' => q{N'a pas pu obtenir l'emplacement courant dans le journal des transactions sur $1},
  400. 'hs-receive-delay' => q{délai de réception},
  401. 'hs-replay-delay' => q{délai de rejeu},
  402. 'index' => q{Index},
  403. 'invalid-option' => q{Option invalide},
  404. 'invalid-query' => q{Une requête invalide a renvoyé : $1},
  405. 'language' => q{Langage},
  406. 'listener-msg' => q{processus LISTEN trouvés : $1},
  407. 'listening' => q{en écoute},
  408. 'locks-msg' => q{total des verrous « $1 » : $2},
  409. 'locks-msg2' => q{total des verrous : $1},
  410. 'logfile-bad' => q{Option logfile invalide « $1 »},
  411. 'logfile-debug' => q{Journal applicatif final : $1},
  412. 'logfile-dne' => q{le journal applicatif $1 n'existe pas !},
  413. 'logfile-fail' => q{échec pour tracer dans : $1},
  414. 'logfile-ok' => q{trace dans : $1},
  415. 'logfile-openfail' => q{échec pour l'ouverture du journal applicatif « $1 » : $2},
  416. 'logfile-opt-bad' => q{Option logfile invalide},
  417. 'logfile-seekfail' => q{Échec de la recherche dans $1 : $2},
  418. 'logfile-stderr' => q{La sortie des traces a été redirigés stderr : merci de fournir un nom de fichier},
  419. 'logfile-syslog' => q{La base de données utiliser syslog, merci de spécifier le chemin avec l'option --logfile (fac=$1)},
  420. 'mode-standby' => q{Serveur en mode standby},
  421. 'mode' => q{mode},
  422. 'mrtg-fail' => q{Échec de l'action $1 : $2},
  423. 'new-ver-nocver' => q{N'a pas pu t??l??charger les informations de version pour $1},
  424. 'new-ver-badver' => q{N'a pas pu analyser les informations de version pour $1},
  425. 'new-ver-dev' => q{Ne peut pas comparer les versions sur des versions de d??veloppement : vous avez $1 version $2},
  426. 'new-ver-nolver' => q{N'a pas pu d??terminer les informations de version locale pour $1},
  427. 'new-ver-ok' => q{La version $1 est la dernière pour $2},
  428. 'new-ver-warn' => q{Merci de mettre à jour vers la version $1 de $2. Vous utilisez actuellement la $3},
  429. 'new-ver-tt' => q{Votre version de $1 ($2) semble ult??rieure ?? la version courante ! ($3)},
  430. 'no-db' => q{Pas de bases de données},
  431. 'no-match-db' => q{Aucune base de données trouvée à cause des options d'exclusion/inclusion},
  432. 'no-match-fs' => q{Aucun système de fichier trouvé à cause des options d'exclusion/inclusion},
  433. 'no-match-rel' => q{Aucune relation trouvée à cause des options d'exclusion/inclusion},
  434. 'no-match-set' => q{Aucun paramètre trouvé à cause des options d'exclusion/inclusion},
  435. 'no-match-table' => q{Aucune table trouvée à cause des options d'exclusion/inclusion},
  436. 'no-match-user' => q{Aucune entrée trouvée à cause options d'exclusion/inclusion},
  437. 'no-parse-psql' => q{N'a pas pu analyser la sortie de psql !},
  438. 'no-time-hires' => q{N'a pas trouvé le module Time::HiRes, nécessaire quand « showtime » est activé},
  439. 'opt-output-invalid' => q{Sortie invalide : doit être 'nagios' ou 'mrtg' ou 'simple' ou 'cacti'},
  440. 'opt-psql-badpath' => q{Argument invalide pour psql : doit être le chemin complet vers un fichier nommé psql},
  441. 'opt-psql-noexec' => q{ Le fichier « $1 » ne paraît pas exécutable},
  442. 'opt-psql-noexist' => q{Ne peut pas trouver l'exécutable psql indiqué : $1},
  443. 'opt-psql-nofind' => q{N'a pas pu trouver un psql exécutable},
  444. 'opt-psql-nover' => q{N'a pas pu déterminer la version de psql},
  445. 'opt-psql-restrict' => q{Ne peut pas utiliser l'option --PGBINDIR ou --PSQL si NO_PSQL_OPTION est activé},
  446. 'pgbouncer-pool' => q{Pool=$1 $2=$3},
  447. 'pgb-backends-mrtg' => q{base=$1 connexions max=$2},
  448. 'pgb-backends-msg' => q{$1 connexions sur $2 ($3%)},
  449. 'pgb-backends-none' => q{Aucune connection},
  450. 'pgb-backends-users' => q{Le nombre d'utilisateurs, $1, doit être un nombre ou un pourcentage},
  451. 'PID' => q{PID},
  452. 'port' => q{port},
  453. 'preptxn-none' => q{Aucune transaction préparée trouvée},
  454. 'psa-disabled' => q{Pas de requ??te - est-ce que stats_command_string ou track_activities sont d??sactiv??s ?},
  455. 'psa-noexact' => q{Erreur inconnue},
  456. 'psa-nosuper' => q{Aucune correspondance - merci de m'ex??cuter en tant que superutilisateur},
  457. 'qtime-count-msg' => q{Requêtes totales : $1},
  458. 'qtime-count-none' => q{pas plus que $1 requêtes},
  459. 'qtime-for-msg' => q{$1 requêtes plus longues que $2s, requête la plus longue : $3s$4 $5},
  460. 'qtime-msg' => q{requête la plus longue : $1s$2 $3},
  461. 'qtime-none' => q{aucune requête},
  462. 'queries' => q{requêtes},
  463. 'query-time' => q{durée de la requête},
  464. 'range-badcs' => q{Option « $1 » invalide : doit être une somme de contrôle},
  465. 'range-badlock' => q{Option « $1 » invalide : doit être un nombre de verrou ou « type1=#:type2=# »},
  466. 'range-badpercent' => q{Option « $1 » invalide : doit être un pourcentage},
  467. 'range-badpercsize' => q{Option « $1 » invalide : doit être une taille ou un pourcentage},
  468. 'range-badsize' => q{Taille invalide pour l'option « $1 »},
  469. 'range-badtype' => q{validate_range appelé avec un type inconnu « $1 »},
  470. 'range-badversion' => q{Chaîne invalide pour l'option « $1 » : $2},
  471. 'range-cactionly' => q{Cette action est pour cacti seulement et ne prend pas les arguments warning et critical},
  472. 'range-int' => q{Argument invalide pour l'option « $1 » : doit être un entier},
  473. 'range-int-pos' => q{Argument invalide pour l'option « $1 » : doit être un entier positif},
  474. 'range-neg-percent' => q{Ne peut pas indiquer un pourcentage négatif !},
  475. 'range-none' => q{Les options warning et critical ne sont pas nécessaires},
  476. 'range-noopt-both' => q{Doit fournir les options warning et critical},
  477. 'range-noopt-one' => q{Doit fournir une option warning ou critical},
  478. 'range-noopt-only' => q{Peut seulement fournir une option warning ou critical},
  479. 'range-noopt-orboth' => q{Doit fournir une option warning, une option critical ou les deux},
  480. 'range-noopt-size' => q{Doit fournir une taille warning et/ou critical},
  481. 'range-nosize' => q{Doit fournir une taille warning et/ou critical},
  482. 'range-notime' => q{Doit fournir une heure warning et/ou critical},
  483. 'range-seconds' => q{Argument invalide pour l'option « $1 » : doit être un nombre de secondes},
  484. 'range-version' => q{doit être dans le format X.Y ou X.Y.Z, où X est le numéro de version majeure, },
  485. 'range-warnbig' => q{L'option warning ne peut pas être plus grand que l'option critical},
  486. 'range-warnbigsize' => q{L'option warning ($1 octets) ne peut pas être plus grand que l'option critical ($2 octets)},
  487. 'range-warnbigtime' => q{L'option warning ($1 s) ne peut pas être plus grand que l'option critical ($2 s)},
  488. 'range-warnsmall' => q{L'option warningne peut pas être plus petit que l'option critical},
  489. 'range-nointfortime' => q{Argument invalide pour l'option '$1' : doit être un entier, une heure ou un entier horaire},
  490. 'relsize-msg-ind' => q{le plus gros index est « $1 » : $2},
  491. 'relsize-msg-reli' => q{la plus grosse relation est l'index « $1 » : $2},
  492. 'relsize-msg-relt' => q{la plus grosse relation est la table « $1 » : $2},
  493. 'relsize-msg-tab' => q{la plus grosse table est « $1 » : $2},
  494. 'rep-badarg' => q{Argument repinfo invalide : 6 valeurs séparées par des virgules attendues},
  495. 'rep-duh' => q{Aucun sens à tester la réplication avec les mêmes valeurs},
  496. 'rep-fail' => q{Ligne non répliquée sur l'esclave $1},
  497. 'rep-noarg' => q{A besoin d'un argument repinfo},
  498. 'rep-norow' => q{Ligne source de la réplication introuvable : $1},
  499. 'rep-noslaves' => q{Aucun esclave trouvé},
  500. 'rep-notsame' => q{Ne peut pas tester la réplication : les valeurs ne sont pas identiques},
  501. 'rep-ok' => q{La ligne a été répliquée},
  502. 'rep-sourcefail' => q{Échec de la mise à jour de la source},
  503. 'rep-timeout' => q{La ligne n'a pas été répliquée. Délai dépassé : $1},
  504. 'rep-unknown' => q{Échec du test de la réplication},
  505. 'rep-wrongvals' => q{Ne peut pas tester la réplication : les valeurs ne sont pas les bonnes (ni $1 ni $2 ni $3)},
  506. 'runcommand-err' => q{Erreur inconnue de la fonction « run_command »},
  507. 'runcommand-nodb' => q{Aucune base de données cible trouvée},
  508. 'runcommand-nodupe' => q{N'a pas pu dupliqué STDERR},
  509. 'runcommand-noerr' => q{N'a pas pu ouvrir STDERR},
  510. 'runcommand-nosys' => q{Échec de l'appel système avec un $1},
  511. 'runcommand-pgpass' => q{Création du fichier pgpass temporaire $1},
  512. 'runcommand-timeout' => q{Délai épuisée pour la commande ! Essayez d'augmenter --timeout à une valeur plus importante que $1},
  513. 'runtime-badmrtg' => q{queryname invalide ?},
  514. 'runtime-badname' => q{Option invalide pour queryname option : doit être le nom d'une vue},
  515. 'runtime-msg' => q{durée d'exécution de la requête : $1 secondes},
  516. 'schema' => q{Schéma},
  517. 'ss-createfile' => q{Création du fichier $1},
  518. 'ss-different' => q{"$1" est différent:},
  519. 'ss-existson' => q{Existe sur :},
  520. 'ss-failed' => q{Les bases de données sont différentes. Éléments différents : $1},
  521. 'ss-matched' => q{Les bases de données ont les mêmes éléments},
  522. 'ss-missingon' => q{Manque sur :},
  523. 'ss-noexist' => q{$1 "$2" n'existe pas sur toutes les bases de données :},
  524. 'ss-notset' => q{"$1" n'est pas configuré sur toutes les bases de données :},
  525. 'ss-suffix' => q{Erreur : ne peut pas utiliser le suffixe sauf à rechercher des schémas basés sur l'horloge},
  526. 'size' => q{taille},
  527. 'slony-noschema' => q{N'a pas pu déterminer le schéma de Slony},
  528. 'slony-nonumber' => q{L'appel à sl_status n'a pas renvoyé un numéro},
  529. 'slony-lagtime' => q{Durée de lag de Slony : $1},
  530. 'seq-die' => q{N'a pas pu récupérer d'informations sur la séquence $1},
  531. 'seq-msg' => q{$1=$2% (appels restant=$3)},
  532. 'seq-none' => q{Aucune sequences trouvée},
  533. 'symlink-create' => q{Création de « $1 »},
  534. 'symlink-done' => q{Création impossible de « $1 »: $2 est déjà lié à "$3"},
  535. 'symlink-exists' => q{Création impossible de « $1 »: le fichier $2 existe déjà},
  536. 'symlink-fail1' => q{Échec de la suppression de « $1 » : $2},
  537. 'symlink-fail2' => q{N'a pas pu supprimer le lien symbolique $1 vers $2 : $3},
  538. 'symlink-name' => q{Cette commande ne fonctionnera pas sauf si le programme contient le mot « postgres »},
  539. 'symlink-unlink' => q{Supression de « $1 » :$2 },
  540. 'table' => q{Table},
  541. 'testmode-end' => q{FIN DU MODE DE TEST},
  542. 'testmode-fail' => q{Échec de la connexion : $1 $2},
  543. 'testmode-norun' => q{N'a pas pu exécuter « $1 » sur $2 : la version doit être supérieure ou égale à $3, mais est $4},
  544. 'testmode-noset' => q{N'a pas pu exécuter « $1 » sur $2 : $3 n'est pas activé},
  545. 'testmode-nover' => q{N'a pas pu trouver la version de $1},
  546. 'testmode-ok' => q{Connexion OK : $1},
  547. 'testmode-start' => q{DÉBUT DU MODE DE TEST},
  548. 'time-day' => q{jour},
  549. 'time-days' => q{jours},
  550. 'time-hour' => q{heure},
  551. 'time-hours' => q{heures},
  552. 'time-minute' => q{minute},
  553. 'time-minutes' => q{minutes},
  554. 'time-month' => q{mois},
  555. 'time-months' => q{mois},
  556. 'time-second' => q{seconde},
  557. 'time-seconds' => q{secondes},
  558. 'time-week' => q{semaine},
  559. 'time-weeks' => q{semaines},
  560. 'time-year' => q{année},
  561. 'time-years' => q{années},
  562. 'timesync-diff' => q{diff},
  563. 'timesync-msg' => q{timediff=$1 Base de données=$2 Local=$3},
  564. 'transactions' => q{transactions},
  565. 'trigger-msg' => q{Triggers désactivés : $1},
  566. 'txn-time' => q{durée de la transaction},
  567. 'txnidle-count-msg' => q{Transactions en attente totales : $1},
  568. 'txnidle-count-none' => q{pas plus de $1 transaction en attente},
  569. 'txnidle-for-msg' => q{$1 transactions en attente plus longues que $2s, transaction la plus longue : $3s$4 $5},
  570. 'txnidle-msg' => q{transaction en attente la plus longue : $1s$2 $3},
  571. 'txnidle-none' => q{Aucun processus en attente dans une transaction},
  572. 'txntime-count-msg' => q{Transactions totales : $1},
  573. 'txntime-count-none' => q{pas plus que $1 transactions},
  574. 'txntime-for-msg' => q{$1 transactions plus longues que $2s, transaction la plus longue : $3s$4 $5},
  575. 'txntime-msg' => q{Transaction la plus longue : $1s$2 $3},
  576. 'txntime-none' => q{Aucune transaction},
  577. 'txnwrap-cbig' => q{La valeur critique doit être inférieure à 2 milliards},
  578. 'txnwrap-wbig' => q{La valeur d'avertissement doit être inférieure à 2 milliards},
  579. 'unknown-error' => q{erreur inconnue},
  580. 'usage' => qq{\nUsage: \$1 <options>\n Essayez « \$1 --help » pour liste complète des options\n\n},
  581. 'username' => q{nom utilisateur},
  582. 'user' => q{Utilisateur},
  583. 'vac-nomatch-a' => q{Aucune des tables correspondantes n'a eu d'opération ANALYZE},
  584. 'vac-nomatch-v' => q{Aucune des tables correspondantes n'a eu d'opération VACUUM},
  585. 'version' => q{version $1},
  586. 'version-badmrtg' => q{Argument invalide pour la version de mrtg},
  587. 'version-fail' => q{version $1, alors que la version attendue est $2},
  588. 'version-ok' => q{version $1},
  589. 'wal-numfound' => q{Fichiers WAL trouvés : $1},
  590. 'wal-numfound2' => q{Fichiers WAL "$2" trouvés : $1},
  591. },
  592. 'af' => {
  593. },
  594. 'cs' => {
  595. 'checkpoint-po' => q{�as posledn�ho kontroln�ho bodu:},
  596. },
  597. 'de' => {
  598. 'backends-po' => q{tut mir leid, schon zu viele Verbindungen},
  599. 'checkpoint-po' => q{Zeit des letzten Checkpoints:},
  600. },
  601. 'es' => {
  602. 'backends-po' => q{lo siento, ya tenemos demasiados clientes},
  603. 'checkpoint-po' => q{Instante de �ltimo checkpoint:},
  604. },
  605. 'fa' => {
  606. 'checkpoint-po' => q{زمان آخرین وارسی:},
  607. },
  608. 'hr' => {
  609. 'backends-po' => q{nažalost, već je otvoreno previše klijentskih veza},
  610. },
  611. 'hu' => {
  612. 'checkpoint-po' => q{A legut�bbi ellen�rz�pont ideje:},
  613. },
  614. 'it' => {
  615. 'checkpoint-po' => q{Orario ultimo checkpoint:},
  616. },
  617. 'ja' => {
  618. 'backends-po' => q{現在クライアント数が多すぎます},
  619. 'checkpoint-po' => q{最終チェックポイント時刻:},
  620. },
  621. 'ko' => {
  622. 'backends-po' => q{최대 동시 접속자 수를 초과했습니다.},
  623. 'checkpoint-po' => q{������ üũ����Ʈ �ð�:},
  624. },
  625. 'nb' => {
  626. 'backends-po' => q{beklager, for mange klienter},
  627. 'checkpoint-po' => q{Tidspunkt for nyeste kontrollpunkt:},
  628. },
  629. 'nl' => {
  630. },
  631. 'pl' => {
  632. 'checkpoint-po' => q{Czas najnowszego punktu kontrolnego:},
  633. },
  634. 'pt_BR' => {
  635. 'backends-po' => q{desculpe, muitos clientes conectados},
  636. 'checkpoint-po' => q{Hora do último ponto de controle:},
  637. },
  638. 'ro' => {
  639. 'checkpoint-po' => q{Timpul ultimului punct de control:},
  640. },
  641. 'ru' => {
  642. 'backends-po' => q{��������, ��� ������� ����� ��������},
  643. 'checkpoint-po' => q{����� ��������� checkpoint:},
  644. },
  645. 'sk' => {
  646. 'backends-po' => q{je mi ��to, je u� pr�li� ve�a klientov},
  647. 'checkpoint-po' => q{Čas posledného kontrolného bodu:},
  648. },
  649. 'sl' => {
  650. 'backends-po' => q{povezanih je �e preve� odjemalcev},
  651. 'checkpoint-po' => q{�as zadnje kontrolne to�ke ............},
  652. },
  653. 'sv' => {
  654. 'backends-po' => q{ledsen, f�r m�nga klienter},
  655. 'checkpoint-po' => q{Tidpunkt f�r senaste kontrollpunkt:},
  656. },
  657. 'ta' => {
  658. 'checkpoint-po' => q{நவீன சோதனை மையத்தின் நேரம்:},
  659. },
  660. 'tr' => {
  661. 'backends-po' => q{üzgünüm, istemci sayısı çok fazla},
  662. 'checkpoint-po' => q{En son checkpoint'in zamanı:},
  663. },
  664. 'zh_CN' => {
  665. 'backends-po' => q{�Բ���, �Ѿ���̫���Ŀͻ�},
  666. 'checkpoint-po' => q{���¼�������ʱ��:},
  667. },
  668. 'zh_TW' => {
  669. 'backends-po' => q{對不起，用戶端過多},
  670. 'checkpoint-po' => q{最新的檢查點時間:},
  671. },
  672. );
  673. ## use critic
  674.  
  675. our $lang = $ENV{LC_ALL} || $ENV{LC_MESSAGES} || $ENV{LANG} || 'en';
  676. $lang = substr($lang,0,2);
  677.  
  678. ## Messages are stored in these until the final output via finishup()
  679. our (%ok, %warning, %critical, %unknown);
  680.  
  681. our $ME = basename($0);
  682. our $ME2 = 'check_postgres.pl';
  683. our $USAGE = msg('usage', $ME);
  684.  
  685. ## This gets turned on for meta-commands which don't hit a Postgres database
  686. our $nohost = 0;
  687.  
  688. ## Global error string, mostly used for MRTG error handling
  689. our $ERROR = '';
  690.  
  691. $opt{test} = 0;
  692. $opt{timeout} = 30;
  693.  
  694. ## Look for any rc files to control additional parameters
  695. ## Command line options always overwrite these
  696. ## Format of these files is simply name=val
  697.  
  698. ## This option must come before the GetOptions call
  699. for my $arg (@ARGV) {
  700. if ($arg eq '--no-check_postgresrc') {
  701. $opt{'no-check_postgresrc'} = 1;
  702. last;
  703. }
  704. }
  705.  
  706. ## Used by same_schema in the find_catalog_info sub
  707. my %catalog_info = (
  708.  
  709. user => {
  710. SQL => q{
  711. SELECT *, usename AS name, quote_ident(usename) AS safeusename
  712. FROM pg_user},
  713. deletecols => [ qw{ passwd } ],
  714. },
  715.  
  716. schema => {
  717. SQL => q{
  718. SELECT n.oid, quote_ident(nspname) AS name, quote_ident(usename) AS owner, nspacl
  719. FROM pg_namespace n
  720. JOIN pg_user u ON (u.usesysid = n.nspowner)},
  721. deletecols => [ ],
  722. exclude => 'temp_schemas',
  723. },
  724. language => {
  725. SQL => q{
  726. SELECT l.*, lanname AS name, quote_ident(usename) AS owner
  727. FROM pg_language l
  728. JOIN pg_user u ON (u.usesysid = l.lanowner)},
  729. SQL2 => q{
  730. SELECT l.*, lanname AS name
  731. FROM pg_language l
  732. },
  733. },
  734. type => {
  735. SQL => q{
  736. SELECT t.oid AS oid, t.*, quote_ident(usename) AS owner, quote_ident(nspname) AS schema,
  737. nspname||'.'||typname AS name
  738. FROM pg_type t
  739. JOIN pg_user u ON (u.usesysid = t.typowner)
  740. JOIN pg_namespace n ON (n.oid = t.typnamespace)
  741. WHERE t.typtype NOT IN ('b','c')},
  742. exclude => 'system',
  743. },
  744. sequence => {
  745. SQL => q{
  746. SELECT c.*, nspname||'.'||relname AS name, quote_ident(usename) AS owner,
  747. (quote_ident(nspname)||'.'||quote_ident(relname)) AS safename,
  748. quote_ident(nspname) AS schema
  749. FROM pg_class c
  750. JOIN pg_user u ON (u.usesysid = c.relowner)
  751. JOIN pg_namespace n ON (n.oid = c.relnamespace)
  752. WHERE c.relkind = 'S'},
  753. innerSQL => 'SELECT * FROM ROWSAFENAME',
  754. },
  755. view => {
  756. SQL => q{
  757. SELECT c.*, nspname||'.'||relname AS name, quote_ident(usename) AS owner,
  758. quote_ident(relname) AS safename, quote_ident(nspname) AS schema,
  759. TRIM(pg_get_viewdef(c.oid, TRUE)) AS viewdef, spcname AS tablespace
  760. FROM pg_class c
  761. JOIN pg_user u ON (u.usesysid = c.relowner)
  762. JOIN pg_namespace n ON (n.oid = c.relnamespace)
  763. LEFT JOIN pg_tablespace s ON (s.oid = c.reltablespace)
  764. WHERE c.relkind = 'v'},
  765. exclude => 'system',
  766. },
  767. table => {
  768. SQL => q{
  769. SELECT c.*, nspname||'.'||relname AS name, quote_ident(usename) AS owner,
  770. quote_ident(relname) AS safename, quote_ident(nspname) AS schema,
  771. spcname AS tablespace
  772. FROM pg_class c
  773. JOIN pg_user u ON (u.usesysid = c.relowner)
  774. JOIN pg_namespace n ON (n.oid = c.relnamespace)
  775. LEFT JOIN pg_tablespace s ON (s.oid = c.reltablespace)
  776. WHERE c.relkind = 'r'},
  777. exclude => 'system',
  778. },
  779. index => {
  780. SQL => q{
  781. SELECT c.*, i.*, nspname||'.'||relname AS name, quote_ident(usename) AS owner,
  782. quote_ident(relname) AS safename, quote_ident(nspname) AS schema,
  783. spcname AS tablespace, amname,
  784. pg_get_indexdef(c.oid) AS indexdef
  785. FROM pg_class c
  786. JOIN pg_user u ON (u.usesysid = c.relowner)
  787. JOIN pg_namespace n ON (n.oid = c.relnamespace)
  788. JOIN pg_index i ON (c.oid = i.indexrelid)
  789. LEFT JOIN pg_tablespace s ON (s.oid = c.reltablespace)
  790. LEFT JOIN pg_am a ON (a.oid = c.relam)
  791. WHERE c.relkind = 'i'},
  792. exclude => 'system',
  793. },
  794. operator => {
  795. SQL => q{
  796. SELECT o.*, o.oid, nspname||'.'||o.oprname AS name, quote_ident(o.oprname) AS safename,
  797. usename AS owner, nspname AS schema,
  798. t1.typname AS resultname,
  799. t2.typname AS leftname, t3.typname AS rightname
  800. FROM pg_operator o
  801. JOIN pg_user u ON (u.usesysid = o.oprowner)
  802. JOIN pg_namespace n ON (n.oid = o.oprnamespace)
  803. JOIN pg_proc p1 ON (p1.oid = o.oprcode)
  804. JOIN pg_type t1 ON (t1.oid = o.oprresult)
  805. LEFT JOIN pg_type t2 ON (t2.oid = o.oprleft)
  806. LEFT JOIN pg_type t3 ON (t3.oid = o.oprright)},
  807. exclude => 'system',
  808. },
  809. trigger => {
  810. SQL => q{
  811. SELECT t.*, n1.nspname||'.'||t.tgname AS name, quote_ident(t.tgname) AS safename, quote_ident(usename) AS owner,
  812. n1.nspname AS tschema, c1.relname AS tname,
  813. n2.nspname AS cschema, c2.relname AS cname,
  814. n3.nspname AS procschema, p.proname AS procname
  815. FROM pg_trigger t
  816. JOIN pg_class c1 ON (c1.oid = t.tgrelid)
  817. JOIN pg_user u ON (u.usesysid = c1.relowner)
  818. JOIN pg_namespace n1 ON (n1.oid = c1.relnamespace)
  819. JOIN pg_proc p ON (p.oid = t.tgfoid)
  820. JOIN pg_namespace n3 ON (n3.oid = p.pronamespace)
  821. LEFT JOIN pg_class c2 ON (c2.oid = t.tgconstrrelid)
  822. LEFT JOIN pg_namespace n2 ON (n2.oid = c2.relnamespace)
  823. WHERE t.tgconstrrelid = 0 AND tgname !~ '^pg_'},
  824. },
  825. function => {
  826. SQL => q{
  827. SELECT p.*, p.oid, nspname||'.'||p.proname AS name, quote_ident(p.proname) AS safename,
  828. md5(prosrc) AS source_checksum,
  829. usename AS owner, nspname AS schema
  830. FROM pg_proc p
  831. JOIN pg_user u ON (u.usesysid = p.proowner)
  832. JOIN pg_namespace n ON (n.oid = p.pronamespace)},
  833. exclude => 'system',
  834. },
  835. constraint => {
  836. SQL => q{
  837. SELECT c.*, c.oid, n.nspname||'.'||c.conname AS name, quote_ident(c.conname) AS safename,
  838. n.nspname AS schema, relname AS tname
  839. FROM pg_constraint c
  840. JOIN pg_namespace n ON (n.oid = c.connamespace)
  841. JOIN pg_class r ON (r.oid = c.conrelid)
  842. JOIN pg_namespace n2 ON (n2.oid = r.relnamespace)},
  843. exclude => 'system',
  844. },
  845. column => {
  846. SQL => q{
  847. SELECT a.*, n.nspname||'.'||c.relname||'.'||attname AS name, quote_ident(a.attname) AS safename,
  848. n.nspname||'.'||c.relname AS tname,
  849. typname, quote_ident(nspname) AS schema,
  850. pg_get_expr(d.adbin, a.attrelid, true) AS default
  851. FROM pg_attribute a
  852. JOIN pg_type t ON (t.oid = a.atttypid)
  853. JOIN pg_class c ON (c.oid = a.attrelid AND c.relkind = 'r')
  854. JOIN pg_namespace n ON (n.oid = c.relnamespace)
  855. LEFT JOIN pg_attrdef d ON (d.adrelid = a.attrelid AND d.adnum = a.attnum)
  856. WHERE attnum >= 1
  857. AND NOT attisdropped},
  858. postSQL => q{ORDER BY n.nspname, c.relname, a.attnum},
  859. exclude => 'system',
  860. },
  861. );
  862.  
  863. my $rcfile;
  864. if (! $opt{'no-check_postgresrc'}) {
  865. if (-e '.check_postgresrc') {
  866. $rcfile = '.check_postgresrc';
  867. }
  868. elsif (-e "$ENV{HOME}/.check_postgresrc") {
  869. $rcfile = "$ENV{HOME}/.check_postgresrc";
  870. }
  871. elsif (-e '/etc/check_postgresrc') {
  872. $rcfile = '/etc/check_postgresrc';
  873. }
  874. elsif (-e '/usr/local/etc/check_postgresrc') {
  875. $rcfile = '/usr/local/etc/check_postgresrc';
  876. }
  877. }
  878. ## We need a temporary hash so that multi-value options can be overridden on the command line
  879. my %tempopt;
  880. if (defined $rcfile) {
  881. open my $rc, '<', $rcfile or die qq{Could not open "$rcfile": $!\n};
  882. RCLINE:
  883. while (<$rc>) {
  884. next if /^\s*#/;
  885. next unless /^\s*(\w+)\s*=\s*(.+?)\s*$/o;
  886. my ($name,$value) = ($1,$2); ## no critic (ProhibitCaptureWithoutTest)
  887. ## Map alternate option spellings to preferred names
  888. if ($name eq 'dbport' or $name eq 'p' or $name eq 'dbport1' or $name eq 'p1' or $name eq 'port1') {
  889. $name = 'port';
  890. }
  891. elsif ($name eq 'dbhost' or $name eq 'H' or $name eq 'dbhost1' or $name eq 'H1' or $name eq 'host1') {
  892. $name = 'host';
  893. }
  894. elsif ($name eq 'db' or $name eq 'db1' or $name eq 'dbname1') {
  895. $name = 'dbname';
  896. }
  897. elsif ($name eq 'u' or $name eq 'u1' or $name eq 'dbuser1') {
  898. $name = 'dbuser';
  899. }
  900. ## Now for all the additional non-1 databases
  901. elsif ($name =~ /^dbport(\d+)$/o or $name eq /^p(\d+)$/o) {
  902. $name = "port$1";
  903. }
  904. elsif ($name =~ /^dbhost(\d+)$/o or $name eq /^H(\d+)$/o) {
  905. $name = "host$1";
  906. }
  907. elsif ($name =~ /^db(\d)$/o) {
  908. $name = "dbname$1";
  909. }
  910. elsif ($name =~ /^u(\d+)$/o) {
  911. $name = "dbuser$1";
  912. }
  913.  
  914. ## These options are multiples ('@s')
  915. for my $arr (qw/include exclude includeuser excludeuser host port
  916. dbuser dbname dbpass dbservice schema/) {
  917. next if $name ne $arr and $name ne "${arr}2";
  918. push @{$tempopt{$name}} => $value;
  919. ## Don't set below as a normal value
  920. next RCLINE;
  921. }
  922. $opt{$name} = $value;
  923. }
  924. close $rc or die;
  925. }
  926.  
  927. die $USAGE if ! @ARGV;
  928.  
  929. GetOptions(
  930. \%opt,
  931. 'version|V',
  932. 'verbose|v+',
  933. 'vv',
  934. 'help|h',
  935. 'quiet|q',
  936. 'man',
  937. 'output=s',
  938. 'simple',
  939. 'showperf=i',
  940. 'perflimit=i',
  941. 'showtime=i',
  942. 'timeout|t=i',
  943. 'test',
  944. 'symlinks',
  945. 'debugoutput=s',
  946. 'no-check_postgresrc',
  947. 'assume-standby-mode',
  948. 'assume-prod',
  949.  
  950. 'action=s',
  951. 'warning=s',
  952. 'critical=s',
  953. 'include=s@',
  954. 'exclude=s@',
  955. 'includeuser=s@',
  956. 'excludeuser=s@',
  957.  
  958. 'host|dbhost|H|dbhost1|H1=s@',
  959. 'port|dbport|p|port1|dbport1|p1=s@',
  960. 'dbname|db|dbname1|db1=s@',
  961. 'dbuser|u|dbuser1|u1=s@',
  962. 'dbpass|dbpass1=s@',
  963. 'dbservice|dbservice1=s@',
  964.  
  965. 'PGBINDIR=s',
  966. 'PSQL=s',
  967.  
  968. 'tempdir=s',
  969. 'get_method=s',
  970. 'language=s',
  971. 'mrtg=s', ## used by MRTG checks only
  972. 'logfile=s', ## used by check_logfile only
  973. 'queryname=s', ## used by query_runtime only
  974. 'query=s', ## used by custom_query only
  975. 'valtype=s', ## used by custom_query only
  976. 'reverse', ## used by custom_query only
  977. 'repinfo=s', ## used by replicate_row only
  978. 'noidle', ## used by backends only
  979. 'datadir=s', ## used by checkpoint only
  980. 'schema=s@', ## used by slony_status only
  981. 'filter=s@', ## used by same_schema only
  982. 'suffix=s', ## used by same_schema only
  983. 'replace', ## used by same_schema only
  984. );
  985.  
  986. die $USAGE if ! keys %opt and ! @ARGV;
  987.  
  988. ## Process the args that are not so easy for Getopt::Long
  989. my @badargs;
  990.  
  991. while (my $arg = pop @ARGV) {
  992.  
  993. ## These must be of the form x=y
  994. if ($arg =~ /^\-?\-?(\w+)\s*=\s*(.+)/o) {
  995. my ($name,$value) = (lc $1, $2);
  996. if ($name =~ /^(?:db)?port(\d+)$/o or $name =~ /^p(\d+)$/o) {
  997. $opt{"port$1"} = $value;
  998. }
  999. elsif ($name =~ /^(?:db)?host(\d+)$/o or $name =~ /^H(\d+)$/o) {
  1000. $opt{"host$1"} = $value;
  1001. }
  1002. elsif ($name =~ /^db(?:name)?(\d+)$/o) {
  1003. $opt{"dbname$1"} = $value;
  1004. }
  1005. elsif ($name =~ /^dbuser(\d+)$/o or $name =~ /^u(\d+)/o) {
  1006. $opt{"dbuser$1"} = $value;
  1007. }
  1008. elsif ($name =~ /^dbpass(\d+)$/o) {
  1009. $opt{"dbpass$1"} = $value;
  1010. }
  1011. elsif ($name =~ /^dbservice(\d+)$/o) {
  1012. $opt{"dbservice$1"} = $value;
  1013. }
  1014. else {
  1015. push @badargs => $arg;
  1016. }
  1017. next;
  1018. }
  1019. push @badargs => $arg;
  1020. }
  1021.  
  1022. if (@badargs) {
  1023. warn "Invalid arguments:\n";
  1024. for (@badargs) {
  1025. warn " $_\n";
  1026. }
  1027. die $USAGE;
  1028. }
  1029.  
  1030. if ( $opt{man} ) {
  1031. require Pod::Usage;
  1032. Pod::Usage::pod2usage({-verbose => 2});
  1033. exit;
  1034. }
  1035.  
  1036. ## Put multi-val options from check_postgresrc in place, only if no command-line args!
  1037. for my $mv (keys %tempopt) {
  1038. $opt{$mv} ||= delete $tempopt{$mv};
  1039. }
  1040.  
  1041. our $VERBOSE = $opt{verbose} || 0;
  1042. $VERBOSE = 5 if $opt{vv};
  1043.  
  1044. our $OUTPUT = lc($opt{output} || '');
  1045.  
  1046. ## Allow the optimization of the get_methods list by an argument
  1047. if ($opt{get_method}) {
  1048. my $found = 0;
  1049. for my $meth (@get_methods) {
  1050. if ($meth =~ /^$opt{get_method}/io) {
  1051. @get_methods = ($meth);
  1052. $found = 1;
  1053. last;
  1054. }
  1055. }
  1056. if (!$found) {
  1057. print "Unknown value for get_method: $opt{get_method}\n";
  1058. print "Valid choices are:\n";
  1059. print (join "\n" => map { s/(\w+).*/$1/; $_ } @get_methods);
  1060. print "\n";
  1061. exit;
  1062. }
  1063. }
  1064.  
  1065. ## Allow the language to be changed by an explicit option
  1066. if ($opt{language}) {
  1067. $lang = substr($opt{language},0,2);
  1068. }
  1069.  
  1070. ## Output the actual string returned by psql in the normal output
  1071. ## Argument is 'a' for all, 'w' for warning, 'c' for critical, 'u' for unknown
  1072. ## Can be grouped together
  1073. our $DEBUGOUTPUT = $opt{debugoutput} || '';
  1074. our $DEBUG_INFO = '?';
  1075.  
  1076. ## If not explicitly given an output, check the current directory,
  1077. ## then fall back to the default.
  1078.  
  1079. if (!$OUTPUT) {
  1080. my $dir = getcwd;
  1081. if ($dir =~ /(nagios|mrtg|simple|cacti)/io) {
  1082. $OUTPUT = lc $1;
  1083. }
  1084. elsif ($opt{simple}) {
  1085. $OUTPUT = 'simple';
  1086. }
  1087. else {
  1088. $OUTPUT = $DEFAULT_OUTPUT;
  1089. }
  1090. }
  1091.  
  1092.  
  1093. ## Extract transforms from the output
  1094. $opt{transform} = '';
  1095. if ($OUTPUT =~ /\b(kb|mb|gb|tb|eb)\b/) {
  1096. $opt{transform} = uc $1;
  1097. }
  1098. if ($OUTPUT =~ /(nagios|mrtg|simple|cacti)/io) {
  1099. $OUTPUT = lc $1;
  1100. }
  1101. ## Check for a valid output setting
  1102. if ($OUTPUT ne 'nagios' and $OUTPUT ne 'mrtg' and $OUTPUT ne 'simple' and $OUTPUT ne 'cacti') {
  1103. die msgn('opt-output-invalid');
  1104. }
  1105.  
  1106. our $MRTG = ($OUTPUT eq 'mrtg' or $OUTPUT eq 'simple') ? 1 : 0;
  1107. our (%stats, %statsmsg);
  1108. our $SIMPLE = $OUTPUT eq 'simple' ? 1 : 0;
  1109.  
  1110. ## See if we need to invoke something based on our name
  1111. our $action = $opt{action} || '';
  1112. if ($ME =~ /check_postgres_(\w+)/ and ! defined $opt{action}) {
  1113. $action = $1;
  1114. }
  1115.  
  1116. $VERBOSE >= 3 and warn Dumper \%opt;
  1117.  
  1118. if ($opt{version}) {
  1119. print qq{$ME2 version $VERSION\n};
  1120. exit 0;
  1121. }
  1122.  
  1123. ## Quick hash to put normal action information in one place:
  1124. our $action_info = {
  1125. # Name # clusterwide? # helpstring
  1126. archive_ready => [1, 'Check the number of WAL files ready in the pg_xlog/archive_status'],
  1127. autovac_freeze => [1, 'Checks how close databases are to autovacuum_freeze_max_age.'],
  1128. backends => [1, 'Number of connections, compared to max_connections.'],
  1129. bloat => [0, 'Check for table and index bloat.'],
  1130. checkpoint => [1, 'Checks how long since the last checkpoint'],
  1131. cluster_id => [1, 'Checks the Database System Identifier'],
  1132. commitratio => [0, 'Report if the commit ratio of a database is too low.'],
  1133. connection => [0, 'Simple connection check.'],
  1134. custom_query => [0, 'Run a custom query.'],
  1135. database_size => [0, 'Report if a database is too big.'],
  1136. dbstats => [1, 'Returns stats from pg_stat_database: Cacti output only'],
  1137. disabled_triggers => [0, 'Check if any triggers are disabled'],
  1138. disk_space => [1, 'Checks space of local disks Postgres is using.'],
  1139. fsm_pages => [1, 'Checks percentage of pages used in free space map.'],
  1140. fsm_relations => [1, 'Checks percentage of relations used in free space map.'],
  1141. hitratio => [0, 'Report if the hit ratio of a database is too low.'],
  1142. hot_standby_delay => [1, 'Check the replication delay in hot standby setup'],
  1143. index_size => [0, 'Checks the size of indexes only.'],
  1144. table_size => [0, 'Checks the size of tables only.'],
  1145. relation_size => [0, 'Checks the size of tables and indexes.'],
  1146. last_analyze => [0, 'Check the maximum time in seconds since any one table has been analyzed.'],
  1147. last_vacuum => [0, 'Check the maximum time in seconds since any one table has been vacuumed.'],
  1148. last_autoanalyze => [0, 'Check the maximum time in seconds since any one table has been autoanalyzed.'],
  1149. last_autovacuum => [0, 'Check the maximum time in seconds since any one table has been autovacuumed.'],
  1150. listener => [0, 'Checks for specific listeners.'],
  1151. locks => [0, 'Checks the number of locks.'],
  1152. logfile => [1, 'Checks that the logfile is being written to correctly.'],
  1153. new_version_bc => [0, 'Checks if a newer version of Bucardo is available.'],
  1154. new_version_box => [0, 'Checks if a newer version of boxinfo is available.'],
  1155. new_version_cp => [0, 'Checks if a newer version of check_postgres.pl is available.'],
  1156. new_version_pg => [0, 'Checks if a newer version of Postgres is available.'],
  1157. new_version_tnm => [0, 'Checks if a newer version of tail_n_mail is available.'],
  1158. pgb_pool_cl_active => [1, 'Check the number of active clients in each pgbouncer pool.'],
  1159. pgb_pool_cl_waiting => [1, 'Check the number of waiting clients in each pgbouncer pool.'],
  1160. pgb_pool_sv_active => [1, 'Check the number of active server connections in each pgbouncer pool.'],
  1161. pgb_pool_sv_idle => [1, 'Check the number of idle server connections in each pgbouncer pool.'],
  1162. pgb_pool_sv_used => [1, 'Check the number of used server connections in each pgbouncer pool.'],
  1163. pgb_pool_sv_tested => [1, 'Check the number of tested server connections in each pgbouncer pool.'],
  1164. pgb_pool_sv_login => [1, 'Check the number of login server connections in each pgbouncer pool.'],
  1165. pgb_pool_maxwait => [1, 'Check the current maximum wait time for client connections in pgbouncer pools.'],
  1166. pgbouncer_backends => [0, 'Check how many clients are connected to pgbouncer compared to max_client_conn.'],
  1167. pgbouncer_checksum => [0, 'Check that no pgbouncer settings have changed since the last check.'],
  1168. pgagent_jobs => [0, 'Check for no failed pgAgent jobs within a specified period of time.'],
  1169. prepared_txns => [1, 'Checks number and age of prepared transactions.'],
  1170. query_runtime => [0, 'Check how long a specific query takes to run.'],
  1171. query_time => [1, 'Checks the maximum running time of current queries.'],
  1172. replicate_row => [0, 'Verify a simple update gets replicated to another server.'],
  1173. same_schema => [0, 'Verify that two databases have the exact same tables, columns, etc.'],
  1174. sequence => [0, 'Checks remaining calls left in sequences.'],
  1175. settings_checksum => [0, 'Check that no settings have changed since the last check.'],
  1176. slony_status => [1, 'Ensure Slony is up to date via sl_status.'],
  1177. timesync => [0, 'Compare database time to local system time.'],
  1178. txn_idle => [1, 'Checks the maximum "idle in transaction" time.'],
  1179. txn_time => [1, 'Checks the maximum open transaction time.'],
  1180. txn_wraparound => [1, 'See how close databases are getting to transaction ID wraparound.'],
  1181. version => [1, 'Check for proper Postgres version.'],
  1182. wal_files => [1, 'Check the number of WAL files in the pg_xlog directory'],
  1183. };
  1184.  
  1185. ## XXX Need to i18n the above
  1186. our $action_usage = '';
  1187. our $longname = 1;
  1188. for (keys %$action_info) {
  1189. $longname = length($_) if length($_) > $longname;
  1190. }
  1191. for (sort keys %$action_info) {
  1192. $action_usage .= sprintf " %-*s - %s\n", 2+$longname, $_, $action_info->{$_}[1];
  1193. }
  1194.  
  1195.  
  1196. if ($opt{help}) {
  1197. print qq{Usage: $ME2 <options>
  1198. Run various tests against one or more Postgres databases.
  1199. Returns with an exit code of 0 (success), 1 (warning), 2 (critical), or 3 (unknown)
  1200. This is version $VERSION.
  1201.  
  1202. Common connection options:
  1203. -H, --host=NAME hostname(s) to connect to; defaults to none (Unix socket)
  1204. -p, --port=NUM port(s) to connect to; defaults to $opt{defaultport}.
  1205. -db, --dbname=NAME database name(s) to connect to; defaults to 'postgres' or 'template1'
  1206. -u --dbuser=NAME database user(s) to connect as; defaults to '$opt{defaultuser}'
  1207. --dbpass=PASS database password(s); use a .pgpass file instead when possible
  1208. --dbservice=NAME service name to use inside of pg_service.conf
  1209.  
  1210. Connection options can be grouped: --host=a,b --host=c --port=1234 --port=3344
  1211. would connect to a-1234, b-1234, and c-3344
  1212.  
  1213. Limit options:
  1214. -w value, --warning=value the warning threshold, range depends on the action
  1215. -c value, --critical=value the critical threshold, range depends on the action
  1216. --include=name(s) items to specifically include (e.g. tables), depends on the action
  1217. --exclude=name(s) items to specifically exclude (e.g. tables), depends on the action
  1218. --includeuser=include objects owned by certain users
  1219. --excludeuser=exclude objects owned by certain users
  1220.  
  1221. Other options:
  1222. --assume-standby-mode assume that server in continious WAL recovery mode
  1223. --assume-prod assume that server in production mode
  1224. --PGBINDIR=PATH path of the postgresql binaries; avoid using if possible
  1225. --PSQL=FILE (deprecated) location of the psql executable; avoid using if possible
  1226. -v, --verbose verbosity level; can be used more than once to increase the level
  1227. -h, --help display this help information
  1228. --man display the full manual
  1229. -t X, --timeout=X how long in seconds before we timeout. Defaults to 30 seconds.
  1230. --symlinks create named symlinks to the main program for each action
  1231.  
  1232. Actions:
  1233. Which test is determined by the --action option, or by the name of the program
  1234. $action_usage
  1235.  
  1236. For a complete list of options and full documentation, view the manual.
  1237.  
  1238. $ME --man
  1239.  
  1240. Or visit: http://bucardo.org/check_postgres/
  1241.  
  1242.  
  1243. };
  1244. exit 0;
  1245. }
  1246.  
  1247. build_symlinks() if $opt{symlinks};
  1248.  
  1249. $action =~ /\w/ or die $USAGE;
  1250.  
  1251. ## Be nice and figure out what they meant
  1252. $action =~ s/\-/_/g;
  1253. $action = lc $action;
  1254.  
  1255. ## Build symlinked copies of this file
  1256. build_symlinks() if $action =~ /build_symlinks/; ## Does not return, may be 'build_symlinks_force'
  1257.  
  1258. ## Die if Time::HiRes is needed but not found
  1259. if ($opt{showtime}) {
  1260. eval {
  1261. require Time::HiRes;
  1262. import Time::HiRes qw/gettimeofday tv_interval sleep/;
  1263. };
  1264. if ($@) {
  1265. die msg('no-time-hires');
  1266. }
  1267. }
  1268.  
  1269. ## We don't (usually) want to die, but want a graceful Nagios-like exit instead
  1270. sub ndie {
  1271. eval { File::Temp::cleanup(); };
  1272. my $msg = shift;
  1273. chomp $msg;
  1274. ## If this message already starts with an ERROR, filter that out for prettiness
  1275. $msg =~ s/^\s*ERROR:\s*/ /;
  1276. ## Trim whitespace
  1277. $msg =~ s/^\s*(.+)\s*$/$1/;
  1278. print "ERROR: $msg\n";
  1279. exit 3;
  1280. }
  1281.  
  1282. sub msg { ## no critic
  1283.  
  1284. my $name = shift || '?';
  1285.  
  1286. my $msg = '';
  1287.  
  1288. if (exists $msg{$lang}{$name}) {
  1289. $msg = $msg{$lang}{$name};
  1290. }
  1291. elsif (exists $msg{'en'}{$name}) {
  1292. $msg = $msg{'en'}{$name};
  1293. }
  1294. else {
  1295. ## Allow for non-matches in certain rare cases
  1296. return '' if $opt{nomsgok};
  1297. my $line = (caller)[2];
  1298. die qq{Invalid message "$name" from line $line\n};
  1299. }
  1300.  
  1301. my $x=1;
  1302. {
  1303. my $val = $_[$x-1];
  1304. $val = '?' if ! defined $val;
  1305. last unless $msg =~ s/\$$x/$val/g;
  1306. $x++;
  1307. redo;
  1308. }
  1309. return $msg;
  1310.  
  1311. } ## end of msg
  1312.  
  1313. sub msgn { ## no critic
  1314. return msg(@_) . "\n";
  1315. }
  1316.  
  1317. sub msg_en {
  1318.  
  1319. my $name = shift || '?';
  1320.  
  1321. return $msg{'en'}{$name};
  1322.  
  1323. } ## end of msg_en
  1324.  
  1325. ## Everything from here on out needs psql, so find and verify a working version:
  1326. if ($NO_PSQL_OPTION) {
  1327. (delete $opt{PGBINDIR} or delete $opt{PSQL}) and ndie msg('opt-psql-restrict');
  1328. }
  1329. if (! defined $PGBINDIR or ! length $PGBINDIR) {
  1330. if (defined $ENV{PGBINDIR} and length $ENV{PGBINDIR}){
  1331. $PGBINDIR = $ENV{PGBINDIR};
  1332. }
  1333. elsif (defined $opt{PGBINDIR} and length $opt{PGBINDIR}){
  1334. $PGBINDIR = $opt{PGBINDIR};
  1335. }
  1336. else {
  1337. undef $PGBINDIR;
  1338. }
  1339. }
  1340. if (exists $opt{PSQL}) {
  1341. $PSQL = $opt{PSQL};
  1342. $PSQL =~ m{^/[\w\d\/]*psql$} or ndie msg('opt-psql-badpath');
  1343. -e $PSQL or ndie msg('opt-psql-noexist', $PSQL);
  1344. }
  1345. else {
  1346. my $psql = (defined $PGBINDIR) ? "$PGBINDIR/psql" : 'psql';
  1347. chomp($PSQL = qx{which "$psql"});
  1348. $PSQL or ndie msg('opt-psql-nofind');
  1349. }
  1350. -x $PSQL or ndie msg('opt-psql-noexec', $PSQL);
  1351. $res = qx{$PSQL --version};
  1352. $res =~ /psql\D+(\d+\.\d+)/ or ndie msg('opt-psql-nover');
  1353. our $psql_version = $1;
  1354.  
  1355. $VERBOSE >= 2 and warn qq{psql=$PSQL version=$psql_version\n};
  1356.  
  1357. $opt{defaultdb} = $psql_version >= 8.0 ? 'postgres' : 'template1';
  1358. $opt{defaultdb} = 'pgbouncer' if $action =~ /^pgb/;
  1359.  
  1360. ## Check the current database mode
  1361. our $STANDBY = 0;
  1362. our $MASTER = 0;
  1363. make_sure_standby_mode() if $opt{'assume-standby-mode'};
  1364. make_sure_prod() if $opt{'assume-prod'};
  1365.  
  1366. ## Create the list of databases we are going to connect to
  1367. my @targetdb = setup_target_databases();
  1368.  
  1369. sub add_response {
  1370.  
  1371. my ($type,$msg) = @_;
  1372.  
  1373. $db->{host} ||= '';
  1374.  
  1375. if ($STANDBY) {
  1376. $action_info->{$action}[0] = 1;
  1377. }
  1378.  
  1379. if ($nohost) {
  1380. push @{$type->{''}} => [$msg, length $nohost > 1 ? $nohost : ''];
  1381. return;
  1382. }
  1383.  
  1384. my $dbservice = $db->{dbservice};
  1385. my $dbname = defined $db->{dbname} ? qq{DB "$db->{dbname}"} : '';
  1386. my $dbhost = (!$db->{host} or $db->{host} eq '<none>') ? '' : qq{ (host:$db->{host})};
  1387. my $dbport = defined $db->{port} ? ($db->{port} eq $opt{defaultport} ? '' : qq{ (port=$db->{port}) }) : '';
  1388.  
  1389. ## Same_schema gets some different output
  1390. my $same_schema_header = '';
  1391. if ($action eq 'same_schema') {
  1392.  
  1393. ## Pretty display of what exactly those numbers mean!
  1394. my $number = 0;
  1395. my $historical = 0;
  1396. for my $row (@targetdb) {
  1397. $number++;
  1398. if (exists $row->{filename}) {
  1399. $historical = 1;
  1400. $same_schema_header .= sprintf "\nDB %s: File=%s\nDB %s: %s: %s %s: %s",
  1401. $number,
  1402. $row->{filename},
  1403. $number,
  1404. 'Creation date',
  1405. $row->{ctime},
  1406. 'CP version',
  1407. $row->{cversion};
  1408. }
  1409. $same_schema_header .= sprintf "\nDB %s: %s%s%s%s%s",
  1410. $number,
  1411. defined $row->{dbservice} ? qq{dbservice=$row->{dbservice} } : '',
  1412. defined $row->{port} ? qq{port=$row->{port} } : '',
  1413. defined $row->{host} ? qq{host=$row->{host} } : '',
  1414. defined $row->{dbname} ? qq{dbname=$row->{dbname} } : '',
  1415. defined $row->{dbuser} ? qq{user=$row->{dbuser} } : '';
  1416. $same_schema_header .= "\nDB $number: PG version: $row->{pgversion}";
  1417. $same_schema_header .= "\nDB $number: Total objects: $row->{objects}";
  1418. }
  1419.  
  1420. ## Databases
  1421. $number = 1;
  1422. my %dlist = map { ($_->{dbname} || ''), $number++; } @targetdb;
  1423. if (keys %dlist > 1 and ! $historical) {
  1424. my $dblist = join ',' => sort { $dlist{$a} <=> $dlist{$b} } keys %dlist;
  1425. $dbname = qq{ (databases:$dblist)};
  1426. }
  1427. ## Hosts
  1428. $number = 1;
  1429. my %hostlist = map { ($_->{host} || ''), $number++; } @targetdb;
  1430. if (keys %hostlist > 1 and ! $historical) {
  1431. my $dblist = join ',' => sort { $hostlist{$a} <=> $hostlist{$b} } keys %hostlist;
  1432. $dbhost = qq{ (hosts:$dblist)};
  1433. }
  1434. ## Ports
  1435. $number = 1;
  1436. my %portlist = map { ($_->{port} || ''), $number++; } @targetdb;
  1437. if (keys %portlist > 1 and ! $historical) {
  1438. my $dblist = join ',' => sort { $portlist{$a} <=> $portlist{$b} } keys %portlist;
  1439. $dbport = qq{ (ports:$dblist)};
  1440. }
  1441. }
  1442.  
  1443. my $header = sprintf q{%s%s%s%s},
  1444. ($action_info->{$action}[0] ? '' : (defined $dbservice and length $dbservice)) ?
  1445. qq{service=$dbservice} : $dbname,
  1446. (defined $db->{showschema} ? qq{ schema:$db->{showschema} } : ''),
  1447. $dbhost,
  1448. $dbport;
  1449. $header =~ s/\s+$//;
  1450. $header =~ s/^ //;
  1451. my $perf = ($opt{showtime} and $db->{totaltime} and $action ne 'bloat') ? "time=$db->{totaltime}s" : '';
  1452. if ($db->{perf}) {
  1453. $db->{perf} =~ s/^ +//;
  1454. if (length $same_schema_header) {
  1455. $db->{perf} =~ s/^\n//;
  1456. $db->{perf} = "$same_schema_header\n$db->{perf}";
  1457. }
  1458. $perf .= sprintf '%s%s', length($perf) ? ' ' : '', $db->{perf};
  1459. }
  1460.  
  1461. ## Strip trailing semicolons as allowed by the Nagios spec
  1462. ## But not for same_schema, where we might have (for example) a view def
  1463. if ($action ne 'same_schema') {
  1464. $perf =~ s/; / /;
  1465. $perf =~ s/;$//;
  1466. }
  1467.  
  1468. push @{$type->{$header}} => [$msg,$perf];
  1469.  
  1470. return;
  1471.  
  1472. } ## end of add_response
  1473.  
  1474.  
  1475. sub add_unknown {
  1476. my $msg = shift || $db->{error};
  1477. $msg =~ s/[\r\n]\s*/\\n /g;
  1478. $msg =~ s/\|/<PIPE>/g if $opt{showperf};
  1479. add_response \%unknown, $msg;
  1480. }
  1481. sub add_critical {
  1482. add_response \%critical, shift;
  1483. }
  1484. sub add_warning {
  1485. add_response \%warning, shift;
  1486. }
  1487. sub add_ok {
  1488. add_response \%ok, shift;
  1489. }
  1490.  
  1491.  
  1492. sub do_mrtg {
  1493. ## Hashref of info to pass out for MRTG or stat
  1494. my $arg = shift;
  1495. my $one = $arg->{one} || 0;
  1496. my $two = $arg->{two} || 0;
  1497. if ($SIMPLE) {
  1498. $one = $two if (length $two and $two > $one);
  1499. if ($opt{transform} eq 'KB' and $one =~ /^\d+$/) {
  1500. $one = int $one/(1024);
  1501. }
  1502. if ($opt{transform} eq 'MB' and $one =~ /^\d+$/) {
  1503. $one = int $one/(1024*1024);
  1504. }
  1505. elsif ($opt{transform} eq 'GB' and $one =~ /^\d+$/) {
  1506. $one = int $one/(1024*1024*1024);
  1507. }
  1508. elsif ($opt{transform} eq 'TB' and $one =~ /^\d+$/) {
  1509. $one = int $one/(1024*1024*1024*1024);
  1510. }
  1511. elsif ($opt{transform} eq 'EB' and $one =~ /^\d+$/) {
  1512. $one = int $one/(1024*1024*1024*1024*1024);
  1513. }
  1514. print "$one\n";
  1515. }
  1516. else {
  1517. my $uptime = $arg->{uptime} || '';
  1518. my $message = $arg->{msg} || '';
  1519. print "$one\n$two\n$uptime\n$message\n";
  1520. }
  1521. exit 0;
  1522. }
  1523.  
  1524.  
  1525. sub bad_mrtg {
  1526. my $msg = shift;
  1527. $ERROR and ndie $ERROR;
  1528. warn msgn('mrtg-fail', $action, $msg);
  1529. exit 3;
  1530. }
  1531.  
  1532.  
  1533. sub do_mrtg_stats {
  1534.  
  1535. ## Show the two highest items for mrtg stats hash
  1536.  
  1537. my $msg = shift;
  1538. defined $msg or ndie('unknown-error');
  1539.  
  1540. keys %stats or bad_mrtg($msg);
  1541. my ($one,$two) = ('','');
  1542. for (sort { $stats{$b} <=> $stats{$a} } keys %stats) {
  1543. if ($one eq '') {
  1544. $one = $stats{$_};
  1545. $msg = exists $statsmsg{$_} ? $statsmsg{$_} : "DB: $_";
  1546. next;
  1547. }
  1548. $two = $stats{$_};
  1549. last;
  1550. }
  1551. do_mrtg({one => $one, two => $two, msg => $msg});
  1552. }
  1553.  
  1554. sub make_sure_mode_is {
  1555.  
  1556. ## Requires $ENV{PGDATA} or --datadir
  1557.  
  1558. $db->{host} = '<none>';
  1559.  
  1560. ## Run pg_controldata, grab the mode
  1561. $res = open_controldata();
  1562.  
  1563. my $regex = msg('checkmode-state');
  1564. if ($res !~ /$regex\s*(.+)/) { ## no critic (ProhibitUnusedCapture)
  1565. ## Just in case, check the English one as well
  1566. $regex = msg_en('checkmode-state');
  1567. if ($res !~ /$regex\s*(.+)/) {
  1568. ndie msg('checkpoint-noregex');
  1569. }
  1570. }
  1571. my $last = $1;
  1572.  
  1573. return $last;
  1574.  
  1575. }
  1576.  
  1577. sub make_sure_standby_mode {
  1578.  
  1579. ## Checks if database in standby mode
  1580. ## Requires $ENV{PGDATA} or --datadir
  1581.  
  1582. my $last = make_sure_mode_is();
  1583.  
  1584. my $regex = msg('checkmode-recovery');
  1585. if ($last =~ /$regex/) {
  1586. $STANDBY = 1;
  1587. }
  1588.  
  1589. return;
  1590.  
  1591. } ## end of make_sure_standby_mode
  1592.  
  1593. sub make_sure_prod {
  1594.  
  1595. ## Checks if database in production mode
  1596. ## Requires $ENV{PGDATA} or --datadir
  1597.  
  1598. my $last = make_sure_mode_is();
  1599.  
  1600. my $regex = msg('checkmode-prod');
  1601. if ($last =~ /$regex/) {
  1602. $MASTER = 1;
  1603. }
  1604.  
  1605. return;
  1606.  
  1607. } ## end of make_sure_production_mode
  1608.  
  1609. sub finishup {
  1610.  
  1611. ## Final output
  1612. ## These are meant to be compact and terse: sometimes messages go to pagers
  1613.  
  1614. if ($MRTG) {
  1615. ## Try hard to ferret out a message in case we short-circuited here
  1616. my $msg = [[]];
  1617. if (keys %critical) {
  1618. ($msg) = values %critical;
  1619. }
  1620. elsif (keys %warning) {
  1621. ($msg) = values %warning;
  1622. }
  1623. elsif (keys %ok) {
  1624. ($msg) = values %ok;
  1625. }
  1626. elsif (keys %unknown) {
  1627. ($msg) = values %unknown;
  1628. }
  1629. do_mrtg_stats($msg->[0][0]);
  1630. }
  1631.  
  1632. $action =~ s/^\s*(\S+)\s*$/$1/;
  1633. my $service = sprintf "%s$action", $FANCYNAME ? 'postgres_' : '';
  1634. if (keys %critical or keys %warning or keys %ok or keys %unknown) {
  1635. ## If in quiet mode, print nothing if all is ok
  1636. if ($opt{quiet} and ! keys %critical and ! keys %warning and ! keys %unknown) {
  1637. }
  1638. else {
  1639. printf '%s ', $YELLNAME ? uc $service : $service;
  1640. }
  1641. }
  1642.  
  1643. sub dumpresult {
  1644. my ($type,$info) = @_;
  1645. my $SEP = ' * ';
  1646. ## Are we showing DEBUG_INFO?
  1647. my $showdebug = 0;
  1648. if ($DEBUGOUTPUT) {
  1649. $showdebug = 1 if $DEBUGOUTPUT =~ /a/io
  1650. or ($DEBUGOUTPUT =~ /c/io and $type eq 'c')
  1651. or ($DEBUGOUTPUT =~ /w/io and $type eq 'w')
  1652. or ($DEBUGOUTPUT =~ /o/io and $type eq 'o')
  1653. or ($DEBUGOUTPUT =~ /u/io and $type eq 'u');
  1654. }
  1655. for (sort keys %$info) {
  1656. printf '%s %s%s ',
  1657. $_,
  1658. $showdebug ? "[DEBUG: $DEBUG_INFO] " : '',
  1659. join $SEP => map { $_->[0] } @{$info->{$_}};
  1660. }
  1661. if ($opt{showperf}) {
  1662. my $pmsg = '';
  1663. for (sort keys %$info) {
  1664. my $m = sprintf '%s ', join ' ' => map { $_->[1] } @{$info->{$_}};
  1665. $pmsg .= $m;
  1666. }
  1667. $pmsg =~ s/^\s+//;
  1668. $pmsg and print "| $pmsg";
  1669. }
  1670. print "\n";
  1671.  
  1672. return;
  1673.  
  1674. }
  1675.  
  1676. if (keys %critical) {
  1677. print 'CRITICAL: ';
  1678. dumpresult(c => \%critical);
  1679. exit 2;
  1680. }
  1681. if (keys %warning) {
  1682. print 'WARNING: ';
  1683. dumpresult(w => \%warning);
  1684. exit 1;
  1685. }
  1686. if (keys %ok) {
  1687. ## We print nothing if in quiet mode
  1688. if (! $opt{quiet}) {
  1689. print 'OK: ';
  1690. dumpresult(o => \%ok);
  1691. }
  1692. exit 0;
  1693. }
  1694. if (keys %unknown) {
  1695. print 'UNKNOWN: ';
  1696. dumpresult(u => \%unknown);
  1697. exit 3;
  1698. }
  1699.  
  1700. die $USAGE;
  1701.  
  1702. } ## end of finishup
  1703.  
  1704.  
  1705. ## For options that take a size e.g. --critical="10 GB"
  1706. our $sizere = qr{^\s*(\d+\.?\d?)\s*([bkmgtepz])?\w*$}i; ## Don't care about the rest of the string
  1707.  
  1708. ## For options that take a time e.g. --critical="10 minutes" Fractions are allowed.
  1709. our $timere = qr{^\s*(\d+(?:\.\d+)?)\s*(\w*)\s*$}i;
  1710.  
  1711. ## For options that must be specified in seconds
  1712. our $timesecre = qr{^\s*(\d+)\s*(?:s(?:econd|ec)?)?s?\s*$};
  1713.  
  1714. ## For simple checksums:
  1715. our $checksumre = qr{^[a-f0-9]{32}$};
  1716.  
  1717. ## If in test mode, verify that we can run each requested action
  1718. our %testaction = (
  1719. autovac_freeze => 'VERSION: 8.2',
  1720. last_vacuum => 'ON: stats_row_level(<8.3) VERSION: 8.2',
  1721. last_analyze => 'ON: stats_row_level(<8.3) VERSION: 8.2',
  1722. last_autovacuum => 'ON: stats_row_level(<8.3) VERSION: 8.2',
  1723. last_autoanalyze => 'ON: stats_row_level(<8.3) VERSION: 8.2',
  1724. prepared_txns => 'VERSION: 8.1',
  1725. database_size => 'VERSION: 8.1',
  1726. disabled_triggers => 'VERSION: 8.1',
  1727. relation_size => 'VERSION: 8.1',
  1728. sequence => 'VERSION: 8.1',
  1729. table_size => 'VERSION: 8.1',
  1730. index_size => 'VERSION: 8.1',
  1731. query_time => 'VERSION: 8.1',
  1732. txn_time => 'VERSION: 8.3',
  1733. wal_files => 'VERSION: 8.1',
  1734. archive_ready => 'VERSION: 8.1',
  1735. fsm_pages => 'VERSION: 8.2 MAX: 8.3',
  1736. fsm_relations => 'VERSION: 8.2 MAX: 8.3',
  1737. hot_standby_delay => 'VERSION: 9.0',
  1738. listener => 'MAX: 8.4',
  1739. );
  1740. if ($opt{test}) {
  1741. print msgn('testmode-start');
  1742. my $info = run_command('SELECT name, setting FROM pg_settings');
  1743. my %set; ## port, host, name, user
  1744. for my $db (@{$info->{db}}) {
  1745. if (exists $db->{fail}) {
  1746. (my $err = $db->{error}) =~ s/\s*\n\s*/ \| /g;
  1747. print msgn('testmode-fail', $db->{pname}, $err);
  1748. next;
  1749. }
  1750. print msgn('testmode-ok', $db->{pname});
  1751. for (@{ $db->{slurp} }) {
  1752. $set{$_->{name}} = $_->{setting};
  1753. }
  1754. }
  1755. for my $ac (split /\s+/ => $action) {
  1756. my $limit = $testaction{lc $ac};
  1757. next if ! defined $limit;
  1758.  
  1759. if ($limit =~ /VERSION: ((\d+)\.(\d+))/) {
  1760. my ($rver,$rmaj,$rmin) = ($1,$2,$3);
  1761. for my $db (@{$info->{db}}) {
  1762. next unless exists $db->{ok};
  1763. if ($set{server_version} !~ /((\d+)\.(\d+))/) {
  1764. print msgn('testmode-nover', $db->{pname});
  1765. next;
  1766. }
  1767. my ($sver,$smaj,$smin) = ($1,$2,$3);
  1768. if ($smaj < $rmaj or ($smaj==$rmaj and $smin < $rmin)) {
  1769. print msgn('testmode-norun', $ac, $db->{pname}, $rver, $sver);
  1770. }
  1771. $db->{version} = $sver;
  1772. }
  1773. }
  1774.  
  1775. if ($limit =~ /MAX: ((\d+)\.(\d+))/) {
  1776. my ($rver,$rmaj,$rmin) = ($1,$2,$3);
  1777. for my $db (@{$info->{db}}) {
  1778. next unless exists $db->{ok};
  1779. if ($set{server_version} !~ /((\d+)\.(\d+))/) {
  1780. print msgn('testmode-nover', $db->{pname});
  1781. next;
  1782. }
  1783. my ($sver,$smaj,$smin) = ($1,$2,$3);
  1784. if ($smaj > $rmaj or ($smaj==$rmaj and $smin > $rmin)) {
  1785. print msgn('testmode-norun', $ac, $db->{pname}, $rver, $sver);
  1786. }
  1787. }
  1788. }
  1789.  
  1790. while ($limit =~ /\bON: (\w+)(?:\(([<>=])(\d+\.\d+)\))?/g) {
  1791. my ($setting,$op,$ver) = ($1,$2||'',$3||0);
  1792. for my $db (@{$info->{db}}) {
  1793. next unless exists $db->{ok};
  1794. if ($ver) {
  1795. next if $op eq '<' and $db->{version} >= $ver;
  1796. next if $op eq '>' and $db->{version} <= $ver;
  1797. next if $op eq '=' and $db->{version} != $ver;
  1798. }
  1799. my $val = $set{$setting};
  1800. if ($val ne 'on') {
  1801. print msgn('testmode-noset', $ac, $db->{pname}, $setting);
  1802. }
  1803. }
  1804. }
  1805. }
  1806. print msgn('testmode-end');
  1807. exit 0;
  1808. }
  1809.  
  1810. ## Expand the list of included/excluded users into a standard format
  1811. our $USERWHERECLAUSE = '';
  1812. if ($opt{includeuser}) {
  1813. my %userlist;
  1814. for my $user (@{$opt{includeuser}}) {
  1815. for my $u2 (split /,/ => $user) {
  1816. $userlist{$u2}++;
  1817. }
  1818. }
  1819. my $safename;
  1820. if (1 == keys %userlist) {
  1821. ($safename = each %userlist) =~ s/'/''/g;
  1822. $USERWHERECLAUSE = " AND usename = '$safename'";
  1823. }
  1824. else {
  1825. $USERWHERECLAUSE = ' AND usename IN (';
  1826. for my $user (sort keys %userlist) {
  1827. ($safename = $user) =~ s/'/''/g;
  1828. $USERWHERECLAUSE .= "'$safename',";
  1829. }
  1830. chop $USERWHERECLAUSE;
  1831. $USERWHERECLAUSE .= ')';
  1832. }
  1833. }
  1834. elsif ($opt{excludeuser}) {
  1835. my %userlist;
  1836. for my $user (@{$opt{excludeuser}}) {
  1837. for my $u2 (split /,/ => $user) {
  1838. $userlist{$u2}++;
  1839. }
  1840. }
  1841. my $safename;
  1842. if (1 == keys %userlist) {
  1843. ($safename = each %userlist) =~ s/'/''/g;
  1844. $USERWHERECLAUSE = " AND usename <> '$safename'";
  1845. }
  1846. else {
  1847. $USERWHERECLAUSE = ' AND usename NOT IN (';
  1848. for my $user (sort keys %userlist) {
  1849. ($safename = $user) =~ s/'/''/g;
  1850. $USERWHERECLAUSE .= "'$safename',";
  1851. }
  1852. chop $USERWHERECLAUSE;
  1853. $USERWHERECLAUSE .= ')';
  1854. }
  1855. }
  1856.  
  1857. ## Check number of connections, compare to max_connections
  1858. check_backends() if $action eq 'backends';
  1859.  
  1860. ## Table and index bloat
  1861. check_bloat() if $action eq 'bloat';
  1862.  
  1863. ## Simple connection, warning or critical options
  1864. check_connection() if $action eq 'connection';
  1865.  
  1866. ## Check the commitratio of one or more databases
  1867. check_commitratio() if $action eq 'commitratio';
  1868.  
  1869. ## Check the hitratio of one or more databases
  1870. check_hitratio() if $action eq 'hitratio';
  1871.  
  1872. ## Check the size of one or more databases
  1873. check_database_size() if $action eq 'database_size';
  1874.  
  1875. ## Check local disk_space - local means it must be run from the same box!
  1876. check_disk_space() if $action eq 'disk_space';
  1877.  
  1878. ## Check the size of relations, or more specifically, tables and indexes
  1879. check_index_size() if $action eq 'index_size';
  1880. check_table_size() if $action eq 'table_size';
  1881. check_relation_size() if $action eq 'relation_size';
  1882.  
  1883. ## Check how long since the last full analyze
  1884. check_last_analyze() if $action eq 'last_analyze';
  1885.  
  1886. ## Check how long since the last full vacuum
  1887. check_last_vacuum() if $action eq 'last_vacuum';
  1888.  
  1889. ## Check how long since the last AUTOanalyze
  1890. check_last_analyze('auto') if $action eq 'last_autoanalyze';
  1891.  
  1892. ## Check how long since the last full AUTOvacuum
  1893. check_last_vacuum('auto') if $action eq 'last_autovacuum';
  1894.  
  1895. ## Check that someone is listening for a specific thing
  1896. check_listener() if $action eq 'listener';
  1897.  
  1898. ## Check number and type of locks
  1899. check_locks() if $action eq 'locks';
  1900.  
  1901. ## Logfile is being written to
  1902. check_logfile() if $action eq 'logfile';
  1903.  
  1904. ## Known query finishes in a good amount of time
  1905. check_query_runtime() if $action eq 'query_runtime';
  1906.  
  1907. ## Check the length of running queries
  1908. check_query_time() if $action eq 'query_time';
  1909.  
  1910. ## Verify that the settings are what we think they should be
  1911. check_settings_checksum() if $action eq 'settings_checksum';
  1912.  
  1913. ## Compare DB time to localtime, alert on number of seconds difference
  1914. check_timesync() if $action eq 'timesync';
  1915.  
  1916. ## Check for transaction ID wraparound in all databases
  1917. check_txn_wraparound() if $action eq 'txn_wraparound';
  1918.  
  1919. ## Compare DB versions. warning = just major.minor, critical = full string
  1920. check_version() if $action eq 'version';
  1921.  
  1922. ## Check the number of WAL files. warning and critical are numbers
  1923. check_wal_files() if $action eq 'wal_files';
  1924.  
  1925. ## Check the number of WAL files ready to archive. warning and critical are numbers
  1926. check_archive_ready() if $action eq 'archive_ready';
  1927.  
  1928. ## Check the replication delay in hot standby setup
  1929. check_hot_standby_delay() if $action eq 'hot_standby_delay';
  1930.  
  1931. ## Check the maximum transaction age of all connections
  1932. check_txn_time() if $action eq 'txn_time';
  1933.  
  1934. ## Check the maximum age of idle in transaction connections
  1935. check_txn_idle() if $action eq 'txn_idle';
  1936.  
  1937. ## Run a custom query
  1938. check_custom_query() if $action eq 'custom_query';
  1939.  
  1940. ## Test of replication
  1941. check_replicate_row() if $action eq 'replicate_row';
  1942.  
  1943. ## Compare database schemas
  1944. check_same_schema() if $action eq 'same_schema';
  1945.  
  1946. ## Check sequence values
  1947. check_sequence() if $action eq 'sequence';
  1948.  
  1949. ## See how close we are to autovacuum_freeze_max_age
  1950. check_autovac_freeze() if $action eq 'autovac_freeze';
  1951.  
  1952. ## See how many pages we have used up compared to max_fsm_pages
  1953. check_fsm_pages() if $action eq 'fsm_pages';
  1954.  
  1955. ## See how many relations we have used up compared to max_fsm_relations
  1956. check_fsm_relations() if $action eq 'fsm_relations';
  1957.  
  1958. ## Spit back info from the pg_stat_database table. Cacti only
  1959. check_dbstats() if $action eq 'dbstats';
  1960.  
  1961. ## Check how long since the last checkpoint
  1962. check_checkpoint() if $action eq 'checkpoint';
  1963.  
  1964. ## Check the Database System Identifier
  1965. check_cluster_id() if $action eq 'cluster_id';
  1966.  
  1967. ## Check for disabled triggers
  1968. check_disabled_triggers() if $action eq 'disabled_triggers';
  1969.  
  1970. ## Check for any prepared transactions
  1971. check_prepared_txns() if $action eq 'prepared_txns';
  1972.  
  1973. ## Make sure Slony is behaving
  1974. check_slony_status() if $action eq 'slony_status';
  1975.  
  1976. ## Verify that the pgbouncer settings are what we think they should be
  1977. check_pgbouncer_checksum() if $action eq 'pgbouncer_checksum';
  1978.  
  1979. ## Check the number of active clients in each pgbouncer pool
  1980. check_pgb_pool('cl_active') if $action eq 'pgb_pool_cl_active';
  1981.  
  1982. ## Check the number of waiting clients in each pgbouncer pool
  1983. check_pgb_pool('cl_waiting') if $action eq 'pgb_pool_cl_waiting';
  1984.  
  1985. ## Check the number of active server connections in each pgbouncer pool
  1986. check_pgb_pool('sv_active') if $action eq 'pgb_pool_sv_active';
  1987.  
  1988. ## Check the number of idle server connections in each pgbouncer pool
  1989. check_pgb_pool('sv_idle') if $action eq 'pgb_pool_sv_idle';
  1990.  
  1991. ## Check the number of used server connections in each pgbouncer pool
  1992. check_pgb_pool('sv_used') if $action eq 'pgb_pool_sv_used';
  1993.  
  1994. ## Check the number of tested server connections in each pgbouncer pool
  1995. check_pgb_pool('sv_tested') if $action eq 'pgb_pool_sv_tested';
  1996.  
  1997. ## Check the number of login server connections in each pgbouncer pool
  1998. check_pgb_pool('sv_login') if $action eq 'pgb_pool_sv_login';
  1999.  
  2000. ## Check the current maximum wait time for client connections in pgbouncer pools
  2001. check_pgb_pool('maxwait') if $action eq 'pgb_pool_maxwait';
  2002.  
  2003. ## Check how many clients are connected to pgbouncer compared to max_client_conn.
  2004. check_pgbouncer_backends() if $action eq 'pgbouncer_backends';
  2005.  
  2006. check_pgagent_jobs() if $action eq 'pgagent_jobs';
  2007.  
  2008. ##
  2009. ## Everything past here does not hit a Postgres database
  2010. ##
  2011. $nohost = 1;
  2012.  
  2013. ## Check for new versions of check_postgres.pl
  2014. check_new_version_cp() if $action eq 'new_version_cp';
  2015.  
  2016. ## Check for new versions of Postgres
  2017. check_new_version_pg() if $action eq 'new_version_pg';
  2018.  
  2019. ## Check for new versions of Bucardo
  2020. check_new_version_bc() if $action eq 'new_version_bc';
  2021.  
  2022. ## Check for new versions of boxinfo
  2023. check_new_version_box() if $action eq 'new_version_box';
  2024.  
  2025. ## Check for new versions of tail_n_mail
  2026. check_new_version_tnm() if $action eq 'new_version_tnm';
  2027.  
  2028. finishup();
  2029.  
  2030. exit 0;
  2031.  
  2032.  
  2033. sub build_symlinks {
  2034.  
  2035. ## Create symlinks to most actions
  2036. $ME =~ /postgres/
  2037. or die msgn('symlink-name');
  2038.  
  2039. my $force = $action =~ /force/ ? 1 : 0;
  2040. for my $action (sort keys %$action_info) {
  2041. my $space = ' ' x ($longname - length $action);
  2042. my $file = "check_postgres_$action";
  2043. if (-l $file) {
  2044. if (!$force) {
  2045. my $source = readlink $file;
  2046. print msgn('symlink-done', $file, $space, $source);
  2047. next;
  2048. }
  2049. print msg('symlink-unlink', $file, $space);
  2050. unlink $file or die msgn('symlink-fail1', $file, $!);
  2051. }
  2052. elsif (-e $file) {
  2053. print msgn('symlink-exists', $file, $space);
  2054. next;
  2055. }
  2056.  
  2057. if (symlink $0, $file) {
  2058. print msgn('symlink-create', $file);
  2059. }
  2060. else {
  2061. print msgn('symlink-fail2', $file, $ME, $!);
  2062. }
  2063. }
  2064.  
  2065. exit 0;
  2066.  
  2067. } ## end of build_symlinks
  2068.  
  2069.  
  2070. sub pretty_size {
  2071.  
  2072. ## Transform number of bytes to a SI display similar to Postgres' format
  2073.  
  2074. my $bytes = shift;
  2075. my $rounded = shift || 0;
  2076.  
  2077. return "$bytes bytes" if $bytes < 10240;
  2078.  
  2079. my @unit = qw/kB MB GB TB PB EB YB ZB/;
  2080.  
  2081. for my $p (1..@unit) {
  2082. if ($bytes <= 1024**$p) {
  2083. $bytes /= (1024**($p-1));
  2084. return $rounded ?
  2085. sprintf ('%d %s', $bytes, $unit[$p-2]) :
  2086. sprintf ('%.2f %s', $bytes, $unit[$p-2]);
  2087. }
  2088. }
  2089.  
  2090. return $bytes;
  2091.  
  2092. } ## end of pretty_size
  2093.  
  2094.  
  2095. sub pretty_time {
  2096.  
  2097. ## Transform number of seconds to a more human-readable format
  2098. ## First argument is number of seconds
  2099. ## Second optional arg is highest transform: s,m,h,d,w
  2100. ## If uppercase, it indicates to "round that one out"
  2101.  
  2102. my $sec = shift;
  2103. my $tweak = shift || '';
  2104.  
  2105. ## Just seconds (< 2:00)
  2106. if ($sec < 120 or $tweak =~ /s/) {
  2107. return sprintf "$sec %s", $sec==1 ? msg('time-second') : msg('time-seconds');
  2108. }
  2109.  
  2110. ## Minutes and seconds (< 60:00)
  2111. if ($sec < 60*60 or $tweak =~ /m/) {
  2112. my $min = int $sec / 60;
  2113. $sec %= 60;
  2114. my $ret = sprintf "$min %s", $min==1 ? msg('time-minute') : msg('time-minutes');
  2115. $sec and $tweak !~ /S/ and $ret .= sprintf " $sec %s", $sec==1 ? msg('time-second') : msg('time-seconds');
  2116. return $ret;
  2117. }
  2118.  
  2119. ## Hours, minutes, and seconds (< 48:00:00)
  2120. if ($sec < 60*60*24*2 or $tweak =~ /h/) {
  2121. my $hour = int $sec / (60*60);
  2122. $sec -= ($hour*60*60);
  2123. my $min = int $sec / 60;
  2124. $sec -= ($min*60);
  2125. my $ret = sprintf "$hour %s", $hour==1 ? msg('time-hour') : msg('time-hours');
  2126. $min and $tweak !~ /M/ and $ret .= sprintf " $min %s", $min==1 ? msg('time-minute') : msg('time-minutes');
  2127. $sec and $tweak !~ /[SM]/ and $ret .= sprintf " $sec %s", $sec==1 ? msg('time-second') : msg('time-seconds');
  2128. return $ret;
  2129. }
  2130.  
  2131. ## Days, hours, minutes, and seconds (< 28 days)
  2132. if ($sec < 60*60*24*28 or $tweak =~ /d/) {
  2133. my $day = int $sec / (60*60*24);
  2134. $sec -= ($day*60*60*24);
  2135. my $our = int $sec / (60*60);
  2136. $sec -= ($our*60*60);
  2137. my $min = int $sec / 60;
  2138. $sec -= ($min*60);
  2139. my $ret = sprintf "$day %s", $day==1 ? msg('time-day') : msg('time-days');
  2140. $our and $tweak !~ /H/ and $ret .= sprintf " $our %s", $our==1 ? msg('time-hour') : msg('time-hours');
  2141. $min and $tweak !~ /[HM]/ and $ret .= sprintf " $min %s", $min==1 ? msg('time-minute') : msg('time-minutes');
  2142. $sec and $tweak !~ /[HMS]/ and $ret .= sprintf " $sec %s", $sec==1 ? msg('time-second') : msg('time-seconds');
  2143. return $ret;
  2144. }
  2145.  
  2146. ## Weeks, days, hours, minutes, and seconds (< 28 days)
  2147. my $week = int $sec / (60*60*24*7);
  2148. $sec -= ($week*60*60*24*7);
  2149. my $day = int $sec / (60*60*24);
  2150. $sec -= ($day*60*60*24);
  2151. my $our = int $sec / (60*60);
  2152. $sec -= ($our*60*60);
  2153. my $min = int $sec / 60;
  2154. $sec -= ($min*60);
  2155. my $ret = sprintf "$week %s", $week==1 ? msg('time-week') : msg('time-weeks');
  2156. $day and $tweak !~ /D/ and $ret .= sprintf " $day %s", $day==1 ? msg('time-day') : msg('time-days');
  2157. $our and $tweak !~ /[DH]/ and $ret .= sprintf " $our %s", $our==1 ? msg('time-hour') : msg('time-hours');
  2158. $min and $tweak !~ /[DHM]/ and $ret .= sprintf " $min %s", $min==1 ? msg('time-minute') : msg('time-minutes');
  2159. $sec and $tweak !~ /[DHMS]/ and $ret .= sprintf " $sec %s", $sec==1 ? msg('time-second') : msg('time-seconds');
  2160. return $ret;
  2161.  
  2162. } ## end of pretty_time
  2163.  
  2164.  
  2165. sub run_command {
  2166.  
  2167. ## Run a command string against each of our databases using psql
  2168. ## Optional args in a hashref:
  2169. ## "failok" - don't report if we failed
  2170. ## "fatalregex" - allow this FATAL regex through
  2171. ## "target" - use this targetlist instead of generating one
  2172. ## "timeout" - change the timeout from the default of $opt{timeout}
  2173. ## "regex" - the query must match this or we throw an error
  2174. ## "emptyok" - it's okay to not match any rows at all
  2175. ## "version" - alternate SQL for different versions of Postgres
  2176. ## "dbnumber" - connect with this specific entry from @targetdb
  2177. ## "conninfo" - return the connection information string without doing anything
  2178.  
  2179. my $string = shift || '';
  2180. my $arg = shift || {};
  2181. my $info = { command => $string, db => [], hosts => 0 };
  2182.  
  2183. ## First of all check if the server in standby mode, if so end this
  2184. ## with OK status.
  2185.  
  2186. if ($STANDBY) {
  2187. $db->{'totaltime'} = '0.00';
  2188. add_ok msg('mode-standby');
  2189. if ($MRTG) {
  2190. do_mrtg({one => 1});
  2191. }
  2192. finishup();
  2193. exit 0;
  2194. }
  2195.  
  2196. $VERBOSE >= 3 and warn qq{Starting run_command with: $string\n};
  2197.  
  2198. my (%host,$passfile,$passfh,$tempdir,$tempfile,$tempfh,$errorfile,$errfh);
  2199. my $offset = -1;
  2200.  
  2201. ## The final list of targets has been set inside @targetdb
  2202.  
  2203. if (! @targetdb) {
  2204. ndie msg('runcommand-nodb');
  2205. }
  2206.  
  2207. ## Create a temp file to store our results
  2208. my @tempdirargs = (CLEANUP => 1);
  2209. if ($opt{tempdir}) {
  2210. push @tempdirargs => 'DIR', $opt{tempdir};
  2211. }
  2212.  
  2213. $tempdir = tempdir(@tempdirargs);
  2214. ($tempfh,$tempfile) = tempfile('check_postgres_psql.XXXXXXX', SUFFIX => '.tmp', DIR => $tempdir);
  2215.  
  2216. ## Create another one to catch any errors
  2217. ($errfh,$errorfile) = tempfile('check_postgres_psql_stderr.XXXXXXX', SUFFIX => '.tmp', DIR => $tempdir);
  2218.  
  2219. ## Mild cleanup of the query
  2220. $string =~ s/^\s*(.+?)\s*$/$1/s;
  2221.  
  2222. ## Set a statement_timeout, as a last-ditch safety measure
  2223. my $timeout = $arg->{timeout} || $opt{timeout};
  2224. my $dbtimeout = $timeout * 1000;
  2225. if ($action !~ /^pgb/) {
  2226. $string = "BEGIN;SET statement_timeout=$dbtimeout;COMMIT;$string";
  2227. }
  2228.  
  2229. ## Keep track of which database we are on, to allow dbnumber to work
  2230. my $num = 0;
  2231.  
  2232. ## Loop through and run the command on each target database
  2233. for $db (@targetdb) {
  2234.  
  2235. ## Skip this one if we are using dbnumber and this is not our choice
  2236. $num++;
  2237. if ($arg->{dbnumber} and $arg->{dbnumber} != $num) {
  2238. next;
  2239. }
  2240.  
  2241. ## Just to keep things clean:
  2242. truncate $tempfh, 0;
  2243. truncate $errfh, 0;
  2244.  
  2245. ## Store this target in the global target list
  2246. push @{$info->{db}}, $db;
  2247.  
  2248. my @args = ('-q', '-t');
  2249. if (defined $db->{dbservice} and length $db->{dbservice}) { ## XX Check for simple names
  2250. $db->{pname} = "service=$db->{dbservice}";
  2251. $ENV{PGSERVICE} = $db->{dbservice};
  2252. }
  2253. else {
  2254. $db->{pname} = "port=$db->{port} host=$db->{host} db=$db->{dbname} user=$db->{dbuser}";
  2255. }
  2256.  
  2257. ## If all we want is a connection string, give it and leave now
  2258. if ($arg->{conninfo}) {
  2259. return $db->{pname};
  2260. }
  2261.  
  2262. defined $db->{dbname} and push @args, '-d', $db->{dbname};
  2263. defined $db->{dbuser} and push @args, '-U', $db->{dbuser};
  2264. defined $db->{port} and push @args => '-p', $db->{port};
  2265. if ($db->{host} ne '<none>') {
  2266. push @args => '-h', $db->{host};
  2267. $host{$db->{host}}++; ## For the overall count
  2268. }
  2269.  
  2270. if (defined $db->{dbpass} and length $db->{dbpass}) {
  2271. ## Make a custom PGPASSFILE. Far better to simply use your own .pgpass of course
  2272. ($passfh,$passfile) = tempfile('check_postgres.XXXXXXXX', SUFFIX => '.tmp', DIR => $tempdir);
  2273. $VERBOSE >= 3 and warn msgn('runcommand-pgpass', $passfile);
  2274. $ENV{PGPASSFILE} = $passfile;
  2275. printf $passfh "%s:%s:%s:%s:%s\n",
  2276. $db->{host} eq '<none>' ? '*' : $db->{host}, $db->{port}, $db->{dbname}, $db->{dbuser}, $db->{dbpass};
  2277. close $passfh or ndie msg('file-noclose', $passfile, $!);
  2278. }
  2279.  
  2280. push @args, '-o', $tempfile;
  2281. push @args => '-x';
  2282.  
  2283. ## If we've got different SQL, use this first run to simply grab the version
  2284. ## Then we'll use that info to pick the real query
  2285. if ($arg->{version}) {
  2286. if (!$db->{version}) {
  2287. $arg->{versiononly} = 1;
  2288. $arg->{oldstring} = $string;
  2289. $string = 'SELECT version()';
  2290. }
  2291. else {
  2292. $string = $arg->{oldstring} || $arg->{string};
  2293. for my $row (@{$arg->{version}}) {
  2294. if ($row !~ s/^([<>]?)(\d+\.\d+)\s+//) {
  2295. ndie msg('die-badversion', $row);
  2296. }
  2297. my ($mod,$ver) = ($1||'',$2);
  2298. if ($mod eq '>' and $db->{version} > $ver) {
  2299. $string = $row;
  2300. last;
  2301. }
  2302. if ($mod eq '<' and $db->{version} < $ver) {
  2303. $string = $row;
  2304. last;
  2305. }
  2306. if ($mod eq '' and $db->{version} eq $ver) {
  2307. $string = $row;
  2308. }
  2309. }
  2310. delete $arg->{version};
  2311. $info->{command} = $string;
  2312. }
  2313. }
  2314.  
  2315. local $SIG{ALRM} = sub { die 'Timed out' };
  2316. alarm 0;
  2317.  
  2318. push @args, '-c', $string;
  2319.  
  2320. $VERBOSE >= 3 and warn Dumper \@args;
  2321.  
  2322. my $start = $opt{showtime} ? [gettimeofday()] : 0;
  2323. open my $oldstderr, '>&', \*STDERR or ndie msg('runcommand-nodupe');
  2324. open STDERR, '>', $errorfile or ndie msg('runcommand-noerr');
  2325. eval {
  2326. alarm $timeout;
  2327. $res = system $PSQL => @args;
  2328. };
  2329. my $err = $@;
  2330. alarm 0;
  2331. open STDERR, '>&', $oldstderr or ndie msg('runcommand-noerr');
  2332. close $oldstderr or ndie msg('file-noclose', 'STDERR copy', $!);
  2333. if ($err) {
  2334. if ($err =~ /Timed out/) {
  2335. ndie msg('runcommand-timeout', $timeout);
  2336. }
  2337. else {
  2338. ndie msg('runcommand-err');
  2339. }
  2340. }
  2341.  
  2342. $db->{totaltime} = sprintf '%.2f', $opt{showtime} ? tv_interval($start) : 0;
  2343.  
  2344. if ($res) {
  2345. $db->{fail} = $res;
  2346. $VERBOSE >= 3 and !$arg->{failok} and warn msgn('runcommand-nosys', $res);
  2347. seek $errfh, 0, 0;
  2348. {
  2349. local $/;
  2350. $db->{error} = <$errfh> || '';
  2351. $db->{error} =~ s/\s*$//;
  2352. $db->{error} =~ s/^psql: //;
  2353. $ERROR = $db->{error};
  2354. }
  2355.  
  2356. if ($db->{error} =~ /FATAL/) {
  2357. ## If we are just trying to connect, this should be a normal error
  2358. if ($action eq 'connection') {
  2359. $info->{fatal} = 1;
  2360. return $info;
  2361. }
  2362.  
  2363. if (exists $arg->{fatalregex} and $db->{error} =~ /$arg->{fatalregex}/) {
  2364. $info->{fatalregex} = $db->{error};
  2365. next;
  2366. }
  2367. else {
  2368. ndie "$db->{error}";
  2369. }
  2370. }
  2371.  
  2372. elsif ($db->{error} =~ /statement timeout/) {
  2373. ndie msg('runcommand-timeout', $timeout);
  2374. }
  2375.  
  2376. if ($db->{fail} and !$arg->{failok} and !$arg->{noverify}) {
  2377.  
  2378. ## Check if problem is due to backend being too old for this check
  2379. verify_version();
  2380.  
  2381. if (exists $db->{error}) {
  2382. ndie $db->{error};
  2383. }
  2384.  
  2385. add_unknown;
  2386. ## Remove it from the returned hash
  2387. pop @{$info->{db}};
  2388. }
  2389. }
  2390. else {
  2391. seek $tempfh, 0, 0;
  2392. {
  2393. local $/;
  2394. $db->{slurp} = <$tempfh>;
  2395. }
  2396. $db->{ok} = 1;
  2397.  
  2398. ## Unfortunately, psql outputs "(No rows)" even with -t and -x
  2399. $db->{slurp} = '' if ! defined $db->{slurp} or index($db->{slurp},'(')==0;
  2400.  
  2401. ## Allow an empty query (no matching rows) if requested
  2402. if ($arg->{emptyok} and $db->{slurp} =~ /^\s*$/o) {
  2403. $arg->{emptyok2} = 1;
  2404. }
  2405. ## If we just want a version, grab it and redo
  2406. if ($arg->{versiononly}) {
  2407. if ($db->{error}) {
  2408. ndie $db->{error};
  2409. }
  2410. if ($db->{slurp} !~ /(\d+\.\d+)/) {
  2411. ndie msg('die-badversion', $db->{slurp});
  2412. }
  2413. $db->{version} = $1;
  2414. $db->{ok} = 0;
  2415. delete $arg->{versiononly};
  2416. ## Remove this from the returned hash
  2417. pop @{$info->{db}};
  2418. redo;
  2419. }
  2420.  
  2421. ## If we were provided with a regex, check and bail if it fails
  2422. if ($arg->{regex} and ! $arg->{emptyok2}) {
  2423. if ($db->{slurp} !~ $arg->{regex}) {
  2424. ## Check if problem is due to backend being too old for this check
  2425.  
  2426. verify_version();
  2427.  
  2428. add_unknown msg('invalid-query', $db->{slurp});
  2429.  
  2430. finishup();
  2431. exit 0;
  2432. }
  2433. }
  2434.  
  2435. ## Transform psql output into an arrayref of hashes
  2436. my @stuff;
  2437. my $lnum = 0;
  2438. my $lastval;
  2439. for my $line (split /\n/ => $db->{slurp}) {
  2440.  
  2441. if (index($line,'-')==0) {
  2442. $lnum++;
  2443. next;
  2444. }
  2445. if ($line =~ /^([\?\w]+)\s+\| (.*)/) {
  2446. $stuff[$lnum]{$1} = $2;
  2447. $lastval = $1;
  2448. }
  2449. elsif ($line =~ /^QUERY PLAN\s+\| (.*)/) {
  2450. $stuff[$lnum]{queryplan} = $1;
  2451. $lastval = 'queryplan';
  2452. }
  2453. elsif ($line =~ /^\s+: (.*)/) {
  2454. $stuff[$lnum]{$lastval} .= "\n$1";
  2455. }
  2456. elsif ($line =~ /^\s+\| (.+)/) {
  2457. $stuff[$lnum]{$lastval} .= "\n$1";
  2458. }
  2459. ## No content: can happen in the source of functions, for example
  2460. elsif ($line =~ /^\s+\|\s+$/) {
  2461. $stuff[$lnum]{$lastval} .= "\n";
  2462. }
  2463. else {
  2464. my $msg = msg('no-parse-psql');
  2465. warn "$msg\n";
  2466. $msg = msg('bug-report');
  2467. warn "$msg\n";
  2468. my $cline = (caller)[2];
  2469. my $args = join ' ' => @args;
  2470. warn "Version: $VERSION\n";
  2471. warn "Action: $action\n";
  2472. warn "Calling line: $cline\n";
  2473. warn "Output: $line\n";
  2474. $args =~ s/ -c (.+)/-c "$1"/s;
  2475. warn "Command: $PSQL $args\n";
  2476. ## Last thing is to see if we can grab the PG version
  2477. if (! $opt{stop_looping}) {
  2478. ## Just in case...
  2479. $opt{stop_looping} = 1;
  2480. my $linfo = run_command('SELECT version() AS version');
  2481. (my $v = $linfo->{db}[0]{slurp}[0]{version}) =~ s/(\w+ \S+).+/$1/;
  2482. warn "Postgres version: $v\n";
  2483. }
  2484. exit 1;
  2485. }
  2486. }
  2487. $db->{slurp} = \@stuff;
  2488. } ## end valid system call
  2489.  
  2490. } ## end each database
  2491.  
  2492. close $errfh or ndie msg('file-noclose', $errorfile, $!);
  2493. close $tempfh or ndie msg('file-noclose', $tempfile, $!);
  2494.  
  2495. eval { File::Temp::cleanup(); };
  2496.  
  2497. $info->{hosts} = keys %host;
  2498.  
  2499. $VERBOSE >= 3 and warn Dumper $info;
  2500.  
  2501. if ($DEBUGOUTPUT) {
  2502. if (defined $info->{db} and defined $info->{db}[0]{slurp}) {
  2503. $DEBUG_INFO = $info->{db}[0]{slurp};
  2504. $DEBUG_INFO =~ s/\n/\\n/g;
  2505. $DEBUG_INFO =~ s/\|/<SEP>/g;
  2506. }
  2507. }
  2508.  
  2509. return $info;
  2510.  
  2511. } ## end of run_command
  2512.  
  2513.  
  2514. sub setup_target_databases {
  2515.  
  2516. ## Build a list of all databases to connect to.
  2517. ## Returns a list of all such databases with connection information:
  2518. ## -- dbuser, --dbpass, --dbservice, --port, --dbname, --host
  2519. ##
  2520. ## Items are determined by host, port, and db arguments
  2521. ## Multi-args are grouped together: host, port, dbuser, dbpass
  2522. ## Groups are kept together for first pass
  2523. ## The final arg in a group is passed on
  2524. ##
  2525. ## Examples:
  2526. ## --host=a,b --port=5433 --db=c
  2527.  
  2528. ## Connects twice to port 5433, using database c, to hosts a and b
  2529. ## a-5433-c b-5433-c
  2530. ##
  2531. ## --host=a,b --port=5433 --db=c,d
  2532. ## Connects four times: a-5433-c a-5433-d b-5433-c b-5433-d
  2533. ##
  2534. ## --host=a,b --host=foo --port=1234 --port=5433 --db=e,f
  2535. ## Connects six times: a-1234-e a-1234-f b-1234-e b-1234-f foo-5433-e foo-5433-f
  2536. ##
  2537. ## --host=a,b --host=x --port=5432,5433 --dbuser=alice --dbuser=bob --db=baz
  2538. ## Connects three times: a-5432-alice-baz b-5433-alice-baz x-5433-bob-baz
  2539.  
  2540. ## Returns a list of targets as a hashref
  2541.  
  2542. my $arg = shift || {};
  2543.  
  2544. ## The final list of targets:
  2545. my @target;
  2546.  
  2547. ## Default connection options
  2548. my $conn =
  2549. {
  2550. host => [$ENV{PGHOST} || '<none>'],
  2551. port => [$ENV{PGPORT} || $opt{defaultport}],
  2552. dbname => [$ENV{PGDATABASE} || $opt{defaultdb}],
  2553. dbuser => [$ENV{PGUSER} || $opt{defaultuser}],
  2554. dbpass => [$ENV{PGPASSWORD} || ''],
  2555. dbservice => [''],
  2556. };
  2557.  
  2558. ## Don't set any default values if a service is being used
  2559. if (defined $opt{dbservice} and defined $opt{dbservice}->[0] and length $opt{dbservice}->[0]) {
  2560. $conn->{dbname} = [];
  2561. $conn->{port} = [];
  2562. $conn->{dbuser} = [];
  2563. }
  2564.  
  2565. ## If we were passed in a target, use that and move on
  2566. if (exists $arg->{target}) {
  2567. ## Make a copy, in case we are passed in a ref
  2568. my $newtarget;
  2569. for my $key (keys %$conn) {
  2570. $newtarget->{$key} = exists $arg->{target}{$key} ? $arg->{target}{$key} : $conn->{$key};
  2571. }
  2572. return [$newtarget];
  2573. }
  2574.  
  2575. ## Global count of total places we are connecting to
  2576. ## We don't mess with this if using {target} above
  2577. $opt{numdbs} = 0;
  2578.  
  2579. ## The current group number we are looking at
  2580. my $group_num = 0;
  2581.  
  2582. GROUP: {
  2583.  
  2584. ## This level controls a "group" of targets
  2585.  
  2586. ## Start bubbling all our targets into other stuff
  2587. my %group;
  2588. my $found_new_var = 0;
  2589.  
  2590. for my $v (keys %$conn) { ## For each connection var such as port, host...
  2591. my $vname = $v;
  2592.  
  2593. ## Check if something exists at the current slot number for this var
  2594. if (defined $opt{$v}->[$group_num]) {
  2595.  
  2596. my $new = $opt{$v}->[$group_num];
  2597.  
  2598. ## Strip out whitespace unless this is a service or host
  2599. $new =~ s/\s+//g unless $vname eq 'dbservice' or $vname eq 'host';
  2600.  
  2601. ## Set this as the new default for this connection var moving forward
  2602. $conn->{$vname} = [split /,/ => $new];
  2603.  
  2604. ## Make a note that we found something new this round
  2605. $found_new_var = 1;
  2606. }
  2607.  
  2608. $group{$vname} = $conn->{$vname};
  2609. }
  2610.  
  2611. ## If we found nothing new, we must be done building our groups
  2612. last GROUP if ! $found_new_var and @target;
  2613.  
  2614. $group_num++;
  2615.  
  2616. ## Now break the newly created group into individual targets
  2617. my $tbin = 0;
  2618. TARGET: {
  2619. my $foundtarget = 0;
  2620. my %temptarget;
  2621. for my $g (keys %group) {
  2622. if (defined $group{$g}->[$tbin]) {
  2623. $conn->{$g} = [$group{$g}->[$tbin]];
  2624. $foundtarget = 1;
  2625. }
  2626. $temptarget{$g} = $conn->{$g}[0];
  2627. }
  2628.  
  2629. ## Leave if nothing new
  2630. last TARGET if ! $foundtarget;
  2631.  
  2632. ## Add to our master list
  2633. push @target => \%temptarget;
  2634.  
  2635. $tbin++;
  2636.  
  2637. redo TARGET;
  2638.  
  2639. } ## end TARGET
  2640.  
  2641. last GROUP if ! $found_new_var;
  2642.  
  2643. redo GROUP;
  2644.  
  2645. } ## end GROUP
  2646.  
  2647. return @target;
  2648.  
  2649. } ## end of setup_target_databases
  2650.  
  2651.  
  2652. sub verify_version {
  2653.  
  2654. ## Check if the backend can handle the current action
  2655. my $limit = $testaction{lc $action} || '';
  2656.  
  2657. my $versiononly = shift || 0;
  2658.  
  2659. return if ! $limit and ! $versiononly;
  2660.  
  2661. ## We almost always need the version, so just grab it for any limitation
  2662. $SQL = q{SELECT setting FROM pg_settings WHERE name = 'server_version'};
  2663. my $oldslurp = $db->{slurp} || '';
  2664. my $info = run_command($SQL, {noverify => 1});
  2665. if (defined $info->{db}[0]
  2666. and exists $info->{db}[0]{error}
  2667. and defined $info->{db}[0]{error}
  2668. ) {
  2669. ndie $info->{db}[0]{error};
  2670. }
  2671.  
  2672. if (!defined $info->{db}[0] or $info->{db}[0]{slurp}[0]{setting} !~ /((\d+)\.(\d+))/) {
  2673. ndie msg('die-badversion', $SQL);
  2674. }
  2675. my ($sver,$smaj,$smin) = ($1,$2,$3);
  2676.  
  2677. if ($versiononly) {
  2678. return $sver;
  2679. }
  2680.  
  2681. if ($limit =~ /VERSION: ((\d+)\.(\d+))/) {
  2682. my ($rver,$rmaj,$rmin) = ($1,$2,$3);
  2683. if ($smaj < $rmaj or ($smaj==$rmaj and $smin < $rmin)) {
  2684. ndie msg('die-action-version', $action, $rver, $sver);
  2685. }
  2686. }
  2687.  
  2688. while ($limit =~ /\bON: (\w+)(?:\(([<>=])(\d+\.\d+)\))?/g) {
  2689. my ($setting,$op,$ver) = ($1,$2||'',$3||0);
  2690. if ($ver) {
  2691. next if $op eq '<' and $sver >= $ver;
  2692. next if $op eq '>' and $sver <= $ver;
  2693. next if $op eq '=' and $sver != $ver;
  2694. }
  2695.  
  2696. $SQL = qq{SELECT setting FROM pg_settings WHERE name = '$setting'};
  2697. my $info2 = run_command($SQL);
  2698. if (!defined $info2->{db}[0]) {
  2699. ndie msg('die-nosetting', $setting);
  2700. }
  2701. my $val = $info2->{db}[0]{slurp}[0]{setting};
  2702. if ($val !~ /^\s*on\b/) {
  2703. ndie msg('die-noset', $action, $setting);
  2704. }
  2705. }
  2706.  
  2707. $db->{slurp} = $oldslurp;
  2708. return;
  2709.  
  2710. } ## end of verify_version
  2711.  
  2712.  
  2713. sub size_in_bytes { ## no critic (RequireArgUnpacking)
  2714.  
  2715. ## Given a number and a unit, return the number of bytes.
  2716. ## Defaults to bytes
  2717.  
  2718. my ($val,$unit) = ($_[0],lc substr($_[1]||'s',0,1));
  2719. return $val * ($unit eq 'b' ? 1 : $unit eq 'k' ? 1024 : $unit eq 'm' ? 1024**2 :
  2720. $unit eq 'g' ? 1024**3 : $unit eq 't' ? 1024**4 :
  2721. $unit eq 'p' ? 1024**5 : $unit eq 'e' ? 1024**6 :
  2722. $unit eq 'z' ? 1024**7 : 1);
  2723.  
  2724. } ## end of size_in_bytes
  2725.  
  2726.  
  2727. sub size_in_seconds {
  2728.  
  2729. my ($string,$type) = @_;
  2730.  
  2731. return '' if ! length $string;
  2732. if ($string !~ $timere) {
  2733. ndie msg('die-badtime', $type, substr($type,0,1));
  2734. }
  2735. my ($val,$unit) = ($1,lc substr($2||'s',0,1));
  2736. my $tempval = sprintf '%.9f', $val * (
  2737. $unit eq 's' ? 1 :
  2738. $unit eq 'm' ? 60 :
  2739. $unit eq 'h' ? 3600 :
  2740. $unit eq 'd' ? 86400 :
  2741. $unit eq 'w' ? 604800 :
  2742. $unit eq 'y' ? 31536000 :
  2743. ndie msg('die-badtime', $type, substr($type,0,1))
  2744. );
  2745. $tempval =~ s/0+$//;
  2746. $tempval = int $tempval if $tempval =~ /\.$/;
  2747. return $tempval;
  2748.  
  2749. } ## end of size_in_seconds
  2750.  
  2751.  
  2752. sub skip_item {
  2753.  
  2754. ## Determine if something should be skipped due to inclusion/exclusion options
  2755. ## Exclusion checked first: inclusion can pull it back in.
  2756. my $name = shift;
  2757. my $schema = shift || '';
  2758.  
  2759. my $stat = 0;
  2760. ## Is this excluded?
  2761. if (defined $opt{exclude}) {
  2762. $stat = 1;
  2763. for (@{$opt{exclude}}) {
  2764. for my $ex (split /\s*,\s*/o => $_) {
  2765. if ($ex =~ s/\.$//) {
  2766. if ($ex =~ s/^~//) {
  2767. ($stat += 2 and last) if $schema =~ /$ex/;
  2768. }
  2769. else {
  2770. ($stat += 2 and last) if $schema eq $ex;
  2771. }
  2772. }
  2773. elsif ($ex =~ s/^~//) {
  2774. ($stat += 2 and last) if $name =~ /$ex/;
  2775. }
  2776. else {
  2777. ($stat += 2 and last) if $name eq $ex;
  2778. }
  2779. }
  2780. }
  2781. }
  2782. if (defined $opt{include}) {
  2783. $stat += 4;
  2784. for (@{$opt{include}}) {
  2785. for my $in (split /\s*,\s*/o => $_) {
  2786. if ($in =~ s/\.$//) {
  2787. if ($in =~ s/^~//) {
  2788. ($stat += 8 and last) if $schema =~ /$in/;
  2789. }
  2790. else {
  2791. ($stat += 8 and last) if $schema eq $in;
  2792. }
  2793. }
  2794. elsif ($in =~ s/^~//) {
  2795. ($stat += 8 and last) if $name =~ /$in/;
  2796. }
  2797. else {
  2798. ($stat += 8 and last) if $name eq $in;
  2799. }
  2800. }
  2801. }
  2802. }
  2803.  
  2804. ## Easiest to state the cases when we DO skip:
  2805. return 1 if
  2806. 3 == $stat ## exclude matched, no inclusion checking
  2807. or 4 == $stat ## include check only, no match
  2808. or 7 == $stat; ## exclude match, no inclusion match
  2809.  
  2810. return 0;
  2811.  
  2812. } ## end of skip_item
  2813.  
  2814.  
  2815. sub validate_range {
  2816.  
  2817. ## Valid that warning and critical are set correctly.
  2818. ## Returns new values of both
  2819.  
  2820. my $arg = shift;
  2821. defined $arg and ref $arg eq 'HASH' or ndie qq{validate_range must be called with a hashref\n};
  2822.  
  2823. return ('','') if $MRTG and !$arg->{forcemrtg};
  2824.  
  2825. my $type = $arg->{type} or ndie qq{validate_range must be provided a 'type'\n};
  2826.  
  2827. ## The 'default default' is an empty string, which should fail all mandatory tests
  2828. ## We only set the 'arg' default if neither option is provided.
  2829. my $warning = exists $opt{warning} ? $opt{warning} :
  2830. exists $opt{critical} ? '' : $arg->{default_warning} || '';
  2831. my $critical = exists $opt{critical} ? $opt{critical} :
  2832. exists $opt{warning} ? '' : $arg->{default_critical} || '';
  2833.  
  2834. if ('string' eq $type) {
  2835. ## Don't use this unless you have to
  2836. }
  2837. elsif ('seconds' eq $type) {
  2838. if (length $warning) {
  2839. if ($warning !~ $timesecre) {
  2840. ndie msg('range-seconds', 'warning');
  2841. }
  2842. $warning = $1;
  2843. }
  2844. if (length $critical) {
  2845. if ($critical !~ $timesecre) {
  2846. ndie msg('range-seconds', 'critical')
  2847. }
  2848. $critical = $1;
  2849. if (!$arg->{any_warning} and length $warning and $warning > $critical) {
  2850. ndie msg('range-warnbigtime', $warning, $critical);
  2851. }
  2852. }
  2853. }
  2854. elsif ('time' eq $type) {
  2855. $critical = size_in_seconds($critical, 'critical');
  2856. $warning = size_in_seconds($warning, 'warning');
  2857. if (! length $critical and ! length $warning) {
  2858. ndie msg('range-notime');
  2859. }
  2860. if (!$arg->{any_warning} and length $warning and length $critical and $warning > $critical) {
  2861. ndie msg('range-warnbigtime', $warning, $critical);
  2862. }
  2863. }
  2864. elsif ('version' eq $type) {
  2865. my $msg = msg('range-version');
  2866. if (length $warning and $warning !~ /^\d+\.\d+\.?[\d\w]*$/) {
  2867. ndie msg('range-badversion', 'warning', $msg);
  2868. }
  2869. if (length $critical and $critical !~ /^\d+\.\d+\.?[\d\w]*$/) {
  2870. ndie msg('range-badversion', 'critical', $msg);
  2871. }
  2872. if (! length $critical and ! length $warning) {
  2873. ndie msg('range-noopt-orboth');
  2874. }
  2875. }
  2876. elsif ('size' eq $type) {
  2877. if (length $critical) {
  2878. if ($critical !~ $sizere) {
  2879. ndie msg('range-badsize', 'critical');
  2880. }
  2881. $critical = size_in_bytes($1,$2);
  2882. }
  2883. if (length $warning) {
  2884. if ($warning !~ $sizere) {
  2885. ndie msg('range-badsize', 'warning');
  2886. }
  2887. $warning = size_in_bytes($1,$2);
  2888. if (!$arg->{any_warning} and length $critical and $warning > $critical) {
  2889. ndie msg('range-warnbigsize', $warning, $critical);
  2890. }
  2891. }
  2892. elsif (!length $critical) {
  2893. ndie msg('range-nosize');
  2894. }
  2895. }
  2896. elsif ($type =~ /integer/) {
  2897. $warning =~ s/_//g;
  2898. if (length $warning and $warning !~ /^[-+]?\d+$/) {
  2899. ndie $type =~ /positive/ ? msg('range-int-pos', 'warning') : msg('range-int', 'warning');
  2900. }
  2901. elsif (length $warning and $type =~ /positive/ and $warning <= 0) {
  2902. ndie msg('range-int-pos', 'warning');
  2903. }
  2904.  
  2905. $critical =~ s/_//g;
  2906. if (length $critical and $critical !~ /^[-+]?\d+$/) {
  2907. ndie $type =~ /positive/ ? msg('range-int-pos', 'critical') : msg('range-int', 'critical');
  2908. }
  2909. elsif (length $critical and $type =~ /positive/ and $critical <= 0) {
  2910. ndie msg('range-int-pos', 'critical');
  2911. }
  2912.  
  2913. if (length $warning
  2914. and length $critical
  2915. and (
  2916. ($opt{reverse} and $warning < $critical)
  2917. or
  2918. (!$opt{reverse} and $warning > $critical)
  2919. )
  2920. ) {
  2921. ndie msg('range-warnbig');
  2922. }
  2923. if ($type !~ /string/) {
  2924. $warning = int $warning if length $warning;
  2925. $critical = int $critical if length $critical;
  2926. }
  2927. }
  2928. elsif ('restringex' eq $type) {
  2929. if (! length $critical and ! length $warning) {
  2930. ndie msg('range-noopt-one');
  2931. }
  2932. if (length $critical and length $warning) {
  2933. ndie msg('range-noopt-only');
  2934. }
  2935. my $string = length $critical ? $critical : $warning;
  2936. my $regex = ($string =~ s/^~//) ? '~' : '=';
  2937. $string =~ /^\w+$/ or ndie msg('invalid-option');
  2938. }
  2939. elsif ('percent' eq $type) {
  2940. if (length $critical) {
  2941. if ($critical !~ /^(\d+)\%$/) {
  2942. ndie msg('range-badpercent', 'critical');
  2943. }
  2944. $critical = $1;
  2945. }
  2946. if (length $warning) {
  2947. if ($warning !~ /^(\d+)\%$/) {
  2948. ndie msg('range-badpercent', 'warning');
  2949. }
  2950. $warning = $1;
  2951. }
  2952. }
  2953. elsif ('size or percent' eq $type) {
  2954. if (length $critical) {
  2955. if ($critical =~ $sizere) {
  2956. $critical = size_in_bytes($1,$2);
  2957. }
  2958. elsif ($critical !~ /^\d+\%$/) {
  2959. ndie msg('range-badpercsize', 'critical');
  2960. }
  2961. }
  2962. if (length $warning) {
  2963. if ($warning =~ $sizere) {
  2964. $warning = size_in_bytes($1,$2);
  2965. }
  2966. elsif ($warning !~ /^\d+\%$/) {
  2967. ndie msg('range-badpercsize', 'warning');
  2968. }
  2969. }
  2970. elsif (! length $critical) {
  2971. ndie msg('range-noopt-size');
  2972. }
  2973. }
  2974. elsif ('checksum' eq $type) {
  2975. if (length $critical and $critical !~ $checksumre and $critical ne '0') {
  2976. ndie msg('range-badcs', 'critical');
  2977. }
  2978. if (length $warning and $warning !~ $checksumre) {
  2979. ndie msg('range-badcs', 'warning');
  2980. }
  2981. }
  2982. elsif ('multival' eq $type) { ## Simple number, or foo=#;bar=#
  2983. ## Note: only used for check_locks
  2984. my %err;
  2985. while ($critical =~ /(\w+)\s*=\s*(\d+)/gi) {
  2986. my ($name,$val) = (lc $1,$2);
  2987. $name =~ s/lock$//;
  2988. $err{$name} = $val;
  2989. }
  2990. if (keys %err) {
  2991. $critical = \%err;
  2992. }
  2993. elsif (length $critical and $critical =~ /^(\d+)$/) {
  2994. $err{total} = $1;
  2995. $critical = \%err;
  2996. }
  2997. elsif (length $critical) {
  2998. ndie msg('range-badlock', 'critical');
  2999. }
  3000. my %warn;
  3001. while ($warning =~ /(\w+)\s*=\s*(\d+)/gi) {
  3002. my ($name,$val) = (lc $1,$2);
  3003. $name =~ s/lock$//;
  3004. $warn{$name} = $val;
  3005. }
  3006. if (keys %warn) {
  3007. $warning = \%warn;
  3008. }
  3009. elsif (length $warning and $warning =~ /^(\d+)$/) {
  3010. $warn{total} = $1;
  3011. $warning = \%warn;
  3012. }
  3013. elsif (length $warning) {
  3014. ndie msg('range-badlock', 'warning');
  3015. }
  3016. }
  3017. elsif ('cacti' eq $type) { ## Takes no args, just dumps data
  3018. if (length $warning or length $critical) {
  3019. ndie msg('range-cactionly');
  3020. }
  3021. }
  3022. else {
  3023. ndie msg('range-badtype', $type);
  3024. }
  3025.  
  3026. if ($arg->{both}) {
  3027. if (! length $warning or ! length $critical) {
  3028. ndie msg('range-noopt-both');
  3029. }
  3030. }
  3031. if ($arg->{leastone}) {
  3032. if (! length $warning and ! length $critical) {
  3033. ndie msg('range-noopt-one');
  3034. }
  3035. }
  3036. elsif ($arg->{onlyone}) {
  3037. if (length $warning and length $critical) {
  3038. ndie msg('range-noopt-only');
  3039. }
  3040. if (! length $warning and ! length $critical) {
  3041. ndie msg('range-noopt-one');
  3042. }
  3043. }
  3044.  
  3045. return ($warning,$critical);
  3046.  
  3047. } ## end of validate_range
  3048.  
  3049.  
  3050. sub validate_size_or_percent_with_oper {
  3051.  
  3052. my $arg = shift || {};
  3053. ndie qq{validate_range must be called with a hashref\n}
  3054. unless ref $arg eq 'HASH';
  3055.  
  3056. my $warning = exists $opt{warning} ? $opt{warning} :
  3057. exists $opt{critical} ? '' : $arg->{default_warning} || '';
  3058. my $critical = exists $opt{critical} ? $opt{critical} :
  3059. exists $opt{warning} ? '' : $arg->{default_critical} || '';
  3060.  
  3061. ndie msg('range-noopt-size') unless length $critical || length $warning;
  3062. my @subs;
  3063. for my $val ($warning, $critical) {
  3064. if ($val =~ /^(.+?)\s([&|]{2}|and|or)\s(.+)$/i) {
  3065. my ($l, $op, $r) = ($1, $2, $3);
  3066. local $opt{warning} = $l;
  3067. local $opt{critical} = 0;
  3068. ($l) = validate_range({ type => 'size or percent' });
  3069. $opt{warning} = $r;
  3070. ($r) = validate_range({ type => 'size or percent' });
  3071. if ($l =~ s/%$//) {
  3072. ($l, $r) = ($r, $l);
  3073. }
  3074. else {
  3075. $r =~ s/%$//;
  3076. }
  3077. push @subs, $op eq '&&' || lc $op eq 'and' ? sub {
  3078. $_[0] >= $l && $_[1] >= $r;
  3079. } : sub {
  3080. $_[0] >= $l || $_[1] >= $r;
  3081. };
  3082. }
  3083. else {
  3084. local $opt{warning} = $val;
  3085. local $opt{critical} = 0;
  3086. my ($v) = validate_range({ type => 'size or percent' });
  3087. push @subs, !length $v ? sub { 0 }
  3088. : $v =~ s/%$// ? sub { $_[1] >= $v }
  3089. : sub { $_[0] >= $v };
  3090. }
  3091. }
  3092.  
  3093. return @subs;
  3094.  
  3095. } ## end of validate_size_or_percent_with_oper
  3096.  
  3097.  
  3098. sub validate_integer_for_time {
  3099.  
  3100. my $arg = shift || {};
  3101. ndie qq{validate_integer_for_time must be called with a hashref\n}
  3102. unless ref $arg eq 'HASH';
  3103.  
  3104. my $warning = exists $opt{warning} ? $opt{warning} :
  3105. exists $opt{critical} ? '' : $arg->{default_warning} || '';
  3106. my $critical = exists $opt{critical} ? $opt{critical} :
  3107. exists $opt{warning} ? '' : $arg->{default_critical} || '';
  3108. ndie msg('range-nointfortime', 'critical') unless length $critical or length $warning;
  3109.  
  3110. my @ret;
  3111. for my $spec ([ warning => $warning], [critical => $critical]) {
  3112. my ($level, $val) = @{ $spec };
  3113. if (length $val) {
  3114. if ($val =~ /^(.+?)\sfor\s(.+)$/i) {
  3115. my ($int, $time) = ($1, $2);
  3116.  
  3117. # Integer first, time second.
  3118. ($int, $time) = ($time, $int)
  3119. if $int =~ /[a-zA-Z]$/ || $time =~ /^[-+]\d+$/;
  3120.  
  3121. # Determine the values.
  3122. $time = size_in_seconds($time, $level);
  3123. ndie msg('range-int', $level) if $time !~ /^[-+]?\d+$/;
  3124. push @ret, int $int, $time;
  3125. }
  3126. else {
  3127. # Disambiguate int from time int by sign.
  3128. if ($val =~ /^[-+]\d+$/) {
  3129. ndie msg('range-int', $level) if $val !~ /^[-+]?\d+$/;
  3130. push @ret, int $val, '';
  3131. }
  3132. else {
  3133. # Assume time for backwards compatibility.
  3134. push @ret, '', size_in_seconds($val, $level);
  3135. }
  3136. }
  3137. }
  3138. else {
  3139. push @ret, '', '';
  3140. }
  3141. }
  3142.  
  3143. return @ret;
  3144.  
  3145. } ## end of validate_integer_for_time
  3146.  
  3147.  
  3148. sub perfname {
  3149.  
  3150. ## Return a safe label name for Nagios performance data
  3151. my $name = shift;
  3152.  
  3153. my $escape = 0;
  3154.  
  3155. $name =~ s/'/''/g and $escape++;
  3156.  
  3157. if ($escape or index($name, ' ') >=0) {
  3158. $name = qq{'$name'};
  3159. }
  3160.  
  3161. return $name;
  3162.  
  3163. } ## end of perfname;
  3164.  
  3165.  
  3166. sub open_controldata {
  3167. ## Requires $ENV{PGDATA} or --datadir
  3168.  
  3169. ## Find the data directory, make sure it exists
  3170. my $dir = $opt{datadir} || $ENV{PGDATA};
  3171.  
  3172. if (!defined $dir or ! length $dir) {
  3173. ndie msg('checkpoint-nodir');
  3174. }
  3175.  
  3176. if (! -d $dir) {
  3177. ndie msg('checkpoint-baddir', $dir);
  3178. }
  3179.  
  3180. ## Run pg_controldata
  3181. ## We still catch deprecated option
  3182. my $pgc;
  3183. if (defined $ENV{PGCONTROLDATA} and length $ENV{PGCONTROLDATA}) {
  3184. # ndie msg('depr-pgcontroldata');
  3185. $pgc = "$ENV{PGCONTROLDATA}";
  3186. }
  3187. else {
  3188. $pgc = (defined $PGBINDIR) ? "$PGBINDIR/pg_controldata" : 'pg_controldata';
  3189. chomp($pgc = qx{which "$pgc"});
  3190. }
  3191. -x $pgc or ndie msg('opt-psql-noexec', $pgc);
  3192.  
  3193. $COM = qq{$pgc "$dir"};
  3194. eval {
  3195. $res = qx{$COM 2>&1};
  3196. };
  3197. if ($@) {
  3198. ndie msg('checkpoint-nosys', $@);
  3199. }
  3200.  
  3201. ## If the path is echoed back, we most likely have an invalid data dir
  3202. if ($res =~ /$dir/) {
  3203. ndie msg('checkpoint-baddir2', $dir);
  3204. }
  3205.  
  3206. if ($res =~ /WARNING: Calculated CRC checksum/) {
  3207. ndie msg('checkpoint-badver', $dir);
  3208. }
  3209. if ($res !~ /^pg_control.+\d+/) {
  3210. ndie msg('checkpoint-badver2');
  3211. }
  3212.  
  3213. ## return the pg_controldata output
  3214. return $res;
  3215. }
  3216.  
  3217.  
  3218. sub check_archive_ready {
  3219.  
  3220. ## Check on the number of WAL archive with status "ready"
  3221. ## Supports: Nagios, MRTG
  3222. ## Must run as a superuser
  3223. ## Critical and warning are the number of files
  3224. ## Example: --critical=10
  3225.  
  3226. return check_wal_files('/archive_status', '.ready', 10, 15);
  3227.  
  3228. } ## end of check_archive_ready
  3229.  
  3230.  
  3231. sub check_autovac_freeze {
  3232.  
  3233. ## Check how close all databases are to autovacuum_freeze_max_age
  3234. ## Supports: Nagios, MRTG
  3235. ## It makes no sense to run this more than once on the same cluster
  3236. ## Warning and criticals are percentages
  3237. ## Can also ignore databases with exclude, and limit with include
  3238.  
  3239. my ($warning, $critical) = validate_range
  3240. ({
  3241. type => 'percent',
  3242. default_warning => '90%',
  3243. default_critical => '95%',
  3244. forcemrtg => 1,
  3245. });
  3246.  
  3247. (my $w = $warning) =~ s/\D//;
  3248. (my $c = $critical) =~ s/\D//;
  3249.  
  3250. my $SQL = q{SELECT freez, txns, ROUND(100*(txns/freez::float)) AS perc, datname}.
  3251. q{ FROM (SELECT foo.freez::int, age(datfrozenxid) AS txns, datname}.
  3252. q{ FROM pg_database d JOIN (SELECT setting AS freez FROM pg_settings WHERE name = 'autovacuum_freeze_max_age') AS foo}.
  3253. q{ ON (true) WHERE d.datallowconn) AS foo2 ORDER BY 3 DESC, 4 ASC};
  3254.  
  3255. my $info = run_command($SQL, {regex => qr{\w+} } );
  3256.  
  3257. $db = $info->{db}[0];
  3258.  
  3259. my (@crit,@warn,@ok);
  3260. my ($maxp,$maxt,$maxdb) = (0,0,''); ## used by MRTG only
  3261. SLURP: for my $r (@{$db->{slurp}}) {
  3262. next SLURP if skip_item($r->{datname});
  3263.  
  3264. if ($MRTG) {
  3265. if ($r->{perc} > $maxp) {
  3266. $maxdb = $r->{datname};
  3267. $maxp = $r->{perc};
  3268. }
  3269. elsif ($r->{perc} == $maxp) {
  3270. $maxdb .= sprintf '%s%s', (length $maxdb ? ' | ' : ''), $r->{datname};
  3271. }
  3272. $maxt = $r->{txns} if $r->{txns} > $maxt;
  3273. next SLURP;
  3274. }
  3275.  
  3276. my $msg = sprintf ' %s=%s%%;%s;%s', perfname($r->{datname}), $r->{perc}, $w, $c;
  3277. $db->{perf} .= " $msg";
  3278. if (length $critical and $r->{perc} >= $c) {
  3279. push @crit => $msg;
  3280. }
  3281. elsif (length $warning and $r->{perc} >= $w) {
  3282. push @warn => $msg;
  3283. }
  3284. else {
  3285. push @ok => $msg;
  3286. }
  3287. }
  3288. if ($MRTG) {
  3289. do_mrtg({one => $maxp, two => $maxt, msg => $maxdb});
  3290. }
  3291. if (@crit) {
  3292. add_critical join ' ' => @crit;
  3293. }
  3294. elsif (@warn) {
  3295. add_warning join ' ' => @warn;
  3296. }
  3297. else {
  3298. add_ok join ' ' => @ok;
  3299. }
  3300.  
  3301. return;
  3302.  
  3303. } ## end of check_autovac_freeze
  3304.  
  3305.  
  3306. sub check_backends {
  3307.  
  3308. ## Check the number of connections
  3309. ## Supports: Nagios, MRTG
  3310. ## It makes no sense to run this more than once on the same cluster
  3311. ## Need to be superuser, else only your queries will be visible
  3312. ## Warning and criticals can take three forms:
  3313. ## critical = 12 -- complain if there are 12 or more connections
  3314. ## critical = 95% -- complain if >= 95% of available connections are used
  3315. ## critical = -5 -- complain if there are only 5 or fewer connection slots left
  3316. ## The former two options only work with simple numbers - no percentage or negative
  3317. ## Can also ignore databases with exclude, and limit with include
  3318.  
  3319. my $warning = $opt{warning} || '90%';
  3320. my $critical = $opt{critical} || '95%';
  3321. my $noidle = $opt{noidle} || 0;
  3322.  
  3323. ## If only critical was used, remove the default warning
  3324. if ($opt{critical} and !$opt{warning}) {
  3325. $warning = $critical;
  3326. }
  3327.  
  3328. my $validre = qr{^(\-?)(\d+)(\%?)$};
  3329. if ($critical !~ $validre) {
  3330. ndie msg('backends-users', 'Critical');
  3331. }
  3332. my ($e1,$e2,$e3) = ($1,$2,$3);
  3333. if ($warning !~ $validre) {
  3334. ndie msg('backends-users', 'Warning');
  3335. }
  3336. my ($w1,$w2,$w3) = ($1,$2,$3);
  3337.  
  3338. ## If number is greater, all else is same, and not minus
  3339. if ($w2 > $e2 and $w1 eq $e1 and $w3 eq $e3 and $w1 eq '') {
  3340. ndie msg('range-warnbig');
  3341. }
  3342. ## If number is less, all else is same, and minus
  3343. if ($w2 < $e2 and $w1 eq $e1 and $w3 eq $e3 and $w1 eq '-') {
  3344. ndie msg('range-warnsmall');
  3345. }
  3346. if (($w1 and $w3) or ($e1 and $e3)) {
  3347. ndie msg('range-neg-percent');
  3348. }
  3349.  
  3350. my $MAXSQL = q{SELECT setting AS mc FROM pg_settings WHERE name = 'max_connections'};
  3351.  
  3352. my $NOIDLE = $noidle ? q{WHERE current_query <> '<IDLE>'} : '';
  3353. $SQL = qq{
  3354. SELECT COUNT(datid) AS current,
  3355. ($MAXSQL) AS mc,
  3356. d.datname
  3357. FROM pg_database d
  3358. LEFT JOIN pg_stat_activity s ON (s.datid = d.oid) $NOIDLE
  3359. GROUP BY 2,3
  3360. ORDER BY datname
  3361. };
  3362. my $info = run_command($SQL, {regex => qr{\d+}, fatalregex => 'too many clients' } );
  3363.  
  3364. $db = $info->{db}[0];
  3365.  
  3366. ## If we cannot connect because of too many clients, we treat as a critical error
  3367. if (exists $info->{fatalregex}) {
  3368. my $regmsg = msg('backends-po');
  3369. my $regmsg2 = msg_en('backends-po');
  3370. if ($info->{fatalregex} =~ /$regmsg/ or $info->{fatalregex} =~ /$regmsg2/) {
  3371. add_critical msg('backends-fatal');
  3372. return;
  3373. }
  3374. }
  3375.  
  3376. ## There may be no entries returned if we catch pg_stat_activity at the right
  3377. ## moment in older versions of Postgres
  3378. if (! defined $db) {
  3379. $info = run_command($MAXSQL, {regex => qr[\d] } );
  3380. $db = $info->{db}[0];
  3381. if (!defined $db->{slurp} or $db->{slurp} !~ /(\d+)/) {
  3382. undef %unknown;
  3383. add_unknown msg('backends-nomax');
  3384. return;
  3385. }
  3386. my $limit = $1;
  3387. if ($MRTG) {
  3388. do_mrtg({one => 1, msg => msg('backends-mrtg', $db->{dbname}, $limit)});
  3389. }
  3390. my $percent = (int 1/$limit*100) || 1;
  3391. add_ok msg('backends-msg', 1, $limit, $percent);
  3392. return;
  3393. }
  3394.  
  3395. my $total = 0;
  3396. my $grandtotal = @{$db->{slurp}};
  3397.  
  3398. ## If no max_connections, something is wrong
  3399. if ($db->{slurp}[0]{mc} !~ /\d/) {
  3400. add_unknown msg('backends-nomax');
  3401. return;
  3402. }
  3403. my $limit = $db->{slurp}[0]{mc};
  3404.  
  3405. for my $r (@{$db->{slurp}}) {
  3406.  
  3407. ## Always want perf to show all
  3408. my $nwarn=$w2;
  3409. my $ncrit=$e2;
  3410. if ($e1) {
  3411. $ncrit = $limit-$e2;
  3412. }
  3413. elsif ($e3) {
  3414. $ncrit = (int $e2*$limit/100);
  3415. }
  3416. if ($w1) {
  3417. $nwarn = $limit-$w2;
  3418. }
  3419. elsif ($w3) {
  3420. $nwarn = (int $w2*$limit/100)
  3421. }
  3422.  
  3423. if (! skip_item($r->{datname})) {
  3424. $db->{perf} .= sprintf ' %s=%s;%s;%s;0;%s',
  3425. perfname($r->{datname}), $r->{current}, $nwarn, $ncrit, $limit;
  3426. $total += $r->{current};
  3427. }
  3428. }
  3429.  
  3430. if ($MRTG) {
  3431. do_mrtg({one => $total, msg => msg('backends-mrtg', $db->{dbname}, $limit)});
  3432. }
  3433.  
  3434. if (!$total) {
  3435. if ($grandtotal) {
  3436. ## We assume that exclude/include rules are correct, and we simply had no entries
  3437. ## at all in the specific databases we wanted
  3438. add_ok msg('backends-oknone');
  3439. }
  3440. else {
  3441. add_unknown msg('no-match-db');
  3442. }
  3443. return;
  3444. }
  3445.  
  3446. my $percent = (int $total / $limit*100) || 1;
  3447. my $msg = msg('backends-msg', $total, $limit, $percent);
  3448. my $ok = 1;
  3449.  
  3450. if ($e1) { ## minus
  3451. $ok = 0 if $limit-$total <= $e2;
  3452. }
  3453. elsif ($e3) { ## percent
  3454. my $nowpercent = $total/$limit*100;
  3455. $ok = 0 if $nowpercent >= $e2;
  3456. }
  3457. else { ## raw number
  3458. $ok = 0 if $total >= $e2;
  3459. }
  3460. if (!$ok) {
  3461. add_critical $msg;
  3462. return;
  3463. }
  3464.  
  3465. if ($w1) {
  3466. $ok = 0 if $limit-$total <= $w2;
  3467. }
  3468. elsif ($w3) {
  3469. my $nowpercent = $total/$limit*100;
  3470. $ok = 0 if $nowpercent >= $w2;
  3471. }
  3472. else {
  3473. $ok = 0 if $total >= $w2;
  3474. }
  3475. if (!$ok) {
  3476. add_warning $msg;
  3477. return;
  3478. }
  3479.  
  3480. add_ok $msg;
  3481.  
  3482. return;
  3483.  
  3484. } ## end of check_backends
  3485.  
  3486.  
  3487. sub check_bloat {
  3488.  
  3489. ## Check how bloated the tables and indexes are
  3490. ## Supports: Nagios, MRTG
  3491. ## NOTE! This check depends on ANALYZE being run regularly
  3492. ## Also requires stats collection to be on
  3493. ## This action may be very slow on large databases
  3494. ## By default, checks all relations
  3495. ## Can check specific one(s) with include; can ignore some with exclude
  3496. ## Begin name with a '~' to make it a regular expression
  3497. ## Warning and critical are in sizes, defaults to bytes
  3498. ## Valid units: b, k, m, g, t, e
  3499. ## All above may be written as plural or with a trailing 'b'
  3500. ## Example: --critical="25 GB" --include="mylargetable"
  3501. ## Can also specify percentages
  3502.  
  3503. ## Don't bother with tables or indexes unless they have at least this many bloated pages
  3504. my $MINPAGES = 0;
  3505. my $MINIPAGES = 10;
  3506.  
  3507. my $LIMIT = 10;
  3508. if ($opt{perflimit}) {
  3509. $LIMIT = $opt{perflimit};
  3510. }
  3511.  
  3512. my ($warning, $critical) = validate_size_or_percent_with_oper
  3513. ({
  3514. default_warning => '1 GB',
  3515. default_critical => '5 GB',
  3516. });
  3517.  
  3518. ## This was fun to write
  3519. $SQL = q{
  3520. SELECT
  3521. current_database() AS db, schemaname, tablename, reltuples::bigint AS tups, relpages::bigint AS pages, otta,
  3522. ROUND(CASE WHEN otta=0 OR sml.relpages=0 OR sml.relpages=otta THEN 0.0 ELSE sml.relpages/otta::numeric END,1) AS tbloat,
  3523. CASE WHEN relpages < otta THEN 0 ELSE relpages::bigint - otta END AS wastedpages,
  3524. CASE WHEN relpages < otta THEN 0 ELSE bs*(sml.relpages-otta)::bigint END AS wastedbytes,
  3525. CASE WHEN relpages < otta THEN '0 bytes'::text ELSE (bs*(relpages-otta))::bigint || ' bytes' END AS wastedsize,
  3526. iname, ituples::bigint AS itups, ipages::bigint AS ipages, iotta,
  3527. ROUND(CASE WHEN iotta=0 OR ipages=0 OR ipages=iotta THEN 0.0 ELSE ipages/iotta::numeric END,1) AS ibloat,
  3528. CASE WHEN ipages < iotta THEN 0 ELSE ipages::bigint - iotta END AS wastedipages,
  3529. CASE WHEN ipages < iotta THEN 0 ELSE bs*(ipages-iotta) END AS wastedibytes,
  3530. CASE WHEN ipages < iotta THEN '0 bytes' ELSE (bs*(ipages-iotta))::bigint || ' bytes' END AS wastedisize,
  3531. CASE WHEN relpages < otta THEN
  3532. CASE WHEN ipages < iotta THEN 0 ELSE ipages-iotta::bigint END
  3533. ELSE CASE WHEN ipages < iotta THEN relpages-otta::bigint
  3534. ELSE relpages-otta::bigint + ipages-iotta::bigint END
  3535. END AS totalwastedbytes
  3536. FROM (
  3537. SELECT
  3538. nn.nspname AS schemaname,
  3539. cc.relname AS tablename,
  3540. COALESCE(cc.reltuples,0) AS reltuples,
  3541. COALESCE(cc.relpages,0) AS relpages,
  3542. COALESCE(bs,0) AS bs,
  3543. COALESCE(CEIL((cc.reltuples*((datahdr+ma-
  3544. (CASE WHEN datahdr%ma=0 THEN ma ELSE datahdr%ma END))+nullhdr2+4))/(bs-20::float)),0) AS otta,
  3545. COALESCE(c2.relname,'?') AS iname, COALESCE(c2.reltuples,0) AS ituples, COALESCE(c2.relpages,0) AS ipages,
  3546. COALESCE(CEIL((c2.reltuples*(datahdr-12))/(bs-20::float)),0) AS iotta -- very rough approximation, assumes all cols
  3547. FROM
  3548. pg_class cc
  3549. JOIN pg_namespace nn ON cc.relnamespace = nn.oid AND nn.nspname <> 'information_schema'
  3550. LEFT JOIN
  3551. (
  3552. SELECT
  3553. ma,bs,schemaname,tablename,
  3554. (datawidth+(hdr+ma-(case when hdr%ma=0 THEN ma ELSE hdr%ma END)))::numeric AS datahdr,
  3555. (maxfracsum*(nullhdr+ma-(case when nullhdr%ma=0 THEN ma ELSE nullhdr%ma END))) AS nullhdr2
  3556. FROM (
  3557. SELECT
  3558. schemaname, tablename, hdr, ma, bs,
  3559. SUM((1-null_frac)*avg_width) AS datawidth,
  3560. MAX(null_frac) AS maxfracsum,
  3561. hdr+(
  3562. SELECT 1+count(*)/8
  3563. FROM pg_stats s2
  3564. WHERE null_frac<>0 AND s2.schemaname = s.schemaname AND s2.tablename = s.tablename
  3565. ) AS nullhdr
  3566. FROM pg_stats s, (
  3567. SELECT
  3568. BLOCK_SIZE,
  3569. CASE WHEN SUBSTRING(SPLIT_PART(v, ' ', 2) FROM '#"[0-9]+.[0-9]+#"%' for '#')
  3570. IN ('8.0','8.1','8.2') THEN 27 ELSE 23 END AS hdr,
  3571. CASE WHEN v ~ 'mingw32' OR v ~ '64-bit' THEN 8 ELSE 4 END AS ma
  3572. FROM (SELECT version() AS v) AS foo
  3573. ) AS constants
  3574. GROUP BY 1,2,3,4,5
  3575. ) AS foo
  3576. ) AS rs
  3577. ON cc.relname = rs.tablename AND nn.nspname = rs.schemaname
  3578. LEFT JOIN pg_index i ON indrelid = cc.oid
  3579. LEFT JOIN pg_class c2 ON c2.oid = i.indexrelid
  3580. ) AS sml
  3581. };
  3582.  
  3583. if (! defined $opt{include} and ! defined $opt{exclude}) {
  3584. $SQL .= " WHERE sml.relpages - otta > $MINPAGES OR ipages - iotta > $MINIPAGES";
  3585. $SQL .= " ORDER BY totalwastedbytes DESC LIMIT $LIMIT";
  3586. }
  3587. else {
  3588. $SQL .= ' ORDER BY totalwastedbytes DESC';
  3589. }
  3590.  
  3591. if ($psql_version <= 7.4) {
  3592. $SQL =~ s/BLOCK_SIZE/(SELECT 8192) AS bs/;
  3593. }
  3594. else {
  3595. $SQL =~ s/BLOCK_SIZE/(SELECT current_setting('block_size')::numeric) AS bs/;
  3596. }
  3597.  
  3598. my $info = run_command($SQL);
  3599.  
  3600. if (defined $info->{db}[0] and exists $info->{db}[0]{error}) {
  3601. ndie $info->{db}[0]{error};
  3602. }
  3603.  
  3604. my %seenit;
  3605.  
  3606. ## Store the perf data for sorting at the end
  3607. my %perf;
  3608.  
  3609. $db = $info->{db}[0];
  3610.  
  3611. if ($db->{slurp} !~ /\w+/o) {
  3612. add_ok msg('bloat-nomin') unless $MRTG;
  3613. return;
  3614. }
  3615. ## Not a 'regex' to run_command as we need to check the above first.
  3616. if ($db->{slurp} !~ /\d+/) {
  3617. add_unknown msg('invalid-query', $db->{slurp}) unless $MRTG;
  3618. return;
  3619. }
  3620.  
  3621. my $max = -1;
  3622. my $maxmsg = '?';
  3623.  
  3624. ## The perf must be added before the add_x, so we defer the settings:
  3625. my (@addwarn, @addcrit);
  3626.  
  3627. for my $r (@{ $db->{slurp} }) {
  3628.  
  3629. for my $v (values %$r) {
  3630. $v =~ s/(\d+) bytes/pretty_size($1,1)/ge;
  3631. }
  3632.  
  3633. my ($dbname,$schema,$table,$tups,$pages,$otta,$bloat,$wp,$wb,$ws) = @$r{
  3634. qw/ db schemaname tablename tups pages otta tbloat wastedpages wastedbytes wastedsize/};
  3635.  
  3636. next if skip_item($table, $schema);
  3637.  
  3638. my ($index,$irows,$ipages,$iotta,$ibloat,$iwp,$iwb,$iws) = @$r{
  3639. qw/ iname irows ipages iotta ibloat wastedipgaes wastedibytes wastedisize/};
  3640.  
  3641. ## Made it past the exclusions
  3642. $max = -2 if $max == -1;
  3643.  
  3644. ## Do the table first if we haven't seen it
  3645. if (! $seenit{"$dbname.$schema.$table"}++) {
  3646. my $nicename = perfname("$schema.$table");
  3647. $perf{$wb}{$nicename}++;
  3648. my $msg = msg('bloat-table', $dbname, $schema, $table, $tups, $pages, $otta, $bloat, $wb, $ws);
  3649. my $ok = 1;
  3650. my $perbloat = $bloat * 100;
  3651.  
  3652. if ($MRTG) {
  3653. $stats{table}{"DB=$dbname TABLE=$schema.$table"} = [$wb, $bloat];
  3654. next;
  3655. }
  3656. if ($critical->($wb, $perbloat)) {
  3657. push @addcrit => $msg;
  3658. $ok = 0;
  3659. }
  3660.  
  3661. if ($ok and $warning->($wb, $perbloat)) {
  3662. push @addwarn => $msg;
  3663. $ok = 0;
  3664. }
  3665. ($max = $wb, $maxmsg = $msg) if $wb > $max and $ok;
  3666. }
  3667.  
  3668. ## Now the index, if it exists
  3669. if ($index ne '?') {
  3670. my $nicename = perfname($index);
  3671. $perf{$iwb}{$nicename}++;
  3672. my $msg = msg('bloat-index', $dbname, $index, $irows, $ipages, $iotta, $ibloat, $iwb, $iws);
  3673. my $ok = 1;
  3674. my $iperbloat = $ibloat * 100;
  3675.  
  3676. if ($MRTG) {
  3677. $stats{index}{"DB=$dbname INDEX=$index"} = [$iwb, $ibloat];
  3678. next;
  3679. }
  3680. if ($critical->($iwb, $iperbloat)) {
  3681. push @addcrit => $msg;
  3682. $ok = 0;
  3683. }
  3684.  
  3685. if ($ok and $warning->($iwb, $iperbloat)) {
  3686. push @addwarn => $msg;
  3687. $ok = 0;
  3688. }
  3689. ($max = $iwb, $maxmsg = $msg) if $iwb > $max and $ok;
  3690. }
  3691. }
  3692.  
  3693. ## Set a sorted limited perf
  3694. $db->{perf} = '';
  3695. my $count = 0;
  3696. PERF: for my $size (sort {$b <=> $a } keys %perf) {
  3697. for my $name (sort keys %{ $perf{$size} }) {
  3698. $db->{perf} .= "$name=${size}B ";
  3699. last PERF if $opt{perflimit} and ++$count >= $opt{perflimit};
  3700. }
  3701. }
  3702.  
  3703. ## Now we can set the critical and warning
  3704. for (@addcrit) {
  3705. add_critical $_;
  3706. $db->{perf} = '';
  3707. }
  3708. for (@addwarn) {
  3709. add_warning $_;
  3710. $db->{perf} = '';
  3711. }
  3712.  
  3713. if ($max == -1) {
  3714. add_unknown msg('no-match-rel');
  3715. }
  3716. elsif ($max != -1) {
  3717. add_ok $maxmsg;
  3718. }
  3719.  
  3720. if ($MRTG) {
  3721. keys %stats or bad_mrtg(msg('unknown-error'));
  3722. ## We are going to report the highest wasted bytes for table and index
  3723. my ($one,$two,$msg) = ('','');
  3724. ## Can also sort by ratio
  3725. my $sortby = exists $opt{mrtg} and $opt{mrtg} eq 'ratio' ? 1 : 0;
  3726. for (sort { $stats{table}{$b}->[$sortby] <=> $stats{table}{$a}->[$sortby] } keys %{$stats{table}}) {
  3727. $one = $stats{table}{$_}->[$sortby];
  3728. $msg = $_;
  3729. last;
  3730. }
  3731. for (sort { $stats{index}{$b}->[$sortby] <=> $stats{index}{$a}->[$sortby] } keys %{$stats{index}}) {
  3732. $two = $stats{index}{$_}->[$sortby];
  3733. $msg .= " $_";
  3734. last;
  3735. }
  3736. do_mrtg({one => $one, two => $two, msg => $msg});
  3737. }
  3738.  
  3739. return;
  3740.  
  3741. } ## end of check_bloat
  3742.  
  3743. sub check_checkpoint {
  3744.  
  3745. ## Checks how long in seconds since the last checkpoint on a WAL slave
  3746.  
  3747. ## Note that this value is actually the last checkpoint on the
  3748. ## *master* (as copied from the WAL checkpoint record), so it more
  3749. ## indicative that the master has been unable to complete a
  3750. ## checkpoint for some other reason (i.e., unable to write dirty
  3751. ## buffers or archive_command failure, etc). As such, this check
  3752. ## may make more sense on the master, or we may want to look at
  3753. ## the WAL segments received/processed instead of the checkpoint
  3754. ## timestamp.
  3755. ## This check can use the optional --assume-standby-mode or
  3756. ## --assume-prod: if the mode found is not the mode assumed, a
  3757. ## CRITICAL is emitted.
  3758.  
  3759. ## Supports: Nagios, MRTG
  3760. ## Warning and critical are seconds
  3761. ## Requires $ENV{PGDATA} or --datadir
  3762.  
  3763. my ($warning, $critical) = validate_range
  3764. ({
  3765. type => 'time',
  3766. leastone => 1,
  3767. forcemrtg => 1,
  3768. });
  3769.  
  3770. $db->{host} = '<none>';
  3771.  
  3772. ## Run pg_controldata, grab the time
  3773. $res = open_controldata();
  3774.  
  3775. my $regex = msg('checkpoint-po');
  3776. if ($res !~ /$regex\s*(.+)/) { ## no critic (ProhibitUnusedCapture)
  3777. ## Just in case, check the English one as well
  3778. $regex = msg_en('checkpoint-po');
  3779. if ($res !~ /$regex\s*(.+)/) {
  3780. ndie msg('checkpoint-noregex');
  3781. }
  3782. }
  3783. my $last = $1;
  3784.  
  3785. ## Convert to number of seconds
  3786. eval {
  3787. require Date::Parse;
  3788. import Date::Parse;
  3789. };
  3790. if ($@) {
  3791. ndie msg('checkpoint-nodp');
  3792. }
  3793. my $dt = str2time($last);
  3794. if ($dt !~ /^\d+$/) {
  3795. ndie msg('checkpoint-noparse', $last);
  3796. }
  3797. my $diff = time - $dt;
  3798. my $msg = $diff==1 ? msg('checkpoint-ok') : msg('checkpoint-ok2', $diff);
  3799. $db->{perf} = sprintf '%s=%s;%s;%s',
  3800. perfname(msg('age')), $diff, $warning, $critical;
  3801.  
  3802. my $mode = '';
  3803. if ($STANDBY) {
  3804. $mode = 'STANDBY';
  3805. }
  3806. if ($MASTER) {
  3807. $mode = 'MASTER';
  3808. }
  3809.  
  3810. ## If we have an assume flag, then honor it.
  3811. my $goodmode = 1;
  3812. if ($opt{'assume-standby-mode'} and not $STANDBY) {
  3813. $goodmode = 0;
  3814. $mode = 'NOT STANDBY';
  3815. }
  3816. elsif ($opt{'assume-prod'} and not $MASTER) {
  3817. $goodmode = 0;
  3818. $mode = 'NOT MASTER';
  3819. }
  3820.  
  3821. if (length($mode) > 0) {
  3822. $db->{perf} .= sprintf ' %s=%s',
  3823. perfname(msg('mode')), $mode;
  3824. }
  3825.  
  3826. if ($MRTG) {
  3827. do_mrtg({one => $diff, msg => $msg});
  3828. }
  3829.  
  3830. if ((length $critical and $diff >= $critical) or not $goodmode) {
  3831. add_critical $msg;
  3832. return;
  3833. }
  3834.  
  3835. if (length $warning and $diff >= $warning) {
  3836. add_warning $msg;
  3837. return;
  3838. }
  3839.  
  3840. add_ok $msg;
  3841.  
  3842. return;
  3843.  
  3844. } ## end of check_checkpoint
  3845.  
  3846.  
  3847. sub check_cluster_id {
  3848.  
  3849.  
  3850. ## Verify the Database System Identifier provided by pg_controldata
  3851. ## Supports: Nagios, MRTG
  3852. ## One of warning or critical must be given (but not both)
  3853. ## It should run one time to find out the expected cluster-id
  3854. ## You can use --critical="0" to find out the current cluster-id
  3855. ## You can include or exclude settings as well
  3856. ## Example:
  3857. ## check_postgres_cluster_id --critical="5633695740047915125"
  3858.  
  3859. my ($warning, $critical) = validate_range({type => 'integer_string', onlyone => 1});
  3860.  
  3861. $db->{host} = '<none>';
  3862.  
  3863. ## Run pg_controldata, grab the cluster-id
  3864. $res = open_controldata();
  3865.  
  3866. my $regex = msg('checkcluster-id');
  3867. if ($res !~ /$regex\s*(.+)/) { ## no critic (ProhibitUnusedCapture)
  3868. ## Just in case, check the English one as well
  3869. $regex = msg_en('checkcluster-id');
  3870. if ($res !~ /$regex\s*(.+)/) {
  3871. ndie msg('checkpoint-noregex');
  3872. }
  3873. }
  3874. my $ident = $1;
  3875.  
  3876. my $msg = msg('checkcluster-msg', $ident);
  3877. if ($MRTG) {
  3878. $opt{mrtg} or ndie msg('checksum-nomrtg');
  3879. do_mrtg({one => $opt{mrtg} eq $ident ? 1 : 0, msg => $ident});
  3880. }
  3881. if ($critical and $critical ne $ident) {
  3882. add_critical $msg;
  3883. }
  3884. elsif ($warning and $warning ne $ident) {
  3885. add_warning $msg;
  3886. }
  3887. elsif (!$critical and !$warning) {
  3888. add_unknown $msg;
  3889. }
  3890. else {
  3891. add_ok $msg;
  3892. }
  3893.  
  3894. return;
  3895.  
  3896. } ## end of check_cluster_id
  3897.  
  3898.  
  3899. sub check_commitratio {
  3900.  
  3901. ## Check the commitratio of one or more databases
  3902. ## Supports: Nagios, MRTG
  3903. ## mrtg reports the largest two databases
  3904. ## By default, checks all databases
  3905. ## Can check specific one(s) with include
  3906. ## Can ignore some with exclude
  3907. ## Warning and criticals are percentages
  3908. ## Limit to a specific user (db owner) with the includeuser option
  3909. ## Exclude users with the excludeuser option
  3910.  
  3911. my ($warning, $critical) = validate_range({type => 'percent'});
  3912.  
  3913. $SQL = qq{
  3914. SELECT
  3915. round(100.*sd.xact_commit/(sd.xact_commit+sd.xact_rollback), 2) AS dcommitratio,
  3916. d.datname,
  3917. u.usename
  3918. FROM pg_stat_database sd
  3919. JOIN pg_database d ON (d.oid=sd.datid)
  3920. JOIN pg_user u ON (u.usesysid=d.datdba)
  3921. WHERE sd.xact_commit+sd.xact_rollback<>0
  3922. $USERWHERECLAUSE
  3923. };
  3924. if ($opt{perflimit}) {
  3925. $SQL .= " ORDER BY 1 DESC LIMIT $opt{perflimit}";
  3926. }
  3927.  
  3928. my $info = run_command($SQL, { regex => qr{\d+}, emptyok => 1, } );
  3929. my $found = 0;
  3930.  
  3931. for $db (@{$info->{db}}) {
  3932. my $min = 101;
  3933. $found = 1;
  3934. my %s;
  3935. for my $r (@{$db->{slurp}}) {
  3936.  
  3937. next if skip_item($r->{datname});
  3938.  
  3939. if ($r->{dcommitratio} <= $min) {
  3940. $min = $r->{dcommitratio};
  3941. }
  3942. $s{$r->{datname}} = $r->{dcommitratio};
  3943. }
  3944.  
  3945. if ($MRTG) {
  3946. do_mrtg({one => $min, msg => "DB: $db->{dbname}"});
  3947. }
  3948. if ($min > 100) {
  3949. $stats{$db->{dbname}} = 0;
  3950. if ($USERWHERECLAUSE) {
  3951. add_ok msg('no-match-user');
  3952. }
  3953. else {
  3954. add_unknown msg('no-match-db');
  3955. }
  3956. next;
  3957. }
  3958.  
  3959. my $msg = '';
  3960. for (reverse sort {$s{$b} <=> $s{$a} or $a cmp $b } keys %s) {
  3961. $msg .= "$_: $s{$_} ";
  3962. $db->{perf} .= sprintf ' %s=%s;%s;%s',
  3963. perfname($_), $s{$_}, $warning, $critical;
  3964. }
  3965. if (length $critical and $min <= $critical) {
  3966. add_critical $msg;
  3967. }
  3968. elsif (length $warning and $min <= $warning) {
  3969. add_warning $msg;
  3970. }
  3971. else {
  3972. add_ok $msg;
  3973. }
  3974. }
  3975.  
  3976. ## If no results, probably a version problem
  3977. if (!$found and keys %unknown) {
  3978. (my $first) = values %unknown;
  3979. if ($first->[0][0] =~ /pg_database_size/) {
  3980. ndie msg('dbsize-version');
  3981. }
  3982. }
  3983.  
  3984. return;
  3985.  
  3986. } ## end of check_commitratio
  3987.  
  3988.  
  3989. sub check_connection {
  3990.  
  3991. ## Check the connection, get the connection time and version
  3992. ## No comparisons made: warning and critical are not allowed
  3993. ## Suports: Nagios, MRTG
  3994.  
  3995. if ($opt{warning} or $opt{critical}) {
  3996. ndie msg('range-none');
  3997. }
  3998.  
  3999. my $info = run_command('SELECT version() AS v');
  4000.  
  4001. for $db (@{$info->{db}}) {
  4002.  
  4003. my $err = $db->{error} || '';
  4004. if ($err =~ /FATAL/) {
  4005. $MRTG and do_mrtg({one => 0});
  4006. add_critical $db->{error};
  4007. return;
  4008. }
  4009.  
  4010. my $ver = ($db->{slurp}[0]{v} =~ /(\d+\.\d+\S+)/o) ? $1 : '';
  4011.  
  4012. $MRTG and do_mrtg({one => $ver ? 1 : 0});
  4013.  
  4014. if ($ver) {
  4015. add_ok msg('version', $ver);
  4016. }
  4017. else {
  4018. add_unknown msg('invalid-query', $db->{slurp}[0]{v});
  4019. }
  4020. }
  4021.  
  4022. return;
  4023.  
  4024. } ## end of check_connection
  4025.  
  4026.  
  4027. sub check_custom_query {
  4028.  
  4029. ## Run a user-supplied query, then parse the results
  4030. ## If you end up using this to make a useful query, consider making it
  4031. ## into a specific action and sending in a patch!
  4032. ## valtype must be one of: string, time, size, integer
  4033.  
  4034. my $valtype = $opt{valtype} || 'integer';
  4035.  
  4036. my ($warning, $critical) = validate_range({type => $valtype, leastone => 1});
  4037.  
  4038. my $query = $opt{query} or ndie msg('custom-nostring');
  4039.  
  4040. my $reverse = $opt{reverse} || 0;
  4041.  
  4042. my $info = run_command($query);
  4043.  
  4044. for $db (@{$info->{db}}) {
  4045.  
  4046. if (! @{$db->{slurp}}) {
  4047. add_unknown msg('custom-norows');
  4048. next;
  4049. }
  4050.  
  4051. my $goodrow = 0;
  4052.  
  4053. ## The other column tells it the name to use as the perfdata value
  4054. my $perfname;
  4055.  
  4056. for my $r (@{$db->{slurp}}) {
  4057. my $result = $r->{result};
  4058. if (! defined $perfname) {
  4059. $perfname = '';
  4060. for my $name (keys %$r) {
  4061. next if $name eq 'result';
  4062. $perfname = $name;
  4063. last;
  4064. }
  4065. }
  4066. $goodrow++;
  4067. if ($perfname) {
  4068. $db->{perf} .= sprintf ' %s=%s;%s;%s',
  4069. perfname($perfname), $r->{$perfname}, $warning, $critical;
  4070. }
  4071. my $gotmatch = 0;
  4072. if (! defined $result) {
  4073. add_unknown msg('custom-invalid');
  4074. return;
  4075. }
  4076. if (length $critical) {
  4077. if (($valtype eq 'string' and $result eq $critical)
  4078. or
  4079. ($valtype ne 'string' and $reverse ? $result <= $critical : $result >= $critical)) { ## covers integer, time, size
  4080. add_critical "$result";
  4081. $gotmatch = 1;
  4082. }
  4083. }
  4084.  
  4085. if (length $warning and ! $gotmatch) {
  4086. if (($valtype eq 'string' and $result eq $warning)
  4087. or
  4088. ($valtype ne 'string' and length $result and $reverse ? $result <= $warning : $result >= $warning)) {
  4089. add_warning "$result";
  4090. $gotmatch = 1;
  4091. }
  4092. }
  4093.  
  4094. if (! $gotmatch) {
  4095. add_ok "$result";
  4096. }
  4097.  
  4098. } ## end each row returned
  4099.  
  4100. if (!$goodrow) {
  4101. add_unknown msg('custom-invalid');
  4102. }
  4103. }
  4104.  
  4105. return;
  4106.  
  4107. } ## end of check_custom_query
  4108.  
  4109.  
  4110. sub check_database_size {
  4111.  
  4112. ## Check the size of one or more databases
  4113. ## Supports: Nagios, MRTG
  4114. ## mrtg reports the largest two databases
  4115. ## By default, checks all databases
  4116. ## Can check specific one(s) with include
  4117. ## Can ignore some with exclude
  4118. ## Warning and critical are bytes
  4119. ## Valid units: b, k, m, g, t, e
  4120. ## All above may be written as plural or with a trailing 'b'
  4121. ## Limit to a specific user (db owner) with the includeuser option
  4122. ## Exclude users with the excludeuser option
  4123.  
  4124. my ($warning, $critical) = validate_range({type => 'size'});
  4125.  
  4126. $USERWHERECLAUSE =~ s/AND/WHERE/;
  4127.  
  4128. $SQL = qq{
  4129. SELECT pg_database_size(d.oid) AS dsize,
  4130. pg_size_pretty(pg_database_size(d.oid)) AS pdsize,
  4131. datname,
  4132. usename
  4133. FROM pg_database d
  4134. LEFT JOIN pg_user u ON (u.usesysid=d.datdba)$USERWHERECLAUSE
  4135. };
  4136. if ($opt{perflimit}) {
  4137. $SQL .= " ORDER BY 1 DESC LIMIT $opt{perflimit}";
  4138. }
  4139.  
  4140. my $info = run_command($SQL, { regex => qr{\d+}, emptyok => 1, } );
  4141. my $found = 0;
  4142.  
  4143. for $db (@{$info->{db}}) {
  4144. my $max = -1;
  4145. $found = 1;
  4146. my %s;
  4147. for my $r (@{$db->{slurp}}) {
  4148.  
  4149. next if skip_item($r->{datname});
  4150.  
  4151. if ($r->{dsize} >= $max) {
  4152. $max = $r->{dsize};
  4153. }
  4154. $s{$r->{datname}} = [$r->{dsize},$r->{pdsize}];
  4155. }
  4156.  
  4157. if ($MRTG) {
  4158. do_mrtg({one => $max, msg => "DB: $db->{dbname}"});
  4159. }
  4160. if ($max < 0) {
  4161. $stats{$db->{dbname}} = 0;
  4162. if ($USERWHERECLAUSE) {
  4163. add_ok msg('no-match-user');
  4164. }
  4165. else {
  4166. add_unknown msg('no-match-db');
  4167. }
  4168. next;
  4169. }
  4170.  
  4171. my $msg = '';
  4172. for (sort {$s{$b}[0] <=> $s{$a}[0] or $a cmp $b } keys %s) {
  4173. $msg .= "$_: $s{$_}[0] ($s{$_}[1]) ";
  4174. $db->{perf} .= sprintf ' %s=%s;%s;%s',
  4175. perfname($_), $s{$_}[0], $warning, $critical;
  4176. }
  4177. if (length $critical and $max >= $critical) {
  4178. add_critical $msg;
  4179. }
  4180. elsif (length $warning and $max >= $warning) {
  4181. add_warning $msg;
  4182. }
  4183. else {
  4184. add_ok $msg;
  4185. }
  4186. }
  4187.  
  4188. ## If no results, probably a version problem
  4189. if (!$found and keys %unknown) {
  4190. (my $first) = values %unknown;
  4191. if ($first->[0][0] =~ /pg_database_size/) {
  4192. ndie msg('dbsize-version');
  4193. }
  4194. }
  4195.  
  4196. return;
  4197.  
  4198. } ## end of check_database_size
  4199.  
  4200.  
  4201. sub check_dbstats {
  4202.  
  4203. ## Returns values from the pg_stat_database view
  4204. ## Supports: Cacti
  4205. ## Assumes psql and target are the same version for the 8.3 check
  4206.  
  4207. my ($warning, $critical) = validate_range
  4208. ({
  4209. type => 'cacti',
  4210. });
  4211.  
  4212. my $SQL = q{SELECT datname,
  4213. numbackends AS backends,xact_commit AS commits,xact_rollback AS rollbacks,
  4214. blks_read AS read, blks_hit AS hit};
  4215. if ($opt{dbname}) {
  4216. $SQL .= q{
  4217. ,(SELECT SUM(idx_scan) FROM pg_stat_user_indexes) AS idxscan
  4218. ,COALESCE((SELECT SUM(idx_tup_read) FROM pg_stat_user_indexes),0) AS idxtupread
  4219. ,COALESCE((SELECT SUM(idx_tup_fetch) FROM pg_stat_user_indexes),0) AS idxtupfetch
  4220. ,COALESCE((SELECT SUM(idx_blks_read) FROM pg_statio_user_indexes),0) AS idxblksread
  4221. ,COALESCE((SELECT SUM(idx_blks_hit) FROM pg_statio_user_indexes),0) AS idxblkshit
  4222. ,COALESCE((SELECT SUM(seq_scan) FROM pg_stat_user_tables),0) AS seqscan
  4223. ,COALESCE((SELECT SUM(seq_tup_read) FROM pg_stat_user_tables),0) AS seqtupread
  4224. };
  4225. }
  4226. $SQL .= q{ FROM pg_stat_database};
  4227. (my $SQL2 = $SQL) =~ s/AS seqtupread/AS seqtupread, tup_returned AS ret, tup_fetched AS fetch, tup_inserted AS ins, tup_updated AS upd, tup_deleted AS del/;
  4228.  
  4229. my $info = run_command($SQL, {regex => qr{\w}, version => [ ">8.2 $SQL2" ] } );
  4230.  
  4231. for $db (@{$info->{db}}) {
  4232. ROW: for my $r (@{$db->{slurp}}) {
  4233.  
  4234. my $dbname = $r->{datname};
  4235.  
  4236. next ROW if skip_item($dbname);
  4237.  
  4238. ## If dbnames were specififed, use those for filtering as well
  4239. if (@{$opt{dbname}}) {
  4240. my $keepit = 0;
  4241. for my $drow (@{$opt{dbname}}) {
  4242. for my $d (split /,/ => $drow) {
  4243. $d eq $dbname and $keepit = 1;
  4244. }
  4245. }
  4246. next ROW unless $keepit;
  4247. }
  4248.  
  4249. my $msg = '';
  4250. for my $col (qw/
  4251. backends commits rollbacks
  4252. read hit
  4253. idxscan idxtupread idxtupfetch idxblksread idxblkshit
  4254. seqscan seqtupread
  4255. ret fetch ins upd del/) {
  4256. $msg .= "$col:";
  4257. $msg .= (exists $r->{$col} and length $r->{$col}) ? $r->{$col} : 0;
  4258. $msg .= ' ';
  4259. }
  4260. print "${msg}dbname:$dbname\n";
  4261. }
  4262. }
  4263.  
  4264. exit 0;
  4265.  
  4266. } ## end of check_dbstats
  4267.  
  4268.  
  4269. sub check_disabled_triggers {
  4270.  
  4271. ## Checks how many disabled triggers are in the database
  4272. ## Supports: Nagios, MRTG
  4273. ## Warning and critical are integers, defaults to 1
  4274.  
  4275. my ($warning, $critical) = validate_range
  4276. ({
  4277. type => 'positive integer',
  4278. default_warning => 1,
  4279. default_critical => 1,
  4280. forcemrtg => 1,
  4281. });
  4282.  
  4283. $SQL = q{
  4284. SELECT tgrelid::regclass AS tname, tgname, tgenabled
  4285. FROM pg_trigger
  4286. WHERE tgenabled IS NOT TRUE ORDER BY tgname
  4287. };
  4288. my $SQL83 = q{
  4289. SELECT tgrelid::regclass AS tname, tgname, tgenabled
  4290. FROM pg_trigger
  4291. WHERE tgenabled = 'D' ORDER BY tgname
  4292. };
  4293. my $SQLOLD = q{SELECT 'FAIL' AS fail};
  4294.  
  4295. my $info = run_command($SQL, { version => [ ">8.2 $SQL83", "<8.1 $SQLOLD" ] } );
  4296.  
  4297. if (exists $info->{db}[0]{fail}) {
  4298. ndie msg('die-action-version', $action, '8.1', $db->{version});
  4299. }
  4300.  
  4301. my $count = 0;
  4302. my $dislis = '';
  4303. for $db (@{$info->{db}}) {
  4304.  
  4305. ROW: for my $r (@{$db->{slurp}}) {
  4306. $count++;
  4307. $dislis .= " $r->{tname}=>$r->{tgname}";
  4308. }
  4309. $MRTG and do_mrtg({one => $count});
  4310.  
  4311. my $msg = msg('trigger-msg', "$count$dislis");
  4312.  
  4313. if ($critical and $count >= $critical) {
  4314. add_critical $msg;
  4315. }
  4316. elsif ($warning and $count >= $warning) {
  4317. add_warning $msg;
  4318. }
  4319. else {
  4320. add_ok $msg;
  4321. }
  4322. }
  4323.  
  4324. return;
  4325.  
  4326. } ## end of check_disabled_triggers
  4327.  
  4328.  
  4329. sub check_disk_space {
  4330.  
  4331. ## Check the available disk space used by postgres
  4332. ## Supports: Nagios, MRTG
  4333. ## Requires the executable "/bin/df"
  4334. ## Must run as a superuser in the database (to examine 'data_directory' setting)
  4335. ## Critical and warning are maximum size, or percentages
  4336. ## Example: --critical="40 GB"
  4337. ## NOTE: Needs to run on the same system (for now)
  4338. ## XXX Allow custom ssh commands for remote df and the like
  4339.  
  4340. my ($warning, $critical) = validate_size_or_percent_with_oper
  4341. ({
  4342. default_warning => '90%',
  4343. default_critical => '95%',
  4344. });
  4345.  
  4346. -x '/bin/df' or ndie msg('diskspace-nodf');
  4347.  
  4348. ## Figure out where everything is.
  4349. $SQL = q{
  4350. SELECT 'S' AS syn, name AS nn, setting AS val
  4351. FROM pg_settings
  4352. WHERE name = 'data_directory'
  4353. OR name ='log_directory'
  4354. UNION ALL
  4355. SELECT 'T' AS syn, spcname AS nn, spclocation AS val
  4356. FROM pg_tablespace
  4357. WHERE spclocation <> ''
  4358. };
  4359.  
  4360. my $info = run_command($SQL);
  4361.  
  4362. my %dir; ## 1 = normal 2 = been checked -1 = does not exist
  4363. my %seenfs;
  4364. for $db (@{$info->{db}}) {
  4365. my %i;
  4366. for my $r (@{$db->{slurp}}) {
  4367. $i{$r->{syn}}{$r->{nn}} = $r->{val};
  4368. }
  4369. if (! exists $i{S}{data_directory}) {
  4370. add_unknown msg('diskspace-nodata');
  4371. next;
  4372. }
  4373. my ($datadir,$logdir) = ($i{S}{data_directory},$i{S}{log_directory}||'');
  4374.  
  4375. if (!exists $dir{$datadir}) {
  4376. if (! -d $datadir) {
  4377. add_unknown msg('diskspace-nodir', $datadir);
  4378. $dir{$datadir} = -1;
  4379. next;
  4380. }
  4381. $dir{$datadir} = 1;
  4382.  
  4383. ## Check if the WAL files are on a separate disk
  4384. my $xlog = "$datadir/pg_xlog";
  4385. if (-l $xlog) {
  4386. my $linkdir = readlink($xlog);
  4387. $dir{$linkdir} = 1 if ! exists $dir{$linkdir};
  4388. }
  4389. }
  4390.  
  4391. ## Check log_directory: relative or absolute
  4392. if (length $logdir) {
  4393. if ($logdir =~ /^\w/) { ## relative, check only if symlinked
  4394. $logdir = "$datadir/$logdir";
  4395. if (-l $logdir) {
  4396. my $linkdir = readlink($logdir);
  4397. $dir{$linkdir} = 1 if ! exists $dir{$linkdir};
  4398. }
  4399. }
  4400. else { ## absolute, always check
  4401. if ($logdir ne $datadir and ! exists $dir{$logdir}) {
  4402. $dir{$logdir} = 1;
  4403. }
  4404. }
  4405. }
  4406.  
  4407. ## Check all tablespaces
  4408. for my $tsname (keys %{$i{T}}) {
  4409. my $tsdir = $i{T}{$tsname};
  4410. $dir{$tsdir} = 1 if ! exists $dir{$tsdir};
  4411. }
  4412.  
  4413. my $gotone = 0;
  4414. for my $dir (keys %dir) {
  4415. next if $dir{$dir} != 1;
  4416.  
  4417. $dir{$dir} = 1;
  4418.  
  4419. $COM = qq{/bin/df -kP "$dir" 2>&1};
  4420. $res = qx{$COM};
  4421.  
  4422. if ($res !~ /^.+\n(\S+)\s+(\d+)\s+(\d+)\s+(\d+)\s+(\d+)\%\s+(\S+)/) {
  4423. ndie msg('diskspace-fail', $COM, $res);
  4424. }
  4425. my ($fs,$total,$used,$avail,$percent,$mount) = ($1,$2*1024,$3*1024,$4*1024,$5,$6);
  4426.  
  4427. ## If we've already done this one, skip it
  4428. next if $seenfs{$fs}++;
  4429.  
  4430. next if skip_item($fs);
  4431.  
  4432. if ($MRTG) {
  4433. $stats{$fs} = [$total,$used,$avail,$percent];
  4434. next;
  4435. }
  4436.  
  4437. $gotone = 1;
  4438.  
  4439. ## Rather than make another call with -h, do it ourselves
  4440. my $prettyused = pretty_size($used);
  4441. my $prettytotal = pretty_size($total);
  4442.  
  4443. my $msg = msg('diskspace-msg', $fs, $mount, $prettyused, $prettytotal, $percent);
  4444.  
  4445. $db->{perf} = sprintf '%s=%sB',
  4446. perfname(msg('size')), $used;
  4447.  
  4448. my $ok = 1;
  4449. if ($critical->($used, $percent)) {
  4450. add_critical $msg;
  4451. $ok = 0;
  4452. }
  4453.  
  4454. if ($ok and $warning->($used, $percent)) {
  4455. add_warning $msg;
  4456. $ok = 0;
  4457. }
  4458.  
  4459. if ($ok) {
  4460. add_ok $msg;
  4461. }
  4462. } ## end each dir
  4463.  
  4464. next if $MRTG;
  4465.  
  4466. if (!$gotone) {
  4467. add_unknown msg('no-match-fs');
  4468. }
  4469. }
  4470.  
  4471. if ($MRTG) {
  4472. keys %stats or bad_mrtg(msg('unknown-error'));
  4473. ## Get the highest by total size or percent (total, used, avail, percent)
  4474. ## We default to 'available'
  4475. my $sortby = exists $opt{mrtg}
  4476. ? $opt{mrtg} eq 'total' ? 0
  4477. : $opt{mrtg} eq 'used' ? 1
  4478. : $opt{mrtg} eq 'avail' ? 2
  4479. : $opt{mrtg} eq 'percent' ? 3 : 2 : 2;
  4480. my ($one,$two,$msg) = ('','','');
  4481. for (sort { $stats{$b}->[$sortby] <=> $stats{$a}->[$sortby] } keys %stats) {
  4482. if ($one eq '') {
  4483. $one = $stats{$_}->[$sortby];
  4484. $msg = $_;
  4485. next;
  4486. }
  4487. $two = $stats{$_}->[$sortby];
  4488. last;
  4489. }
  4490. do_mrtg({one => $one, two => $two, msg => $msg});
  4491. }
  4492.  
  4493. return;
  4494.  
  4495. } ## end of check_disk_space
  4496.  
  4497.  
  4498. sub check_fsm_pages {
  4499.  
  4500. ## Check on the percentage of free space map pages in use
  4501. ## Supports: Nagios, MRTG
  4502. ## Must run as superuser
  4503. ## Requires pg_freespacemap contrib module
  4504. ## Critical and warning are a percentage of max_fsm_pages
  4505. ## Example: --critical=95
  4506.  
  4507. my ($warning, $critical) = validate_range
  4508. ({
  4509. type => 'percent',
  4510. default_warning => '85%',
  4511. default_critical => '95%',
  4512. });
  4513.  
  4514. (my $w = $warning) =~ s/\D//;
  4515. (my $c = $critical) =~ s/\D//;
  4516. my $SQL = q{
  4517. SELECT pages, maxx, ROUND(100*(pages/maxx)) AS percent
  4518. FROM
  4519. (SELECT (sumrequests+numrels)*chunkpages AS pages
  4520. FROM (SELECT SUM(CASE WHEN avgrequest IS NULL
  4521. THEN interestingpages/32 ELSE interestingpages/16 END) AS sumrequests,
  4522. COUNT(relfilenode) AS numrels, 16 AS chunkpages FROM pg_freespacemap_relations) AS foo) AS foo2,
  4523. (SELECT setting::NUMERIC AS maxx FROM pg_settings WHERE name = 'max_fsm_pages') AS foo3
  4524. };
  4525. my $SQLNOOP = q{SELECT 'FAIL' AS fail};
  4526.  
  4527. my $info = run_command($SQL, { version => [ ">8.3 $SQLNOOP" ] } );
  4528.  
  4529. if (exists $info->{db}[0]{slurp}[0]{fail}) {
  4530. add_unknown msg('fsm-page-highver');
  4531. return;
  4532. }
  4533.  
  4534. for $db (@{$info->{db}}) {
  4535. for my $r (@{$db->{slurp}}) {
  4536. my ($pages,$max,$percent) = ($r->{pages}||0,$r->{maxx},$r->{percent}||0);
  4537.  
  4538. $MRTG and do_mrtg({one => $percent, two => $pages});
  4539.  
  4540. my $msg = msg('fsm-page-msg', $pages, $max, $percent);
  4541.  
  4542. if (length $critical and $percent >= $c) {
  4543. add_critical $msg;
  4544. }
  4545. elsif (length $warning and $percent >= $w) {
  4546. add_warning $msg;
  4547. }
  4548. else {
  4549. add_ok $msg;
  4550. }
  4551. }
  4552. }
  4553.  
  4554. return;
  4555.  
  4556. } ## end of check_fsm_pages
  4557.  
  4558.  
  4559. sub check_fsm_relations {
  4560.  
  4561. ## Check on the % of free space map relations in use
  4562. ## Supports: Nagios, MRTG
  4563. ## Must run as superuser
  4564. ## Requires pg_freespacemap contrib module
  4565. ## Critical and warning are a percentage of max_fsm_relations
  4566. ## Example: --critical=95
  4567.  
  4568. my ($warning, $critical) = validate_range
  4569. ({
  4570. type => 'percent',
  4571. default_warning => '85%',
  4572. default_critical => '95%',
  4573. });
  4574.  
  4575. (my $w = $warning) =~ s/\D//;
  4576. (my $c = $critical) =~ s/\D//;
  4577.  
  4578. my $SQL = q{
  4579. SELECT maxx, cur, ROUND(100*(cur/maxx)) AS percent
  4580. FROM (SELECT
  4581. (SELECT COUNT(*) FROM pg_freespacemap_relations) AS cur,
  4582. (SELECT setting::NUMERIC FROM pg_settings WHERE name='max_fsm_relations') AS maxx) x
  4583. };
  4584. my $SQLNOOP = q{SELECT 'FAIL' AS fail};
  4585.  
  4586. my $info = run_command($SQL, { version => [ ">8.3 $SQLNOOP" ] } );
  4587.  
  4588. if (exists $info->{db}[0]{slurp}[0]{fail}) {
  4589. add_unknown msg('fsm-rel-highver');
  4590. return;
  4591. }
  4592.  
  4593. for $db (@{$info->{db}}) {
  4594.  
  4595. for my $r (@{$db->{slurp}}) {
  4596. my ($max,$cur,$percent) = ($r->{maxx},$r->{cur},$r->{percent}||0);
  4597.  
  4598. $MRTG and do_mrtg({one => $percent, two => $cur});
  4599.  
  4600. my $msg = msg('fsm-rel-msg', $cur, $max, $percent);
  4601.  
  4602. if (length $critical and $percent >= $c) {
  4603. add_critical $msg;
  4604. }
  4605. elsif (length $warning and $percent >= $w) {
  4606. add_warning $msg;
  4607. }
  4608. else {
  4609. add_ok $msg;
  4610. }
  4611. }
  4612.  
  4613. }
  4614.  
  4615. return;
  4616.  
  4617. } ## end of check_fsm_relations
  4618.  
  4619.  
  4620. sub check_hitratio {
  4621.  
  4622. ## Check the hitratio of one or more databases
  4623. ## Supports: Nagios, MRTG
  4624. ## mrtg reports the largest two databases
  4625. ## By default, checks all databases
  4626. ## Can check specific one(s) with include
  4627. ## Can ignore some with exclude
  4628. ## Warning and criticals are percentages
  4629. ## Limit to a specific user (db owner) with the includeuser option
  4630. ## Exclude users with the excludeuser option
  4631.  
  4632. my ($warning, $critical) = validate_range({type => 'percent'});
  4633.  
  4634. $SQL = qq{
  4635. SELECT
  4636. round(100.*sd.blks_hit/(sd.blks_read+sd.blks_hit), 2) AS dhitratio,
  4637. d.datname,
  4638. u.usename
  4639. FROM pg_stat_database sd
  4640. JOIN pg_database d ON (d.oid=sd.datid)
  4641. JOIN pg_user u ON (u.usesysid=d.datdba)
  4642. WHERE sd.blks_read+sd.blks_hit<>0
  4643. $USERWHERECLAUSE
  4644. };
  4645. if ($opt{perflimit}) {
  4646. $SQL .= " ORDER BY 1 DESC LIMIT $opt{perflimit}";
  4647. }
  4648.  
  4649. my $info = run_command($SQL, { regex => qr{\d+}, emptyok => 1, } );
  4650. my $found = 0;
  4651.  
  4652. for $db (@{$info->{db}}) {
  4653. my $min = 101;
  4654. $found = 1;
  4655. my %s;
  4656. for my $r (@{$db->{slurp}}) {
  4657.  
  4658. next if skip_item($r->{datname});
  4659.  
  4660. if ($r->{dhitratio} <= $min) {
  4661. $min = $r->{dhitratio};
  4662. }
  4663. $s{$r->{datname}} = $r->{dhitratio};
  4664. }
  4665.  
  4666. if ($MRTG) {
  4667. do_mrtg({one => $min, msg => "DB: $db->{dbname}"});
  4668. }
  4669. if ($min > 100) {
  4670. $stats{$db->{dbname}} = 0;
  4671. if ($USERWHERECLAUSE) {
  4672. add_ok msg('no-match-user');
  4673. }
  4674. else {
  4675. add_unknown msg('no-match-db');
  4676. }
  4677. next;
  4678. }
  4679.  
  4680. my $msg = '';
  4681. for (reverse sort {$s{$b} <=> $s{$a} or $a cmp $b } keys %s) {
  4682. $msg .= "$_: $s{$_} ";
  4683. $db->{perf} .= sprintf ' %s=%s;%s;%s',
  4684. perfname($_), $s{$_}, $warning, $critical;
  4685. }
  4686. if (length $critical and $min <= $critical) {
  4687. add_critical $msg;
  4688. }
  4689. elsif (length $warning and $min <= $warning) {
  4690. add_warning $msg;
  4691. }
  4692. else {
  4693. add_ok $msg;
  4694. }
  4695. }
  4696.  
  4697. ## If no results, probably a version problem
  4698. if (!$found and keys %unknown) {
  4699. (my $first) = values %unknown;
  4700. if ($first->[0][0] =~ /pg_database_size/) {
  4701. ndie msg('dbsize-version');
  4702. }
  4703. }
  4704.  
  4705. return;
  4706.  
  4707. } ## end of check_hitratio
  4708.  
  4709.  
  4710. sub check_hot_standby_delay {
  4711.  
  4712. ## Check on the delay in PITR replication between master and slave
  4713. ## Supports: Nagios, MRTG
  4714. ## Critical and warning are the delay between master and slave xlog locations
  4715. ## Example: --critical=1024
  4716.  
  4717. my ($warning, $critical) = validate_range({type => 'integer', leastone => 1});
  4718.  
  4719. # check if master and slave comply with the check using pg_is_in_recovery()
  4720. my ($master, $slave);
  4721. $SQL = q{SELECT pg_is_in_recovery() AS recovery;};
  4722.  
  4723. # Check if master is online (e.g. really a master)
  4724. for my $x (1..2) {
  4725. my $info = run_command($SQL, { dbnumber => $x, regex => qr(t|f) });
  4726.  
  4727. for $db (@{$info->{db}}) {
  4728. my $status = $db->{slurp}[0];
  4729. if ($status->{recovery} eq 't') {
  4730. $slave = $x;
  4731. last;
  4732. }
  4733. if ($status->{recovery} eq 'f') {
  4734. $master = $x;
  4735. last;
  4736. }
  4737. }
  4738. }
  4739. if (! defined $slave and ! defined $master) {
  4740. add_unknown msg('hs-no-role');
  4741. return;
  4742. }
  4743.  
  4744. ## If the slave is "db1" and master "db2", go ahead and switch them around for clearer output
  4745. if (1 == $slave) {
  4746. ($slave, $master) = (2, 1);
  4747. for my $k (qw(host port dbname dbuser dbpass)) {
  4748. ($opt{$k}, $opt{$k . 2}) = ($opt{$k . 2}, $opt{$k}); ## no critic (ProhibitMismatchedOperators)
  4749. }
  4750. }
  4751.  
  4752. ## Get xlog positions
  4753. my ($moffset, $s_rec_offset, $s_rep_offset);
  4754. ## On master
  4755. $SQL = q{SELECT pg_current_xlog_location() AS location};
  4756. my $info = run_command($SQL, { dbnumber => $master });
  4757. my $saved_db;
  4758. for $db (@{$info->{db}}) {
  4759. my $location = $db->{slurp}[0]{location};
  4760. next if ! defined $location;
  4761.  
  4762. my ($x, $y) = split(/\//, $location);
  4763. $moffset = (hex('ff000000') * hex($x)) + hex($y);
  4764. $saved_db = $db if ! defined $saved_db;
  4765. }
  4766.  
  4767. if (! defined $moffset) {
  4768. add_unknown msg('hs-no-location', 'master');
  4769. return;
  4770. }
  4771.  
  4772. ## On slave
  4773. $SQL = q{SELECT pg_last_xlog_receive_location() AS receive, pg_last_xlog_replay_location() AS replay};
  4774.  
  4775. $info = run_command($SQL, { dbnumber => $slave, regex => qr/\// });
  4776.  
  4777. for $db (@{$info->{db}}) {
  4778. my $receive = $db->{slurp}[0]{receive};
  4779. my $replay = $db->{slurp}[0]{replay};
  4780.  
  4781. if (defined $receive) {
  4782. my ($a, $b) = split(/\//, $receive);
  4783. $s_rec_offset = (hex('ff000000') * hex($a)) + hex($b);
  4784. }
  4785.  
  4786. if (defined $replay) {
  4787. my ($a, $b) = split(/\//, $replay);
  4788. $s_rep_offset = (hex('ff000000') * hex($a)) + hex($b);
  4789. }
  4790.  
  4791. $saved_db = $db if ! defined $saved_db;
  4792. }
  4793.  
  4794. if (! defined $s_rec_offset and ! defined $s_rep_offset) {
  4795. add_unknown msg('hs-no-location', 'slave');
  4796. return;
  4797. }
  4798.  
  4799. ## Compute deltas
  4800. $db = $saved_db;
  4801. my $rec_delta = $moffset - $s_rec_offset;
  4802. my $rep_delta = $moffset - $s_rep_offset;
  4803.  
  4804. $MRTG and do_mrtg({one => $rep_delta, two => $rec_delta});
  4805.  
  4806. $db->{perf} = sprintf ' %s=%s;%s;%s ',
  4807. perfname(msg('hs-replay-delay')), $rep_delta, $warning, $critical;
  4808. $db->{perf} .= sprintf ' %s=%s;%s;%s',
  4809. perfname(msg('hs-receive-delay')), $rec_delta, $warning, $critical;
  4810.  
  4811. ## Do the check on replay delay in case SR has disconnected because it way too far behind
  4812. my $msg = qq{$rep_delta};
  4813. if (length $critical and $rep_delta > $critical) {
  4814. add_critical $msg;
  4815. }
  4816. elsif (length $warning and $rep_delta > $warning) {
  4817. add_warning $msg;
  4818. }
  4819. else {
  4820. add_ok $msg;
  4821. }
  4822.  
  4823. return;
  4824.  
  4825. } ## end of check_hot_standby_delay
  4826.  
  4827.  
  4828. sub check_last_analyze {
  4829. my $auto = shift || '';
  4830. return check_last_vacuum_analyze('analyze', $auto);
  4831. }
  4832.  
  4833.  
  4834. sub check_last_vacuum {
  4835. my $auto = shift || '';
  4836. return check_last_vacuum_analyze('vacuum', $auto);
  4837. }
  4838.  
  4839.  
  4840. sub check_last_vacuum_analyze {
  4841.  
  4842. my $type = shift || 'vacuum';
  4843. my $auto = shift || 0;
  4844.  
  4845. ## Check the last time things were vacuumed or analyzed
  4846. ## Supports: Nagios, MRTG
  4847. ## NOTE: stats_row_level must be set to on in your database (if version 8.2)
  4848. ## By default, reports on the oldest value in the database
  4849. ## Can exclude and include tables
  4850. ## Warning and critical are times, default to seconds
  4851. ## Valid units: s[econd], m[inute], h[our], d[ay]
  4852. ## All above may be written as plural as well (e.g. "2 hours")
  4853. ## Limit to a specific user (relation owner) with the includeuser option
  4854. ## Exclude users with the excludeuser option
  4855. ## Example:
  4856. ## --exclude=~pg_ --include=pg_class,pg_attribute
  4857.  
  4858. my ($warning, $critical) = validate_range
  4859. ({
  4860. type => 'time',
  4861. default_warning => '1 day',
  4862. default_critical => '2 days',
  4863. });
  4864.  
  4865. my $criteria = $auto ?
  4866. qq{pg_stat_get_last_auto${type}_time(c.oid)}
  4867. : qq{GREATEST(pg_stat_get_last_${type}_time(c.oid), pg_stat_get_last_auto${type}_time(c.oid))};
  4868.  
  4869. ## Do include/exclude earlier for large pg_classes?
  4870. $SQL = qq{
  4871. SELECT current_database() AS datname, nspname AS sname, relname AS tname,
  4872. CASE WHEN v IS NULL THEN -1 ELSE round(extract(epoch FROM now()-v)) END AS ltime,
  4873. CASE WHEN v IS NULL THEN '?' ELSE TO_CHAR(v, '$SHOWTIME') END AS ptime
  4874. FROM (SELECT nspname, relname, $criteria AS v
  4875. FROM pg_class c, pg_namespace n
  4876. WHERE relkind = 'r'
  4877. AND n.oid = c.relnamespace
  4878. AND n.nspname <> 'information_schema'
  4879. ORDER BY 3) AS foo
  4880. };
  4881. if ($opt{perflimit}) {
  4882. $SQL .= ' ORDER BY 4 DESC';
  4883. }
  4884.  
  4885. if ($USERWHERECLAUSE) {
  4886. $SQL =~ s/ WHERE/, pg_user u WHERE u.usesysid=c.relowner$USERWHERECLAUSE AND/;
  4887. }
  4888.  
  4889. my $info = run_command($SQL, { regex => qr{\w}, emptyok => 1 } );
  4890.  
  4891. for $db (@{$info->{db}}) {
  4892.  
  4893. if (! @{$db->{slurp}} and $USERWHERECLAUSE) {
  4894. $stats{$db->{dbname}} = 0;
  4895. add_ok msg('no-match-user');
  4896. return;
  4897. }
  4898.  
  4899. ## -1 means no tables found at all
  4900. ## -2 means exclusion rules took effect
  4901. ## -3 means no tables were ever vacuumed/analyzed
  4902. my $maxtime = -1;
  4903. my $maxptime = '?';
  4904. my ($minrel,$maxrel) = ('?','?'); ## no critic
  4905. my $mintime = 0; ## used for MRTG only
  4906. my $count = 0;
  4907. my $found = 0;
  4908. ROW: for my $r (@{$db->{slurp}}) {
  4909. my ($dbname,$schema,$name,$time,$ptime) = @$r{qw/ datname sname tname ltime ptime/};
  4910. if (skip_item($name, $schema)) {
  4911. $maxtime = -2 if $maxtime < 1;
  4912. next ROW;
  4913. }
  4914. $found++;
  4915. if ($time >= 0) {
  4916. $db->{perf} .= sprintf ' %s=%ss;%s;%s',
  4917. perfname("$dbname.$schema.$name"),$time, $warning, $critical;
  4918. }
  4919. if ($time > $maxtime) {
  4920. $maxtime = $time;
  4921. $maxrel = "DB: $dbname TABLE: $schema.$name";
  4922. $maxptime = $ptime;
  4923. }
  4924. if ($time > 0 and ($time < $mintime or !$mintime)) {
  4925. $mintime = $time;
  4926. $minrel = "DB: $dbname TABLE: $schema.$name";
  4927. }
  4928. if ($opt{perflimit}) {
  4929. last if ++$count >= $opt{perflimit};
  4930. }
  4931. }
  4932. if ($MRTG) {
  4933. $maxrel eq '?' and $maxrel = "DB: $db->{dbname} TABLE: ?";
  4934. do_mrtg({one => $mintime, msg => $maxrel});
  4935. return;
  4936. }
  4937. if ($maxtime == -2) {
  4938. add_unknown (
  4939. $found ? $type eq 'vacuum' ? msg('vac-nomatch-v')
  4940. : msg('vac-nomatch-a')
  4941. : msg('no-match-table') ## no critic (RequireTrailingCommaAtNewline)
  4942. );
  4943. }
  4944. elsif ($maxtime < 0) {
  4945. add_unknown $type eq 'vacuum' ? msg('vac-nomatch-v') : msg('vac-nomatch-a');
  4946. }
  4947. else {
  4948. my $showtime = pretty_time($maxtime, 'S');
  4949. my $msg = "$maxrel: $maxptime ($showtime)";
  4950. if ($critical and $maxtime >= $critical) {
  4951. add_critical $msg;
  4952. }
  4953. elsif ($warning and $maxtime >= $warning) {
  4954. add_warning $msg;
  4955. }
  4956. else {
  4957. add_ok $msg;
  4958. }
  4959. }
  4960. }
  4961.  
  4962. return;
  4963.  
  4964. } ## end of check_last_vacuum_analyze
  4965.  
  4966.  
  4967. sub check_listener {
  4968.  
  4969. ## Check for a specific listener
  4970. ## Supports: Nagios, MRTG
  4971. ## Critical and warning are simple strings, or regex if starts with a ~
  4972. ## Example: --critical="~bucardo"
  4973.  
  4974. if ($MRTG and exists $opt{mrtg}) {
  4975. $opt{critical} = $opt{mrtg};
  4976. }
  4977.  
  4978. my ($warning, $critical) = validate_range({type => 'restringex', forcemrtg => 1});
  4979.  
  4980. my $string = length $critical ? $critical : $warning;
  4981. my $regex = ($string =~ s/^~//) ? '~' : '=';
  4982.  
  4983. $SQL = "SELECT count(*) AS c FROM pg_listener WHERE relname $regex '$string'";
  4984. my $info = run_command($SQL);
  4985.  
  4986. for $db (@{$info->{db}}) {
  4987. if ($db->{slurp}[0]{c} !~ /(\d+)/) {
  4988. add_unknown msg('invalid-query', $db->{slurp});
  4989. next;
  4990. }
  4991. my $count = $1;
  4992. if ($MRTG) {
  4993. do_mrtg({one => $count});
  4994. }
  4995. $db->{perf} .= sprintf '%s=%s',
  4996. perfname(msg('listening')), $count;
  4997. my $msg = msg('listener-msg', $count);
  4998. if ($count >= 1) {
  4999. add_ok $msg;
  5000. }
  5001. elsif ($critical) {
  5002. add_critical $msg;
  5003. }
  5004. else {
  5005. add_warning $msg;
  5006. }
  5007. }
  5008. return;
  5009.  
  5010. } ## end of check_listener
  5011.  
  5012.  
  5013. sub check_locks {
  5014.  
  5015. ## Check the number of locks
  5016. ## Supports: Nagios, MRTG
  5017. ## By default, checks all databases
  5018. ## Can check specific databases with include
  5019. ## Can ignore databases with exclude
  5020. ## Warning and critical are either simple numbers, or more complex:
  5021. ## Use locktype=number:locktype2=number
  5022. ## The locktype can be "total", "waiting", or the name of a lock
  5023. ## Lock names are case-insensitive, and do not need the "lock" at the end.
  5024. ## Example: --warning=100 --critical="total=200;exclusive=20;waiting=5"
  5025.  
  5026. my ($warning, $critical) = validate_range
  5027. ({
  5028. type => 'multival',
  5029. default_warning => 100,
  5030. default_critical => 150,
  5031. });
  5032.  
  5033. $SQL = q{SELECT granted, mode, datname FROM pg_locks l RIGHT JOIN pg_database d ON (d.oid=l.database) WHERE d.datallowconn};
  5034. my $info = run_command($SQL, { regex => qr[\s*\w+\s*\|\s*] });
  5035.  
  5036. # Locks are counted globally not by db.
  5037. # We output for each db, following the specific warning and critical :
  5038. # time=00.1 foodb.exclusive=2;;3 foodb.total=10;;30 postgres.exclusive=0;;3 postgres.total=1;;3
  5039. for $db (@{$info->{db}}) {
  5040. my $gotone = 0;
  5041. my %dblock;
  5042. my %totallock = (total => 0);
  5043. ROW: for my $r (@{$db->{slurp}}) {
  5044. my ($granted,$mode,$dbname) = ($r->{granted}, lc $r->{mode}, $r->{datname});
  5045.  
  5046. ## May be forcibly skipping this database via arguments
  5047. next ROW if skip_item($dbname);
  5048.  
  5049. ## If we hit the right join, simply make an empty total entry
  5050. if (! length $granted) {
  5051. $dblock{$dbname}{total} ||= 0;
  5052. }
  5053. else {
  5054. $dblock{$dbname}{total}++;
  5055. $gotone = 1;
  5056. $mode =~ s{lock$}{};
  5057. $dblock{$dbname}{$mode}++;
  5058. $dblock{$dbname}{waiting}++ if $granted ne 't';
  5059. }
  5060. }
  5061.  
  5062. # Compute total, add hash key for critical and warning specific check
  5063. for my $k (keys %dblock) {
  5064.  
  5065. if ($warning) {
  5066. for my $l (keys %{$warning}) {
  5067. $dblock{$k}{$l} = 0 if ! exists $dblock{$k}{$l};
  5068. }
  5069. }
  5070. if ($critical) {
  5071. for my $l (keys %{$critical}) {
  5072. #$dblock{$k}{$l} = 0 if ! exists $dblock{$k}{$l};
  5073. }
  5074. }
  5075. for my $m (keys %{$dblock{$k}}){
  5076. $totallock{$m} += $dblock{$k}{$m};
  5077. }
  5078. }
  5079.  
  5080. if ($MRTG) {
  5081. do_mrtg( {one => $totallock{total}, msg => "DB: $db->{dbname}" } );
  5082. }
  5083.  
  5084. # Nagios perfdata output
  5085. for my $dbname (sort keys %dblock) {
  5086. for my $type (sort keys %{ $dblock{$dbname} }) {
  5087. next if ((! $critical or ! exists $critical->{$type})
  5088. and (!$warning or ! exists $warning->{$type}));
  5089. $db->{perf} .= sprintf ' %s=%s;',
  5090. perfname("$dbname.$type"), $dblock{$dbname}{$type};
  5091. if ($warning and exists $warning->{$type}) {
  5092. $db->{perf} .= $warning->{$type};
  5093. }
  5094. if ($critical and $critical->{$type}) {
  5095. $db->{perf} .= ";$critical->{$type}";
  5096. }
  5097. }
  5098. }
  5099.  
  5100. if (!$gotone) {
  5101. add_unknown msg('no-match-db');
  5102. next;
  5103. }
  5104.  
  5105. ## If not specific errors, just use the total
  5106. my $ok = 1;
  5107. for my $type (keys %totallock) {
  5108. if ($critical and exists $critical->{$type} and $totallock{$type} >= $critical->{$type}) {
  5109. ($type eq 'total')
  5110. ? add_critical msg('locks-msg2', $totallock{total})
  5111. : add_critical msg('locks-msg', $type, $totallock{$type});
  5112. $ok = 0;
  5113. }
  5114. if ($warning and exists $warning->{$type} and $totallock{$type} >= $warning->{$type}) {
  5115. ($type eq 'total')
  5116. ? add_warning msg('locks-msg2', $totallock{total})
  5117. : add_warning msg('locks-msg', $type, $totallock{$type});
  5118. $ok = 0;
  5119. }
  5120. }
  5121. if ($ok) {
  5122. my %show;
  5123. if (!keys %critical and !keys %warning) {
  5124. $show{total} = 1;
  5125. }
  5126. for my $type (keys %critical) {
  5127. $show{$type} = 1;
  5128. }
  5129. for my $type (keys %warning) {
  5130. $show{$type} = 1;
  5131. }
  5132. my $msg = '';
  5133. for (sort keys %show) {
  5134. $msg .= sprintf "$_=%d ", $totallock{$_} || 0;
  5135. }
  5136. add_ok $msg;
  5137. }
  5138. }
  5139.  
  5140. return;
  5141.  
  5142. } ## end of check_locks
  5143.  
  5144.  
  5145. sub check_logfile {
  5146.  
  5147. ## Make sure the logfile is getting written to
  5148. ## Supports: Nagios, MRTG
  5149. ## Especially useful for syslog redirectors
  5150. ## Should be run on the system housing the logs
  5151. ## Optional argument "logfile" tells where the logfile is
  5152. ## Allows for some conversion characters.
  5153. ## Example: --logfile="/syslog/%Y-m%-d%/H%/postgres.log"
  5154. ## Critical and warning are not used: it's either ok or critical.
  5155.  
  5156. my $critwarn = $opt{warning} ? 0 : 1;
  5157.  
  5158. $SQL = q{
  5159. SELECT name, CASE WHEN length(setting)<1 THEN '?' ELSE setting END AS s
  5160. FROM pg_settings
  5161. WHERE name IN ('log_destination','log_directory','log_filename','redirect_stderr','syslog_facility')
  5162. ORDER BY name
  5163. };
  5164.  
  5165. my $logfilere = qr{^[\w_\s\/%\-\.]+$};
  5166. if (exists $opt{logfile} and $opt{logfile} !~ $logfilere) {
  5167. ndie msg('logfile-opt-bad');
  5168. }
  5169.  
  5170. my $info = run_command($SQL);
  5171. $VERBOSE >= 3 and warn Dumper $info;
  5172.  
  5173. for $db (@{$info->{db}}) {
  5174. my $i;
  5175. for my $r (@{$db->{slurp}}) {
  5176. $i->{$r->{name}} = $r->{s} || '?';
  5177. }
  5178. for my $word (qw{ log_destination log_directory log_filename redirect_stderr syslog_facility }) {
  5179. $i->{$word} = '?' if ! exists $i->{$word};
  5180. }
  5181.  
  5182. ## Figure out what we think the log file will be
  5183. my $logfile ='';
  5184. if (exists $opt{logfile} and $opt{logfile} =~ /\w/) {
  5185. $logfile = $opt{logfile};
  5186. }
  5187. else {
  5188. if ($i->{log_destination} eq 'syslog') {
  5189. ## We'll make a best effort to figure out where it is. Using the --logfile option is preferred.
  5190. $logfile = '/var/log/messages';
  5191. if (open my $cfh, '<', '/etc/syslog.conf') {
  5192. while (<$cfh>) {
  5193. if (/\b$i->{syslog_facility}\.(?!none).+?([\w\/]+)$/i) {
  5194. $logfile = $1;
  5195. }
  5196. }
  5197. }
  5198. if (!$logfile or ! -e $logfile) {
  5199. ndie msg('logfile-syslog', $i->{syslog_facility});
  5200. }
  5201. }
  5202. elsif ($i->{log_destination} eq 'stderr') {
  5203. if ($i->{redirect_stderr} ne 'yes') {
  5204. ndie msg('logfile-stderr');
  5205. }
  5206. }
  5207. }
  5208.  
  5209. ## We now have a logfile (or a template)..parse it into pieces.
  5210. ## We need at least hour, day, month, year
  5211. my @t = localtime;
  5212. my ($H,$d,$m,$Y) = (sprintf ('%02d',$t[2]),sprintf('%02d',$t[3]),sprintf('%02d',$t[4]+1),$t[5]+1900);
  5213. my $y = substr($Y,2,4);
  5214. if ($logfile !~ $logfilere) {
  5215. ndie msg('logfile-bad',$logfile);
  5216. }
  5217. $logfile =~ s/%%/~~/g;
  5218. $logfile =~ s/%Y/$Y/g;
  5219. $logfile =~ s/%y/$y/g;
  5220. $logfile =~ s/%m/$m/g;
  5221. $logfile =~ s/%d/$d/g;
  5222. $logfile =~ s/%H/$H/g;
  5223.  
  5224. $VERBOSE >= 3 and warn msg('logfile-debug', $logfile);
  5225.  
  5226. if (! -e $logfile) {
  5227. my $msg = msg('logfile-dne', $logfile);
  5228. $MRTG and ndie $msg;
  5229. if ($critwarn) {
  5230. add_unknown $msg;
  5231. }
  5232. else {
  5233. add_warning $msg;
  5234. }
  5235. next;
  5236. }
  5237. my $logfh;
  5238. unless (open $logfh, '<', $logfile) {
  5239. add_unknown msg('logfile-openfail', $logfile, $!);
  5240. next;
  5241. }
  5242. seek($logfh, 0, 2) or ndie msg('logfile-seekfail', $logfile, $!);
  5243.  
  5244. ## Throw a custom error string.
  5245. ## We do the number first as old versions only show part of the string.
  5246. my $random_number = int rand(999999999999);
  5247. my $funky = sprintf "check_postgres_logfile_error_$random_number $ME DB=$db->{dbname} PID=$$ Time=%s",
  5248. scalar localtime;
  5249.  
  5250. ## Cause an error on just this target
  5251. delete @{$db}{qw(ok slurp totaltime)};
  5252. my $badinfo = run_command("$funky", {failok => 1, target => $db} );
  5253.  
  5254. my $MAXSLEEPTIME = $opt{timeout} || 20;
  5255. my $SLEEP = 1;
  5256. my $found = 0;
  5257. LOGWAIT: {
  5258. sleep $SLEEP;
  5259. seek $logfh, 0, 1 or ndie msg('logfile-seekfail', $logfile, $!);
  5260. while (<$logfh>) {
  5261. if (/logfile_error_$random_number/) { ## Some logs break things up, so we don't use funky
  5262. $found = 1;
  5263. last LOGWAIT;
  5264. }
  5265. }
  5266. $MAXSLEEPTIME -= $SLEEP;
  5267. redo if $MAXSLEEPTIME > 0;
  5268. my $msg = msg('logfile-fail', $logfile);
  5269. $MRTG and do_mrtg({one => 0, msg => $msg});
  5270. if ($critwarn) {
  5271. add_critical $msg;
  5272. }
  5273. else {
  5274. add_warning $msg;
  5275. }
  5276. }
  5277. close $logfh or ndie msg('file-noclose', $logfile, $!);
  5278.  
  5279. if ($found == 1) {
  5280. $MRTG and do_mrtg({one => 1});
  5281. add_ok msg('logfile-ok', $logfile);
  5282. }
  5283. }
  5284. return;
  5285.  
  5286. } ## end of check_logfile
  5287.  
  5288.  
  5289. sub find_new_version {
  5290.  
  5291. ## Check for newer versions of some program
  5292.  
  5293. my $program = shift or die;
  5294. my $exec = shift or die;
  5295. my $url = shift or die;
  5296.  
  5297. ## The format is X.Y.Z [optional message]
  5298. my $versionre = qr{((\d+)\.(\d+)\.(\d+))\s*(.*)};
  5299. my ($cversion,$cmajor,$cminor,$crevision,$cmessage) = ('','','','','');
  5300. my $found = 0;
  5301.  
  5302. ## Try to fetch the current version from the web
  5303. for my $meth (@get_methods) {
  5304. eval {
  5305. my $COM = "$meth $url";
  5306. $VERBOSE >= 1 and warn "TRYING: $COM\n";
  5307. my $info = qx{$COM 2>/dev/null};
  5308. ## Postgres is slightly different
  5309. if ($program eq 'Postgres') {
  5310. $cmajor = {};
  5311. while ($info =~ /<title>(\d+)\.(\d+)\.(\d+)/g) {
  5312. $found = 1;
  5313. $cmajor->{"$1.$2"} = $3;
  5314. }
  5315. }
  5316. elsif ($info =~ $versionre) {
  5317. $found = 1;
  5318. ($cversion,$cmajor,$cminor,$crevision,$cmessage) = ($1, int $2, int $3, int $4, $5);
  5319. if ($VERBOSE >= 1) {
  5320. $info =~ s/\s+$//s;
  5321. warn "Remote version string: $info\n";
  5322. warn "Remote version: $cversion\n";
  5323. }
  5324. }
  5325. };
  5326. last if $found;
  5327. }
  5328.  
  5329. if (! $found) {
  5330. add_unknown msg('new-ver-nocver', $program);
  5331. return;
  5332. }
  5333.  
  5334. ## Figure out the local copy's version
  5335. my $output;
  5336. eval {
  5337. ## We may already know the version (e.g. ourselves)
  5338. $output = ($exec =~ /\d+\.\d+/) ? $exec : qx{$exec --version 2>&1};
  5339. };
  5340. if ($@ or !$output) {
  5341. if ($program eq 'tail_n_mail') {
  5342. ## Check for the old name
  5343. eval {
  5344. $output = qx{tail_n_mail.pl --version 2>&1};
  5345. };
  5346. }
  5347. if ($@ or !$output) {
  5348. add_unknown msg('new-ver-badver', $program);
  5349. return;
  5350. }
  5351. }
  5352.  
  5353. if ($output !~ $versionre) {
  5354. add_unknown msg('new-ver-nolver', $program);
  5355. return;
  5356. }
  5357. my ($lversion,$lmajor,$lminor,$lrevision) = ($1, int $2, int $3, int $4);
  5358. if ($VERBOSE >= 1) {
  5359. $output =~ s/\s+$//s;
  5360. warn "Local version string: $output\n";
  5361. warn "Local version: $lversion\n";
  5362. }
  5363.  
  5364. ## Postgres is a special case
  5365. if ($program eq 'Postgres') {
  5366. my $lver = "$lmajor.$lminor";
  5367. if (! exists $cmajor->{$lver}) {
  5368. add_unknown msg('new-ver-nocver', $program);
  5369. return;
  5370. }
  5371. $crevision = $cmajor->{$lver};
  5372. $cmajor = $lmajor;
  5373. $cminor = $lminor;
  5374. $cversion = "$cmajor.$cminor.$crevision";
  5375. }
  5376.  
  5377. ## Most common case: everything matches
  5378. if ($lversion eq $cversion) {
  5379. add_ok msg('new-ver-ok', $lversion, $program);
  5380. return;
  5381. }
  5382.  
  5383. ## Check for a revision update
  5384. if ($lmajor==$cmajor and $lminor==$cminor and $lrevision<$crevision) {
  5385. add_critical msg('new-ver-warn', $cversion, $program, $lversion);
  5386. return;
  5387. }
  5388.  
  5389. ## Check for a major update
  5390. if ($lmajor<$cmajor or ($lmajor==$cmajor and $lminor<$cminor)) {
  5391. add_warning msg('new-ver-warn', $cversion, $program, $lversion);
  5392. return;
  5393. }
  5394.  
  5395. ## Anything else must be time travel, which we cannot handle
  5396. add_unknown msg('new-ver-tt', $program, $lversion, $cversion);
  5397. return;
  5398.  
  5399. } ## end of find_new_version
  5400.  
  5401.  
  5402. sub check_new_version_bc {
  5403.  
  5404. ## Check if a newer version of Bucardo is available
  5405.  
  5406. my $url = 'http://bucardo.org/bucardo/latest_version.txt';
  5407. find_new_version('Bucardo', 'bucardo_ctl', $url);
  5408.  
  5409. return;
  5410.  
  5411. } ## end of check_new_version_bc
  5412.  
  5413.  
  5414. sub check_new_version_box {
  5415.  
  5416. ## Check if a newer version of boxinfo is available
  5417.  
  5418. my $url = 'http://bucardo.org/boxinfo/latest_version.txt';
  5419. find_new_version('boxinfo', 'boxinfo.pl', $url);
  5420.  
  5421. return;
  5422.  
  5423. } ## end of check_new_version_box
  5424.  
  5425.  
  5426. sub check_new_version_cp {
  5427.  
  5428. ## Check if a new version of check_postgres.pl is available
  5429.  
  5430. my $url = 'http://bucardo.org/check_postgres/latest_version.txt';
  5431. find_new_version('check_postgres', $VERSION, $url);
  5432.  
  5433. return;
  5434.  
  5435. } ## end of check_new_version_cp
  5436.  
  5437.  
  5438. sub check_new_version_pg {
  5439.  
  5440. ## Check if a new version of Postgres is available
  5441.  
  5442. my $url = 'http://www.postgresql.org/versions.rss';
  5443.  
  5444. ## Grab the local version
  5445. my $info = run_command('SELECT version() AS version');
  5446. my $lversion = $info->{db}[0]{slurp}[0]{version};
  5447. ## Make sure it is parseable and check for development versions
  5448. if ($lversion !~ /\d+\.\d+\.\d+/) {
  5449. if ($lversion =~ /(\d+\.\d+\S+)/) {
  5450. add_ok msg('new-ver-dev', 'Postgres', $1);
  5451. return;
  5452. }
  5453. add_unknown msg('new-ver-nolver', 'Postgres');
  5454. return;
  5455. }
  5456.  
  5457. find_new_version('Postgres', $lversion, $url);
  5458.  
  5459. return;
  5460.  
  5461. } ## end of check_new_version_pg
  5462.  
  5463.  
  5464. sub check_new_version_tnm {
  5465.  
  5466. ## Check if a new version of tail_n_mail is available
  5467.  
  5468. my $url = 'http://bucardo.org/tail_n_mail/latest_version.txt';
  5469. find_new_version('tail_n_mail', 'tail_n_mail', $url);
  5470.  
  5471. return;
  5472.  
  5473. } ## end of check_new_version_tnm
  5474.  
  5475.  
  5476. sub check_pgagent_jobs {
  5477. ## Check for failed pgAgent jobs.
  5478. ## Supports: Nagios
  5479. ## Critical and warning are intervals.
  5480. ## Example: --critical="1 hour"
  5481. ## Example: --warning="2 hours"
  5482.  
  5483. my ($warning, $critical) = validate_range({ type => 'time', any_warning => 1 });
  5484.  
  5485. # Determine critcal warning column contents.
  5486. my $is_crit = $critical && $warning
  5487. ? "GREATEST($critical - EXTRACT('epoch' FROM NOW() - (jlog.jlgstart + jlog.jlgduration)), 0)"
  5488. : $critical ? 1 : 0;
  5489.  
  5490. # Determine max time to examine.
  5491. my $seconds = do {
  5492. no warnings;
  5493. $warning > $critical ? $warning : $critical;
  5494. };
  5495.  
  5496. $SQL = qq{
  5497. SELECT jlog.jlgid
  5498. , job.jobname
  5499. , step.jstname
  5500. , slog.jslresult
  5501. , slog.jsloutput
  5502. , $is_crit AS critical
  5503. FROM pgagent.pga_job job
  5504. JOIN pgagent.pga_joblog jlog ON job.jobid = jlog.jlgjobid
  5505. JOIN pgagent.pga_jobstep step ON job.jobid = step.jstjobid
  5506. JOIN pgagent.pga_jobsteplog slog ON jlog.jlgid = slog.jsljlgid AND step.jstid = slog.jsljstid
  5507. WHERE slog.jslresult <> 0
  5508. AND EXTRACT('epoch' FROM NOW() - (jlog.jlgstart + jlog.jlgduration)) < $seconds
  5509. };
  5510.  
  5511. my $info = run_command($SQL);
  5512.  
  5513. for $db (@{$info->{db}}) {
  5514. my @rows = @{ $db->{slurp} } or do {
  5515. add_ok msg('pgagent-jobs-ok');
  5516. next;
  5517. };
  5518.  
  5519. if ($rows[0]{critical} !~ /^(?:[01]|\d+[.]\d+)$/) {
  5520. add_unknown msg('invalid-query', $db->{slurp});
  5521. next;
  5522. }
  5523.  
  5524. my ($is_crit, @msg);
  5525. my $log_id = -1;
  5526. for my $step (@rows) {
  5527. my $output = $step->{jsloutput} || '(NO OUTPUT)';
  5528. push @msg => "$step->{jslresult} $step->{jobname}/$step->{jstname}: $output";
  5529. $is_crit ||= $step->{critical};
  5530. }
  5531.  
  5532. (my $msg = join '; ' => @msg) =~ s{\r?\n}{ }g;
  5533. if ($is_crit) {
  5534. add_critical $msg;
  5535. } else {
  5536. add_warning $msg;
  5537. }
  5538. }
  5539.  
  5540. return;
  5541. }
  5542.  
  5543. sub check_pgbouncer_checksum {
  5544.  
  5545. ## Verify the checksum of all pgbouncer settings
  5546. ## Supports: Nagios, MRTG
  5547. ## Not that the connection will be done on the pgbouncer database
  5548. ## One of warning or critical must be given (but not both)
  5549. ## It should run one time to find out the expected checksum
  5550. ## You can use --critical="0" to find out the checksum
  5551. ## You can include or exclude settings as well
  5552. ## Example:
  5553. ## check_postgres_pgbouncer_checksum --critical="4e7ba68eb88915d3d1a36b2009da4acd"
  5554.  
  5555. my ($warning, $critical) = validate_range({type => 'checksum', onlyone => 1});
  5556.  
  5557. eval {
  5558. require Digest::MD5;
  5559. };
  5560. if ($@) {
  5561. ndie msg('checksum-nomd');
  5562. }
  5563.  
  5564. $SQL = 'SHOW CONFIG';
  5565. my $info = run_command($SQL, { regex => qr[log_pooler_errors] });
  5566.  
  5567. $db = $info->{db}[0];
  5568.  
  5569. my $newstring = '';
  5570. for my $r (@{$db->{slurp}}) {
  5571. my $key = $r->{key};
  5572. next if skip_item($key);
  5573. $newstring .= "$r->{key} = $r->{value}\n";
  5574. }
  5575.  
  5576. if (! length $newstring) {
  5577. add_unknown msg('no-match-set');
  5578. }
  5579.  
  5580. my $checksum = Digest::MD5::md5_hex($newstring);
  5581.  
  5582. my $msg = msg('checksum-msg', $checksum);
  5583. if ($MRTG) {
  5584. $opt{mrtg} or ndie msg('checksum-nomrtg');
  5585. do_mrtg({one => $opt{mrtg} eq $checksum ? 1 : 0, msg => $checksum});
  5586. }
  5587. if ($critical and $critical ne $checksum) {
  5588. add_critical $msg;
  5589. }
  5590. elsif ($warning and $warning ne $checksum) {
  5591. add_warning $msg;
  5592. }
  5593. elsif (!$critical and !$warning) {
  5594. add_unknown $msg;
  5595. }
  5596. else {
  5597. add_ok $msg;
  5598. }
  5599.  
  5600. return;
  5601.  
  5602. } ## end of check_pgbouncer_checksum
  5603.  
  5604. sub check_pgbouncer_backends {
  5605.  
  5606. ## Check the number of connections to pgbouncer compared to
  5607. ## max_client_conn
  5608. ## Supports: Nagios, MRTG
  5609. ## It makes no sense to run this more than once on the same cluster
  5610. ## Need to be superuser, else only your queries will be visible
  5611. ## Warning and criticals can take three forms:
  5612. ## critical = 12 -- complain if there are 12 or more connections
  5613. ## critical = 95% -- complain if >= 95% of available connections are used
  5614. ## critical = -5 -- complain if there are only 5 or fewer connection slots left
  5615. ## The former two options only work with simple numbers - no percentage or negative
  5616. ## Can also ignore databases with exclude, and limit with include
  5617.  
  5618. my $warning = $opt{warning} || '90%';
  5619. my $critical = $opt{critical} || '95%';
  5620. my $noidle = $opt{noidle} || 0;
  5621.  
  5622. ## If only critical was used, remove the default warning
  5623. if ($opt{critical} and !$opt{warning}) {
  5624. $warning = $critical;
  5625. }
  5626.  
  5627. my $validre = qr{^(\-?)(\d+)(\%?)$};
  5628. if ($critical !~ $validre) {
  5629. ndie msg('pgb-backends-users', 'Critical');
  5630. }
  5631. my ($e1,$e2,$e3) = ($1,$2,$3);
  5632. if ($warning !~ $validre) {
  5633. ndie msg('pgb-backends-users', 'Warning');
  5634. }
  5635. my ($w1,$w2,$w3) = ($1,$2,$3);
  5636.  
  5637. ## If number is greater, all else is same, and not minus
  5638. if ($w2 > $e2 and $w1 eq $e1 and $w3 eq $e3 and $w1 eq '') {
  5639. ndie msg('range-warnbig');
  5640. }
  5641. ## If number is less, all else is same, and minus
  5642. if ($w2 < $e2 and $w1 eq $e1 and $w3 eq $e3 and $w1 eq '-') {
  5643. ndie msg('range-warnsmall');
  5644. }
  5645. if (($w1 and $w3) or ($e1 and $e3)) {
  5646. ndie msg('range-neg-percent');
  5647. }
  5648.  
  5649. ## Grab information from the config
  5650. $SQL = 'SHOW CONFIG';
  5651.  
  5652. my $info = run_command($SQL, { regex => qr{\d+}, emptyok => 1 } );
  5653.  
  5654. ## Default values for information gathered
  5655. my $limit = 0;
  5656.  
  5657. ## Determine max_client_conn
  5658. for my $r (@{$info->{db}[0]{slurp}}) {
  5659. if ($r->{key} eq 'max_client_conn') {
  5660. $limit = $r->{value};
  5661. last;
  5662. }
  5663. }
  5664.  
  5665. ## Grab information from pools
  5666. $SQL = 'SHOW POOLS';
  5667.  
  5668. $info = run_command($SQL, { regex => qr{\d+}, emptyok => 1 } );
  5669.  
  5670. $db = $info->{db}[0];
  5671.  
  5672. my $total = 0;
  5673. my $grandtotal = @{$db->{slurp}};
  5674.  
  5675. for my $r (@{$db->{slurp}}) {
  5676.  
  5677. ## Always want perf to show all
  5678. my $nwarn=$w2;
  5679. my $ncrit=$e2;
  5680. if ($e1) {
  5681. $ncrit = $limit-$e2;
  5682. }
  5683. elsif ($e3) {
  5684. $ncrit = (int $e2*$limit/100);
  5685. }
  5686. if ($w1) {
  5687. $nwarn = $limit-$w2;
  5688. }
  5689. elsif ($w3) {
  5690. $nwarn = (int $w2*$limit/100)
  5691. }
  5692.  
  5693. if (! skip_item($r->{database})) {
  5694. my $current = $r->{cl_active} + $r->{cl_waiting};
  5695. $db->{perf} .= " '$r->{database}'=$current;$nwarn;$ncrit;0;$limit";
  5696. $total += $current;
  5697. }
  5698. }
  5699.  
  5700. if ($MRTG) {
  5701. $stats{$db->{dbname}} = $total;
  5702. $statsmsg{$db->{dbname}} = msg('pgb-backends-mrtg', $db->{dbname}, $limit);
  5703. return;
  5704. }
  5705.  
  5706. if (!$total) {
  5707. if ($grandtotal) {
  5708. ## We assume that exclude/include rules are correct, and we simply had no entries
  5709. ## at all in the specific databases we wanted
  5710. add_ok msg('pgb-backends-none');
  5711. }
  5712. else {
  5713. add_unknown msg('no-match-db');
  5714. }
  5715. return;
  5716. }
  5717.  
  5718. my $percent = (int $total / $limit*100) || 1;
  5719. my $msg = msg('pgb-backends-msg', $total, $limit, $percent);
  5720. my $ok = 1;
  5721.  
  5722. if ($e1) { ## minus
  5723. $ok = 0 if $limit-$total <= $e2;
  5724. }
  5725. elsif ($e3) { ## percent
  5726. my $nowpercent = $total/$limit*100;
  5727. $ok = 0 if $nowpercent >= $e2;
  5728. }
  5729. else { ## raw number
  5730. $ok = 0 if $total >= $e2;
  5731. }
  5732. if (!$ok) {
  5733. add_critical $msg;
  5734. return;
  5735. }
  5736.  
  5737. if ($w1) {
  5738. $ok = 0 if $limit-$total <= $w2;
  5739. }
  5740. elsif ($w3) {
  5741. my $nowpercent = $total/$limit*100;
  5742. $ok = 0 if $nowpercent >= $w2;
  5743. }
  5744. else {
  5745. $ok = 0 if $total >= $w2;
  5746. }
  5747. if (!$ok) {
  5748. add_warning $msg;
  5749. return;
  5750. }
  5751.  
  5752. add_ok $msg;
  5753.  
  5754. return;
  5755.  
  5756. } ## end of check_pgbouncer_backends
  5757.  
  5758.  
  5759.  
  5760. sub check_pgb_pool {
  5761.  
  5762. # Check various bits of the pgbouncer SHOW POOLS ouptut
  5763. my $stat = shift;
  5764. my ($warning, $critical) = validate_range({type => 'positive integer'});
  5765.  
  5766. $SQL = 'SHOW POOLS';
  5767. my $info = run_command($SQL, { regex => qr[$stat] });
  5768.  
  5769. $db = $info->{db}[0];
  5770. my $output = $db->{slurp};
  5771. my $gotone = 0;
  5772. for my $i (@$output) {
  5773. next if skip_item($i->{database});
  5774. my $msg = "$i->{database}=$i->{$stat}";
  5775.  
  5776. if ($MRTG) {
  5777. $stats{$i->{database}} = $i->{$stat};
  5778. $statsmsg{$i->{database}} = msg('pgbouncer-pool', $i->{database}, $stat, $i->{$stat});
  5779. next;
  5780. }
  5781.  
  5782. if ($critical and $i->{$stat} >= $critical) {
  5783. add_critical $msg;
  5784. }
  5785. elsif ($warning and $i->{$stat} >= $warning) {
  5786. add_warning $msg;
  5787. }
  5788. else {
  5789. add_ok $msg;
  5790. }
  5791. }
  5792.  
  5793. return;
  5794.  
  5795. } ## end of check_pgb_pool
  5796.  
  5797.  
  5798. sub check_prepared_txns {
  5799.  
  5800. ## Checks age of prepared transactions
  5801. ## Most installations probably want no prepared_transactions
  5802. ## Supports: Nagios, MRTG
  5803.  
  5804. my ($warning, $critical) = validate_range
  5805. ({
  5806. type => 'seconds',
  5807. default_warning => '1',
  5808. default_critical => '30',
  5809. });
  5810.  
  5811. my $SQL = q{
  5812. SELECT database, ROUND(EXTRACT(epoch FROM now()-prepared)) AS age, prepared
  5813. FROM pg_prepared_xacts
  5814. ORDER BY prepared ASC
  5815. };
  5816.  
  5817. my $info = run_command($SQL, {regex => qr[\w+], emptyok => 1 } );
  5818.  
  5819. my $msg = msg('preptxn-none');
  5820. my $found = 0;
  5821. for $db (@{$info->{db}}) {
  5822. my (@crit,@warn,@ok);
  5823. my ($maxage,$maxdb) = (0,''); ## used by MRTG only
  5824. ROW: for my $r (@{$db->{slurp}}) {
  5825. my ($dbname,$age,$date) = ($r->{database},$r->{age},$r->{prepared});
  5826. $found = 1 if ! $found;
  5827. next ROW if skip_item($dbname);
  5828. $found = 2;
  5829. if ($MRTG) {
  5830. if ($age > $maxage) {
  5831. $maxdb = $dbname;
  5832. $maxage = $age;
  5833. }
  5834. elsif ($age == $maxage) {
  5835. $maxdb .= sprintf "%s$dbname", length $maxdb ? ' | ' : '';
  5836. }
  5837. next;
  5838. }
  5839.  
  5840. $msg = "$dbname=$date ($age)";
  5841. $db->{perf} .= sprintf ' %s=%ss;%s;%s',
  5842. perfname($dbname), $age, $warning, $critical;
  5843. if (length $critical and $age >= $critical) {
  5844. push @crit => $msg;
  5845. }
  5846. elsif (length $warning and $age >= $warning) {
  5847. push @warn => $msg;
  5848. }
  5849. else {
  5850. push @ok => $msg;
  5851. }
  5852. }
  5853. if ($MRTG) {
  5854. do_mrtg({one => $maxage, msg => $maxdb});
  5855. }
  5856. elsif (0 == $found) {
  5857. add_ok msg('preptxn-none');
  5858. }
  5859. elsif (1 == $found) {
  5860. add_unknown msg('no-match-db');
  5861. }
  5862. elsif (@crit) {
  5863. add_critical join ' ' => @crit;
  5864. }
  5865. elsif (@warn) {
  5866. add_warning join ' ' => @warn;
  5867. }
  5868. else {
  5869. add_ok join ' ' => @ok;
  5870. }
  5871. }
  5872.  
  5873. return;
  5874.  
  5875. } ## end of check_prepared_txns
  5876.  
  5877.  
  5878. sub check_query_runtime {
  5879.  
  5880. ## Make sure a known query runs at least as fast as we think it should
  5881. ## Supports: Nagios, MRTG
  5882. ## Warning and critical are time limits, defaulting to seconds
  5883. ## Valid units: s[econd], m[inute], h[our], d[ay]
  5884. ## Does a "EXPLAIN ANALYZE SELECT COUNT(1) FROM xyz"
  5885. ## where xyz is given by the option --queryname
  5886. ## This could also be a table or a function, or course, but must be a
  5887. ## single word. If a function, it must be empty (with "()")
  5888. ## Examples:
  5889. ## --warning="100s" --critical="120s" --queryname="speedtest1"
  5890. ## --warning="5min" --critical="15min" --queryname="speedtest()"
  5891.  
  5892. my ($warning, $critical) = validate_range({type => 'time'});
  5893.  
  5894. my $queryname = $opt{queryname} || '';
  5895.  
  5896. if ($queryname !~ /^[\w\_\.]+(?:\(\))?$/) {
  5897. ndie msg('runtime-badname');
  5898. }
  5899.  
  5900. $SQL = "EXPLAIN ANALYZE SELECT COUNT(1) FROM $queryname";
  5901. my $info = run_command($SQL);
  5902.  
  5903. for $db (@{$info->{db}}) {
  5904. if (! exists $db->{slurp}[0]{queryplan}) {
  5905. add_unknown msg('invalid-query', $db->{slurp});
  5906. next;
  5907. }
  5908. my $totalms = -1;
  5909. for my $r (@{$db->{slurp}}) {
  5910. if ($r->{queryplan} =~ / (\d+\.\d+) ms/) {
  5911. $totalms = $1;
  5912. }
  5913. }
  5914. my $totalseconds = sprintf '%.2f', $totalms / 1000.0;
  5915. if ($MRTG) {
  5916. $stats{$db->{dbname}} = $totalseconds;
  5917. next;
  5918. }
  5919. $db->{perf} = sprintf '%s=%ss;%s;%s',
  5920. perfname(msg('query-time')), $totalseconds, $warning, $critical;
  5921. my $msg = msg('runtime-msg', $totalseconds);
  5922. if (length $critical and $totalseconds >= $critical) {
  5923. add_critical $msg;
  5924. }
  5925. elsif (length $warning and $totalseconds >= $warning) {
  5926. add_warning $msg;
  5927. }
  5928. else {
  5929. add_ok $msg;
  5930. }
  5931. }
  5932.  
  5933. $MRTG and do_mrtg_stats(msg('runtime-badmrtg'));
  5934.  
  5935. return;
  5936.  
  5937. } ## end of check_query_runtime
  5938.  
  5939.  
  5940. sub check_query_time {
  5941.  
  5942. ## Check the length of running queries
  5943.  
  5944. check_txn_idle('qtime',
  5945. msg('queries'),
  5946. msg('query-time'),
  5947. 'query_start',
  5948. q{query_start IS NOT NULL AND current_query NOT LIKE '<IDLE>%'});
  5949.  
  5950. return;
  5951.  
  5952. } ## end of check_query_time
  5953.  
  5954.  
  5955. sub check_relation_size {
  5956.  
  5957. my $relkind = shift || 'relation';
  5958.  
  5959. ## Check the size of one or more relations
  5960. ## Supports: Nagios, MRTG
  5961. ## By default, checks all relations
  5962. ## Can check specific one(s) with include
  5963. ## Can ignore some with exclude
  5964. ## Warning and critical are bytes
  5965. ## Valid units: b, k, m, g, t, e
  5966. ## All above may be written as plural or with a trailing 'g'
  5967. ## Limit to a specific user (relation owner) with the includeuser option
  5968. ## Exclude users with the excludeuser option
  5969.  
  5970. my ($warning, $critical) = validate_range({type => 'size'});
  5971.  
  5972. $SQL = sprintf q{
  5973. SELECT pg_relation_size(c.oid) AS rsize,
  5974. pg_size_pretty(pg_relation_size(c.oid)) AS psize,
  5975. relkind, relname, nspname
  5976. FROM pg_class c, pg_namespace n WHERE (relkind = %s) AND n.oid = c.relnamespace
  5977. },
  5978. $relkind eq 'table' ? q{'r'}
  5979. : $relkind eq 'index' ? q{'i'}
  5980. : q{'r' OR relkind = 'i'};
  5981.  
  5982. if ($opt{perflimit}) {
  5983. $SQL .= " ORDER BY 1 DESC LIMIT $opt{perflimit}";
  5984. }
  5985.  
  5986. if ($USERWHERECLAUSE) {
  5987. $SQL =~ s/ WHERE/, pg_user u WHERE u.usesysid=c.relowner$USERWHERECLAUSE AND/;
  5988. }
  5989.  
  5990. my $info = run_command($SQL, {emptyok => 1});
  5991.  
  5992. my $found = 0;
  5993. for $db (@{$info->{db}}) {
  5994.  
  5995. $found = 1;
  5996. if ($db->{slurp}[0]{rsize} !~ /\d/ and $USERWHERECLAUSE) {
  5997. $stats{$db->{dbname}} = 0;
  5998. add_ok msg('no-match-user');
  5999. next;
  6000. }
  6001.  
  6002. my ($max,$pmax,$kmax,$nmax,$smax) = (-1,0,0,'?','?');
  6003.  
  6004. ROW: for my $r (@{$db->{slurp}}) {
  6005. my ($size,$psize,$kind,$name,$schema) = @$r{qw/ rsize psize relkind relname nspname/};
  6006.  
  6007. next ROW if skip_item($name, $schema);
  6008.  
  6009. my $nicename = $kind eq 'r' ? "$schema.$name" : $name;
  6010.  
  6011. $db->{perf} .= sprintf '%s%s=%sB;%s;%s',
  6012. $VERBOSE==1 ? "\n" : ' ',
  6013. perfname($nicename), $size, $warning, $critical;
  6014. ($max=$size, $pmax=$psize, $kmax=$kind, $nmax=$name, $smax=$schema) if $size > $max;
  6015. }
  6016. if ($max < 0) {
  6017. add_unknown msg('no-match-rel');
  6018. next;
  6019. }
  6020. if ($MRTG) {
  6021. my $msg = sprintf 'DB: %s %s %s%s',
  6022. $db->{dbname},
  6023. $kmax eq 'i' ? 'INDEX:' : 'TABLE:',
  6024. $kmax eq 'i' ? '' : "$smax.",
  6025. $nmax;
  6026. do_mrtg({one => $max, msg => $msg});
  6027. next;
  6028. }
  6029.  
  6030. my $msg;
  6031. if ($relkind eq 'relation') {
  6032. if ($kmax eq 'r') {
  6033. $msg = msg('relsize-msg-relt', "$smax.$nmax", $pmax);
  6034. }
  6035. else {
  6036. $msg = msg('relsize-msg-reli', $nmax, $pmax);
  6037. }
  6038. }
  6039. elsif ($relkind eq 'table') {
  6040. $msg = msg('relsize-msg-tab', "$smax.$nmax", $pmax);
  6041. }
  6042. else {
  6043. $msg = msg('relsize-msg-ind', $nmax, $pmax);
  6044. }
  6045. if (length $critical and $max >= $critical) {
  6046. add_critical $msg;
  6047. }
  6048. elsif (length $warning and $max >= $warning) {
  6049. add_warning $msg;
  6050. }
  6051. else {
  6052. add_ok $msg;
  6053. }
  6054. }
  6055.  
  6056. return;
  6057.  
  6058. } ## end of check_relation_size
  6059.  
  6060.  
  6061. sub check_table_size {
  6062. return check_relation_size('table');
  6063. }
  6064. sub check_index_size {
  6065. return check_relation_size('index');
  6066. }
  6067.  
  6068.  
  6069. sub check_replicate_row {
  6070.  
  6071. ## Make an update on one server, make sure it propogates to others
  6072. ## Supports: Nagios, MRTG
  6073. ## Warning and critical are time to replicate to all slaves
  6074.  
  6075. my ($warning, $critical) = validate_range({type => 'time', leastone => 1, forcemrtg => 1});
  6076.  
  6077. if ($warning and $critical and $warning > $critical) {
  6078. ndie msg('range-warnbig');
  6079. }
  6080.  
  6081. if (!$opt{repinfo}) {
  6082. ndie msg('rep-noarg');
  6083. }
  6084. my @repinfo = split /,/ => ($opt{repinfo} || '');
  6085. if ($#repinfo != 5) {
  6086. ndie msg('rep-badarg');
  6087. }
  6088. my ($table,$pk,$id,$col,$val1,$val2) = (@repinfo);
  6089.  
  6090. ## Quote everything, just to be safe (e.g. columns named 'desc')
  6091. $table = qq{"$table"};
  6092. $pk = qq{"$pk"};
  6093. $col = qq{"$col"};
  6094.  
  6095. if ($val1 eq $val2) {
  6096. ndie msg('rep-duh');
  6097. }
  6098.  
  6099. $SQL = qq{UPDATE $table SET $col = 'X' WHERE $pk = '$id'};
  6100. (my $update1 = $SQL) =~ s/X/$val1/;
  6101. (my $update2 = $SQL) =~ s/X/$val2/;
  6102. my $select = qq{SELECT $col AS c FROM $table WHERE $pk = '$id'};
  6103.  
  6104. ## Are they the same on both sides? Must be yes, or we error out
  6105.  
  6106. ## We assume this is a single server
  6107. my $info1 = run_command($select);
  6108. ## Squirrel away the $db setting for later
  6109. my $sourcedb = $info1->{db}[0];
  6110. if (!defined $sourcedb) {
  6111. ndie msg('rep-norow', "$table.$col");
  6112. }
  6113. my $value1 = $info1->{db}[0]{slurp}[0]{c};
  6114.  
  6115. my $info2 = run_command($select, { dbnumber => 2 });
  6116. my $slave = 0;
  6117. for my $d (@{$info2->{db}}) {
  6118. $slave++;
  6119. my $value2 = $d->{slurp}[0]{c};
  6120. if ($value1 ne $value2) {
  6121. ndie msg('rep-notsame');
  6122. }
  6123. }
  6124. my $numslaves = $slave;
  6125. if ($numslaves < 1) {
  6126. ndie msg('rep-noslaves');
  6127. }
  6128.  
  6129. my ($update,$newval);
  6130. if ($value1 eq $val1) {
  6131. $update = $update2;
  6132. $newval = $val2;
  6133. }
  6134. elsif ($value1 eq $val2) {
  6135. $update = $update1;
  6136. $newval = $val1;
  6137. }
  6138. else {
  6139. ndie msg('rep-wrongvals', $value1, $val1, $val2);
  6140. }
  6141.  
  6142. $info1 = run_command($update, { dbnumber => 1, failok => 1 } );
  6143.  
  6144. ## Make sure the update worked
  6145. if (! defined $info1->{db}[0]) {
  6146. ndie msg('rep-sourcefail');
  6147. }
  6148.  
  6149. my $err = $info1->{db}[0]{error} || '';
  6150. if ($err) {
  6151. $err =~ s/ERROR://; ## e.g. Slony read-only
  6152. ndie $err;
  6153. }
  6154.  
  6155. ## Start the clock
  6156. my $starttime = time();
  6157.  
  6158. ## Loop until we get a match, check each in turn
  6159. my %slave;
  6160. my $time = 0;
  6161. LOOP: {
  6162. $info2 = run_command($select, { dbnumber => 2 } );
  6163. ## Reset for final output
  6164. $db = $sourcedb;
  6165.  
  6166. $slave = 0;
  6167. for my $d (@{$info2->{db}}) {
  6168. $slave++;
  6169. next if exists $slave{$slave};
  6170. my $value2 = $d->{slurp}[0]{c};
  6171. $time = $db->{totaltime} = time - $starttime;
  6172. if ($value2 eq $newval) {
  6173. $slave{$slave} = $time;
  6174. next;
  6175. }
  6176. if ($warning and $time > $warning) {
  6177. $MRTG and do_mrtg({one => 0, msg => $time});
  6178. add_warning msg('rep-fail', $slave);
  6179. return;
  6180. }
  6181. elsif ($critical and $time > $critical) {
  6182. $MRTG and do_mrtg({one => 0, msg => $time});
  6183. add_critical msg('rep-fail', $slave);
  6184. return;
  6185. }
  6186. }
  6187. ## Did they all match?
  6188. my $k = keys %slave;
  6189. if (keys %slave >= $numslaves) {
  6190. $MRTG and do_mrtg({one => $time});
  6191. add_ok msg('rep-ok');
  6192. return;
  6193. }
  6194. sleep 1;
  6195. redo;
  6196. }
  6197.  
  6198. $MRTG and ndie msg('rep-timeout', $time);
  6199. add_unknown msg('rep-unknown');
  6200. return;
  6201.  
  6202. } ## end of check_replicate_row
  6203.  
  6204.  
  6205. sub check_same_schema {
  6206.  
  6207. ## Verify that all relations inside two or more databases are the same
  6208. ## Supports: Nagios
  6209. ## Include and exclude are supported
  6210. ## Warning and critical are not used
  6211. ## The filter argument is supported
  6212.  
  6213. ## We override the usual $db->{totaltime} with our own counter
  6214. my $start = [gettimeofday()];
  6215.  
  6216. ## Check for filtering rules, then store inside opt{filtered}
  6217. my %filter;
  6218. if (exists $opt{filter}) {
  6219. for my $item (@{ $opt{filter} }) {
  6220. ## Can separate by whitespace or commas
  6221. for my $phrase (split /[\s,]+/ => $item) {
  6222.  
  6223. ## Can be plain (e.g. nouser) or regex based exclusion, e.g. nouser=bob
  6224. next if $phrase !~ /(\w+)=?\s*(.*)/o;
  6225. my ($name,$regex) = (lc $1,$2||'');
  6226.  
  6227. ## Names are standardized with regards to plurals and casing
  6228. $name =~ s/([aeiou])s$/$1/o;
  6229. $name =~ s/s$//o;
  6230.  
  6231. if (! length $regex) {
  6232. $filter{"$name"} = 1;
  6233. }
  6234. else {
  6235. push @{$filter{"${name}_regex"}} => $regex;
  6236. }
  6237. }
  6238. $VERBOSE >= 3 and warn Dumper \%filter;
  6239. }
  6240. }
  6241. $opt{filtered} = \%filter;
  6242.  
  6243. ## See how many databases we are using
  6244. my $numdbs = @targetdb;
  6245. $VERBOSE >= 3 and warn "Number of databases is $numdbs\n";
  6246.  
  6247. ## If only a single database is given, this is a time-based comparison
  6248. ## In other words, we write and read a local file
  6249. my $samedb = 0;
  6250. if (1 == $numdbs) {
  6251. $samedb = 1;
  6252. $numdbs = 2;
  6253. }
  6254.  
  6255. ## Sanity check
  6256. if ($opt{suffix} and ! $samedb) {
  6257. ndie msg('ss-suffix');
  6258. }
  6259.  
  6260. ## Version information about each database, by number
  6261. my %dbver;
  6262.  
  6263. ## Verify we can connect to each database, and grab version information
  6264. for my $num (1..$numdbs) {
  6265.  
  6266. ## No need to check the same database twice!
  6267. last if $samedb and $num > 1;
  6268.  
  6269. $SQL = 'SELECT version()';
  6270. my $info = run_command($SQL, { dbnumber => $num } );
  6271.  
  6272. ## We need a global $db, so we'll use the first database
  6273. $db = $info->{db}[0] if 1 == $num;
  6274.  
  6275. my $foo = $info->{db}[0];
  6276. my $version = $foo->{slurp}[0]{version};
  6277. $version =~ /\D+(\d+\.\d+)(\S+)/i or die qq{Invalid version: $version\n};
  6278. my ($full,$major,$revision) = ("$1$2",$1,$2);
  6279. $revision =~ s/^\.//;
  6280. $dbver{$num} = {
  6281. full => $version,
  6282. version => $full,
  6283. major => $major,
  6284. revision => $revision,
  6285. };
  6286.  
  6287. $targetdb[$num-1]{pgversion} = $full;
  6288.  
  6289. }
  6290.  
  6291. ## An ordered list of all the things we check.
  6292. ## Order is important here, as when reporting, some things
  6293. ## can mask reporting on others (e.g. no need to report missing tables
  6294. ## if the entire schema has already been reported as missing)
  6295. ## We also indicate which columns should be ignored when comparing,
  6296. ## as well as which columns are of a 'list' nature
  6297. my @catalog_items = (
  6298. [user => 'usesysid', 'useconfig' ],
  6299. [language => 'laninline,lanplcallfoid,lanvalidator', '' ],
  6300. [operator => '', '' ],
  6301. [type => '', '' ],
  6302. [schema => '', '' ],
  6303. [function => 'source_checksum,prolang,prorettype', '' ],
  6304. [table => 'reltype,relfrozenxid,relpages,
  6305. reltuples,relnatts', '' ],
  6306. [view => 'reltype', '' ],
  6307. [sequence => 'reltype,log_cnt,relnatts,is_called', '' ],
  6308. [index => 'relpages,reltuples,indpred,indclass,
  6309. indexprs,indcheckxmin', '' ],
  6310. [trigger => '', '' ],
  6311. [constraint => 'conbin', '' ],
  6312. [column => 'atttypid,attnum,attbyval', '' ],
  6313. );
  6314.  
  6315. ## Where we store all the information, per-database
  6316. my %thing;
  6317.  
  6318. my $saved_db;
  6319. for my $x (1..$numdbs) {
  6320.  
  6321. if ($x > 1 and $samedb) {
  6322. ## This means we are looking at a single database over time
  6323. ## We load the stored information into the current $dbinfo
  6324. my $filename = audit_filename();
  6325.  
  6326. if (! -e $filename) {
  6327. ## We've not saved any information about this database yet
  6328. ## Store the info and exit!
  6329. my $version = $dbver{1}{version};
  6330. write_audit_file({ file => $filename, 'same_schema' => 1,
  6331. info => $thing{1}, pgversion => $version });
  6332. print msg('ss-createfile', $filename) . "\n";
  6333. exit 0;
  6334. }
  6335.  
  6336. ## Meta-information from the file
  6337. my ($conninfo,$ctime,$cversion,$pgversion,$cdbname,$chost,$cport,$cuser);
  6338.  
  6339. ($thing{$x},$conninfo,$ctime,$cversion,$pgversion,$cdbname,$chost,$cport,$cuser)
  6340. = read_audit_file($filename);
  6341.  
  6342. ## Count total objects
  6343. my $totalcount = 0;
  6344. for (keys %{ $thing{$x} }) {
  6345. $totalcount += keys %{ $thing{$x}{$_} };
  6346. }
  6347.  
  6348. ## Add the meta info back into the targetdb
  6349. push @targetdb, {
  6350. filename => $filename,
  6351. conninfo => $conninfo,
  6352. ctime => $ctime,
  6353. cversion => $cversion,
  6354. dbname => $cdbname,
  6355. port => $cport,
  6356. host => $chost,
  6357. dbuser => $cuser,
  6358. pgversion => $pgversion,
  6359. objects => $totalcount,
  6360. };
  6361.  
  6362. next;
  6363.  
  6364. } ## end if samedb
  6365.  
  6366. ## Hash of this round's information
  6367. my $dbinfo;
  6368.  
  6369. for (@catalog_items) {
  6370. my $name = $_->[0];
  6371. $dbinfo->{$name} = find_catalog_info($name, $x, $dbver{$x});
  6372. }
  6373.  
  6374. ## TODO:
  6375. ## operator class, cast, aggregate, conversion, domain, tablespace, foreign tables
  6376. ## foreign server, wrapper, collation, extensions, roles?
  6377.  
  6378. ## Map the oid back to the user, for ease later on
  6379. for my $row (values %{ $dbinfo->{user} }) {
  6380. $dbinfo->{useroid}{$row->{usesysid}} = $row->{usename};
  6381. }
  6382.  
  6383. $thing{$x} = $dbinfo;
  6384.  
  6385. ## Count total objects
  6386. my $totalcount = 0;
  6387. for (keys %{ $thing{$x} }) {
  6388. $totalcount += keys %{ $thing{$x}{$_} };
  6389. }
  6390.  
  6391. $targetdb[$x-1]{objects} = $totalcount;
  6392.  
  6393.  
  6394. } ## end each database to query
  6395.  
  6396. ## Start comparing, and put any differences into %fail
  6397. my %fail;
  6398.  
  6399. ## Ugly, but going to use this as a global for the subroutines below:
  6400. $opt{failcount} = 0;
  6401.  
  6402. ## Simple checks that items exist on each database
  6403. for (@catalog_items) {
  6404. my $name = $_->[0];
  6405. $fail{$name}{exists} = schema_item_exists($name, \%thing);
  6406. }
  6407.  
  6408. ## Now check for some more specific items for each item class.
  6409. ## For many of these, we want to compare all columns except for
  6410. ## certain known exceptions (e.g. anything oid-based)
  6411. ## Because we may go across versions, if the column does not exist
  6412. ## somewhere, it is simply silently ignored
  6413. ## Some items are lists (e.g. acls) and must be treated differently
  6414.  
  6415. for (@catalog_items) {
  6416. my ($name,$ignore,$lists) = @$_;
  6417. $fail{$name}{diff} = schema_item_differences({
  6418. items => \%thing,
  6419. name => $name,
  6420. ignore => $ignore,
  6421. lists => $lists,
  6422. });
  6423. }
  6424.  
  6425. ## Remove empty hashes for a cleaner debug dump
  6426. for (keys %fail) {
  6427. if (exists $fail{$_}{diff} and ! keys %{ $fail{$_}{diff} }) {
  6428. delete $fail{$_}{diff};
  6429. }
  6430. }
  6431.  
  6432. ## Set the total time
  6433. $db->{totaltime} = sprintf '%.2f', tv_interval($start);
  6434.  
  6435. ## Before we outpu any results, rewrite the audit file if needed
  6436. ## We do this if we are reading from a saved file,
  6437. ## and the "replace" argument is set
  6438. if ($samedb and $opt{replace}) {
  6439. my $filename = audit_filename();
  6440. if ( -e $filename) {
  6441. ## Move the old one to a backup version
  6442. my $backupfile = "$filename.backup";
  6443. rename $filename, $backupfile;
  6444. }
  6445. my $version = $dbver{1}{version};
  6446. write_audit_file({ file => $filename, 'same_schema' => 1,
  6447. info => $thing{1}, pgversion => $version });
  6448. ## Cannot print this message as we are outputting Nagios stuff
  6449. #print msg('ss-createfile', $filename) . "\n";
  6450. }
  6451.  
  6452. ## Comparison is done, let's report the results
  6453. if (! $opt{failcount}) {
  6454. add_ok msg('ss-matched');
  6455. return;
  6456. }
  6457.  
  6458. ## Build a pretty message giving all the gory details
  6459. my $msg = '';
  6460.  
  6461. ## Adjust the output based on the leading message sizes
  6462. my $maxsize = 1;
  6463. my $msg_exists = msg('ss-existson');
  6464. my $msg_missing = msg('ss-missingon');
  6465. $maxsize = length $msg_exists if length $msg_exists > $maxsize;
  6466. $maxsize = length $msg_missing if length $msg_missing > $maxsize;
  6467.  
  6468. ## Walk through each item type in alphabetical order and output the differences
  6469. for (@catalog_items) {
  6470. my $item = $_->[0];
  6471.  
  6472. ## Pretty name for this type of item. No matches is okay!
  6473. $opt{nomsgok} = 1;
  6474. my $pitem = msg($item) || ucfirst $item;
  6475. $opt{nomsgok} = 0;
  6476.  
  6477. ## See if there are any items of this class that only exist on some
  6478. my $e = $fail{$item}{exists};
  6479. if (keys %$e) {
  6480. for my $name (sort keys %$e) {
  6481. ## We do not want to report twice on things that appear inside of schemas
  6482. ## However, we do report if the schema *does* exist on any of the missing databases
  6483. if ($item ne 'schema' and $name =~ /(.+?)\./) {
  6484. my $schema = $1;
  6485. ## How many databases do not have this?
  6486. my $missingcount = keys %{ $e->{$name}{nothere} };
  6487. my $noschemacount = 0;
  6488. for my $db (keys %{ $e->{$name}{nothere} }) {
  6489. if (exists $fail{schema}{exists}{$schema}{nothere}{$db}) {
  6490. $noschemacount++;
  6491. }
  6492. }
  6493. if ($missingcount == $noschemacount) {
  6494. next;
  6495. }
  6496. }
  6497.  
  6498. ## Show the list of the item, and a CSV of which databases have it and which don't
  6499. my $isthere = join ', ' => sort { $a<=>$b } keys %{ $e->{$name}{isthere} };
  6500. my $nothere = join ', ' => sort { $a<=>$b } keys %{ $e->{$name}{nothere} };
  6501. $msg .= sprintf "%s\n %-*s %s\n %-*s %s\n",
  6502. msg('ss-noexist', $pitem, $name),
  6503. $maxsize, $msg_exists,
  6504. $isthere,
  6505. $maxsize, $msg_missing,
  6506. $nothere;
  6507. }
  6508. }
  6509.  
  6510. ## See if there are any items for this class that have differences
  6511. my $d = $fail{$item}{diff};
  6512. if (keys %$d) {
  6513.  
  6514. for my $name (sort keys %$d) {
  6515. my $tdiff = $d->{$name};
  6516.  
  6517. ## Any raw column differences?
  6518. if (exists $tdiff->{coldiff}) {
  6519. my @msg;
  6520.  
  6521. for my $col (sort keys %{ $tdiff->{coldiff} }) {
  6522.  
  6523. ## Do not show index 'owners': already covered by the table itself
  6524. if ($col eq 'owner' and $item eq 'index') {
  6525. next;
  6526. }
  6527.  
  6528. ## Do not show column number differences if filtered out with "noposition"
  6529. if ($item eq 'column'
  6530. and $col eq 'column_number'
  6531. and $opt{filtered}{noposition}) {
  6532. next;
  6533. }
  6534.  
  6535. ## Do not show function body differences if filtered out with "nofuncbody"
  6536. ## Also skip if the equivalent 'dash' and 'empty'
  6537. if ($item eq 'function'
  6538. and $col eq 'prosrc') {
  6539.  
  6540. next if $opt{filtered}{nofuncbody};
  6541. my ($one,$two);
  6542. for my $db (sort keys %{ $tdiff->{coldiff}{$col} }) {
  6543. if (defined $one) {
  6544. $two = $tdiff->{coldiff}{$col}{$db};
  6545. }
  6546. else {
  6547. $one = $tdiff->{coldiff}{$col}{$db};
  6548. }
  6549. }
  6550. next if $one eq '-' and $two eq '';
  6551. next if $one eq '' and $two eq '-';
  6552. }
  6553.  
  6554. ## If we are doing a historical comparison, skip some items
  6555. if ($samedb) {
  6556. if ($item eq 'sequence'
  6557. and $col eq 'last_value') {
  6558. next;
  6559. }
  6560. }
  6561.  
  6562. push @msg => sprintf " %s\n", msg('ss-different', $col);
  6563. for my $db (sort keys %{ $tdiff->{coldiff}{$col} }) {
  6564. push @msg => sprintf " %s %s: %s\n",
  6565. ucfirst msg('database'),
  6566. $db,
  6567. $tdiff->{coldiff}{$col}{$db};
  6568. }
  6569. }
  6570.  
  6571. if (@msg) {
  6572. $msg .= qq{$pitem "$name":\n};
  6573. $msg .= $_ for @msg;
  6574. }
  6575. else {
  6576. ## Everything got filtered out, so decrement this item
  6577. $opt{failcount}--;
  6578. }
  6579. }
  6580.  
  6581. ## Any multi-item column differences?
  6582. if (exists $tdiff->{list}) {
  6583.  
  6584. my @msg;
  6585. for my $col (sort keys %{ $tdiff->{list} }) {
  6586.  
  6587. ## Exclude permissions if 'noperm' filter is set
  6588. if ($col =~ /.acl$/ and $opt{filtered}{noperm}) {
  6589. next;
  6590. }
  6591.  
  6592. if (exists $tdiff->{list}{$col}{exists}) {
  6593. my $ex = $tdiff->{list}{$col}{exists};
  6594. for my $name (sort keys %$ex) {
  6595. push @msg => sprintf qq{ "%s":\n %s\n},
  6596. $col,
  6597. msg('ss-notset', $name);
  6598. my $isthere = join ', ' => sort { $a<=>$b } keys %{ $ex->{$name}{isthere} };
  6599. my $nothere = join ', ' => sort { $a<=>$b } keys %{ $ex->{$name}{nothere} };
  6600. push @msg => sprintf " %-*s %s\n %-*s %s\n",
  6601. $maxsize, $msg_exists,
  6602. $isthere,
  6603. $maxsize, $msg_missing,
  6604. $nothere;
  6605. }
  6606. }
  6607. if (exists $tdiff->{list}{$col}{diff}) {
  6608. for my $setting (sort keys %{ $tdiff->{list}{$col}{diff} }) {
  6609.  
  6610. push @msg => sprintf qq{ "%s":\n %s\n},
  6611. $col,
  6612. msg('ss-different', $setting);
  6613. for my $db (sort keys %{ $tdiff->{list}{$col}{diff}{$setting} }) {
  6614. my $val = $tdiff->{list}{$col}{diff}{$setting}{$db};
  6615. push @msg => sprintf " %s %s: %s\n",
  6616. ucfirst msg('database'),
  6617. $db,
  6618. $val;
  6619. }
  6620. }
  6621. }
  6622. }
  6623.  
  6624. if (@msg) {
  6625. $msg .= qq{$pitem "$name":\n};
  6626. $msg .= $_ for @msg;
  6627. }
  6628. else {
  6629. ## No message means it was all filtered out, so we decrment the master count
  6630. $opt{failcount}--;
  6631. }
  6632. }
  6633. }
  6634. }
  6635. }
  6636.  
  6637. ## We may have no items due to exclusions!
  6638. if (! $opt{failcount}) {
  6639. add_ok msg('ss-matched');
  6640. return;
  6641. }
  6642.  
  6643. $db->{perf} = "\n$msg";
  6644. add_critical msg('ss-failed', $opt{failcount});
  6645. return;
  6646.  
  6647. } ## end of check_same_schema
  6648.  
  6649.  
  6650. sub audit_filename {
  6651.  
  6652. ## Generate the name of the file to store audit information
  6653.  
  6654. ## Get the connection information for this connection
  6655. my $filename = run_command('foo', { conninfo => 1 });
  6656. ## Do not care about the username
  6657. $filename =~ s/ user=(.+)//;
  6658. ## Strip out the host if not used
  6659. $filename =~ s/ host=<none>//;
  6660. ## Replace any other spaces
  6661. $filename =~ s/ /./g;
  6662. ## Equals have to be escaped, so we'll change them to a dot
  6663. $filename =~ s/=/./g;
  6664. ## The final filename to use
  6665. $filename = "check_postgres.audit.$filename";
  6666.  
  6667. ## The host name may have slashes, so change to underscores
  6668. $filename =~ s{\/}{_}g;
  6669.  
  6670. ## Got a user-supplied extension? Add it now.
  6671. if ($opt{suffix}) {
  6672. $filename .= ".$opt{suffix}";
  6673. }
  6674.  
  6675. return $filename;
  6676.  
  6677. } ## end of audit_filename
  6678.  
  6679.  
  6680. sub write_audit_file {
  6681.  
  6682. ## Save a new copy of the audit file
  6683. my $arg = shift || {};
  6684. my $filename = $arg->{filename} || audit_filename();
  6685. my $info = $arg->{info} || die;
  6686.  
  6687. ## Create a connection information string
  6688. my $row = $targetdb[0];
  6689. my $conninfo = sprintf '%s%s%s%s',
  6690. defined $row->{port} ? qq{port=$row->{port} } : '',
  6691. defined $row->{host} ? qq{host=$row->{host} } : '',
  6692. defined $row->{dbname} ? qq{dbname=$row->{dbname} } : '',
  6693. defined $row->{dbuser} ? qq{user=$row->{dbuser} } : '';
  6694.  
  6695. open my $fh, '>', $filename or die qq{Could not open "$filename": $!\n};
  6696. print {$fh} "## Audit file for check_postgres\n";
  6697. print {$fh} "## CP version: $VERSION\n";
  6698. print {$fh} "## PG version: $arg->{pgversion}\n";
  6699. printf {$fh} "## Created: %s\n", scalar localtime();
  6700. print {$fh} "## Connection: $conninfo\n";
  6701. print {$fh} "## Database name: $row->{dbname}\n";
  6702. print {$fh} "## Host: $row->{host}\n";
  6703. print {$fh} "## Port: $row->{port}\n";
  6704. print {$fh} "## User: $row->{dbuser}\n";
  6705. if ($arg->{same_schema}) {
  6706. print {$fh} "## Start of same_schema information:\n";
  6707. {
  6708. local $Data::Dumper::Indent = 1;
  6709. print {$fh} Dumper $info;
  6710. }
  6711. print {$fh} "## End of same_schema information\n";
  6712. }
  6713.  
  6714. close $fh or warn qq{Could not close "$filename": $!\n};
  6715.  
  6716. } ## end of write_audit_file
  6717.  
  6718.  
  6719. sub read_audit_file {
  6720.  
  6721. ## Read in the data from a historical file
  6722. ## Returns four items:
  6723. ## 1. The standard catalog structure that was saved
  6724. ## 2. Connection information string
  6725. ## 3. Date the file was created
  6726. ## 4. The version it was created with
  6727.  
  6728. my $filename = shift;
  6729.  
  6730. open my $fh, '<', $filename or die qq{Could not open "$filename": $!\n};
  6731. my $inside = 0;
  6732. my $data = '';
  6733. my ($conninfo,$ctime,$cversion,$pgversion) = ('???','???','???','???');
  6734. my ($cdbname,$chost,$cport,$cuser) = ('???','???','???','???');
  6735. while (<$fh>) {
  6736. if (!$inside) {
  6737. if (/Start of same_schema/) {
  6738. $inside = 1;
  6739. }
  6740. elsif (/CP version: (.+)/) {
  6741. $cversion = $1;
  6742. }
  6743. elsif (/PG version: (.+)/) {
  6744. $pgversion = $1;
  6745. }
  6746. elsif (/Created: (.+)/) {
  6747. $ctime = $1;
  6748. }
  6749. elsif (/Connection: (.+)/) {
  6750. $conninfo = $1;
  6751. }
  6752. elsif (/Database name: (.+)/) {
  6753. $cdbname = $1;
  6754. }
  6755. elsif (/Host: (.+)/) {
  6756. $chost = $1;
  6757. }
  6758. elsif (/Port: (.+)/) {
  6759. $cport = $1;
  6760. }
  6761. elsif (/User: (.+)/) {
  6762. $cuser = $1;
  6763. }
  6764. }
  6765. elsif (/End of same_schema/) {
  6766. last;
  6767. }
  6768. else {
  6769. $data .= $_;
  6770. }
  6771. }
  6772. close $fh or warn qq{Could not close "$filename": $!\n};
  6773.  
  6774. my $POSTGRES1;
  6775. eval $data; ## no critic (ProhibitStringyEval)
  6776. if ($@) {
  6777. die qq{Failed to parse file "$filename": $@\n};
  6778. }
  6779. return $POSTGRES1, $conninfo, $ctime, $cversion,
  6780. $pgversion, $cdbname, $chost, $cport, $cuser;
  6781.  
  6782. } ## end of read_audit_file
  6783.  
  6784.  
  6785. sub schema_item_exists {
  6786.  
  6787. ## Compare a certain class of items across all databases for existence
  6788. ## Returns a hashref of item names, with "isthere" and "nothere"
  6789. ## with keys of database numbers underneath that
  6790.  
  6791. my $item_class = shift;
  6792. my $itemhash = shift;
  6793.  
  6794. ## Things that failed to match:
  6795. my %nomatch;
  6796.  
  6797. my $key = "no${item_class}_regex";
  6798. my $exclude_regex = exists $opt{filtered}->{$key} ? $opt{filtered}->{$key} : [];
  6799.  
  6800. for my $db1 (sort keys %$itemhash) {
  6801. for my $db2 (sort keys %$itemhash) {
  6802. next if $db1 == $db2;
  6803. for my $name (sort keys %{ $itemhash->{$db1}{$item_class} }) {
  6804.  
  6805. ## Can exclude by 'filter' based regex
  6806. next if grep { $name eq $_ } @$exclude_regex;
  6807.  
  6808. if (! exists $itemhash->{$db2}{$item_class}{$name}) {
  6809.  
  6810. ## Special exception for columns: do not add if the table is non-existent
  6811. if ($item_class eq 'column') {
  6812. (my $tablename = $name) =~ s/(.+)\..+/$1/;
  6813. next if ! exists $itemhash->{$db2}{table}{$tablename};
  6814. }
  6815.  
  6816. ## Special exception for triggers: do not add if the table is non-existent
  6817. if ($item_class eq 'trigger') {
  6818. my $it = $itemhash->{$db1}{$item_class}{$name};
  6819. my $tablename = "$it->{tschema}.$it->{tname}";
  6820. next if ! exists $itemhash->{$db2}{table}{$tablename};
  6821. }
  6822.  
  6823. $nomatch{$name}{isthere}{$db1} = 1;
  6824. $nomatch{$name}{nothere}{$db2} = 1;
  6825. }
  6826. }
  6827. }
  6828. }
  6829.  
  6830. ## Increment our fail count once per item mismatch
  6831. $opt{failcount} += keys %nomatch;
  6832.  
  6833. return \%nomatch;
  6834.  
  6835. } ## end of schema_item_exists
  6836.  
  6837.  
  6838. sub schema_item_differences {
  6839.  
  6840. ## Compare a certain class of items across all databases for differences
  6841. ## Takes a hashref of argument, including:
  6842. ## name: the item class name
  6843. ## items: the main hashref of all items
  6844. ## ignore: which fields to ignore. CSV
  6845. ## lists: which fields are lists. CSV
  6846. ## Modified the items hashref by incrementing items->{failcount}
  6847. ## Returns s hashref of item names, with details as to the diffs therein
  6848.  
  6849. my $arg = shift;
  6850.  
  6851. my $item_class = $arg->{name} or die;
  6852. my $itemhash = $arg->{items} or die;
  6853.  
  6854. ## Things we completely ignore:
  6855. my $ignore = { oid => 1 };
  6856. if (exists $arg->{ignore}) {
  6857. for my $item (split /\s*,\s*/ => $arg->{ignore}) {
  6858. $ignore->{$item} = 1;
  6859. }
  6860. }
  6861.  
  6862. ## Things that are handled as lists:
  6863. my $lists = {};
  6864. if (exists $arg->{lists}) {
  6865. for my $item (split /\s*,\s*/ => $arg->{lists}) {
  6866. $lists->{$item} = 1;
  6867. }
  6868. }
  6869.  
  6870. ## The final lists of mismatched items we pass back
  6871. my %nomatch;
  6872.  
  6873. my $key = "no${item_class}_regex";
  6874. my $exclude_regex = exists $opt{filtered}->{$key} ? $opt{filtered}->{$key} : [];
  6875.  
  6876. for my $db1 (sort keys %$itemhash) {
  6877. for my $db2 (sort keys %$itemhash) {
  6878. next if $db1 >= $db2;
  6879. for my $name (sort keys %{ $itemhash->{$db1}{$item_class} }) {
  6880.  
  6881. ## Can exclude by 'filter' based regex
  6882. next if grep { $name eq $_ } @$exclude_regex;
  6883.  
  6884. ## This case has already been handled:
  6885. next if ! exists $itemhash->{$db2}{$item_class}{$name};
  6886.  
  6887. ## Special exception for columns: do not add if the table is non-existent
  6888. if ($item_class eq 'column') {
  6889. (my $tablename = $name) =~ s/(.+)\..+/$1/;
  6890. next if ! exists $itemhash->{$db2}{table}{$tablename};
  6891. }
  6892.  
  6893. my $one = $itemhash->{$db1}{$item_class}{$name};
  6894. my $two = $itemhash->{$db2}{$item_class}{$name};
  6895.  
  6896. for my $col (keys %$one) {
  6897.  
  6898. ## Skip if this col is ignored
  6899. next if exists $ignore->{$col};
  6900.  
  6901. ## If it doesn't exist on the other, just ignore it
  6902. next if ! exists $two->{$col};
  6903.  
  6904. ## If they are the same, move on!
  6905. next if $one->{$col} eq $two->{$col};
  6906.  
  6907. ## Skip certain known numeric fields that have text versions:
  6908. next if $col =~ /.(?:namespace|owner|filenode|oid|relid)$/;
  6909.  
  6910. ## If not a list, just report on the exact match here and move on:
  6911. if (! exists $lists->{$col} and $col !~ /.acl$/) {
  6912. $nomatch{$name}{coldiff}{$col}{$db1} = $one->{$col};
  6913. $nomatch{$name}{coldiff}{$col}{$db2} = $two->{$col};
  6914. next;
  6915. }
  6916.  
  6917. ## This is a list, so we have to break it down to see if it is really different
  6918. ## May be empty or of the form {foo=bar,baz=yak}
  6919.  
  6920. my (%list1,%list2);
  6921. my ($uno,$dos) = ($one->{$col}, $two->{$col});
  6922.  
  6923. if (length $uno) {
  6924. die "Invalid list: $uno for db $db1:$name:$col\n" if $uno !~ /^{(.+)}$/;
  6925. my $t = $1;
  6926. my @tlist = ();
  6927. push(@tlist, $+) while $t =~ m{"([^\"\\]*(?:\\.[^\"\\]*)*)",?
  6928. | ([^,]+),?
  6929. | ,
  6930. }gx;
  6931. push(@tlist, undef) if substr($t, -1,1) eq ',';
  6932. %list1 = map { /(.*)=(.+)/ or die "Invalid list: $uno"; $1,$2 } @tlist;
  6933. }
  6934. if (length $dos) {
  6935. die "Invalid list: $dos for db $db2:$name:$col\n" if $dos !~ /^{(.+)}$/;
  6936. my $t = $1;
  6937. my @tlist = ();
  6938. push(@tlist, $+) while $t =~ m{"([^\"\\]*(?:\\.[^\"\\]*)*)",?
  6939. | ([^,]+),?
  6940. | ,
  6941. }gx;
  6942. push(@tlist, undef) if substr($t, -1,1) eq ',';
  6943. %list2 = map { /(.*)=(.+)/ or die "Invalid list: $uno"; $1,$2 } @tlist;
  6944. }
  6945.  
  6946. ## Items in 1 but not 2?
  6947. for my $setting (sort keys %list1) {
  6948. if (! exists $list2{$setting}) {
  6949. $nomatch{$name}{list}{$col}{exists}{$setting}{isthere}{$db1} = 1;
  6950. $nomatch{$name}{list}{$col}{exists}{$setting}{nothere}{$db2} = 1;
  6951. }
  6952. }
  6953.  
  6954. ## Items in 2 but not 1? Value diferences?
  6955. for my $setting (sort keys %list2) {
  6956. if (! exists $list1{$setting}) {
  6957. $nomatch{$name}{list}{$col}{exists}{$setting}{isthere}{$db2} = 1;
  6958. $nomatch{$name}{list}{$col}{exists}{$setting}{nothere}{$db1} = 1;
  6959. }
  6960. elsif ($list1{$setting} ne $list2{$setting}) {
  6961. $nomatch{$name}{list}{$col}{diff}{$setting}{$db1} = $list1{$setting};
  6962. $nomatch{$name}{list}{$col}{diff}{$setting}{$db2} = $list2{$setting};
  6963. }
  6964. }
  6965. }
  6966. }
  6967. }
  6968. }
  6969.  
  6970. $opt{failcount} += keys %nomatch;
  6971.  
  6972. return \%nomatch;
  6973.  
  6974. } ## end of schema_item_differences
  6975.  
  6976.  
  6977. sub find_catalog_info {
  6978.  
  6979. ## Grab information from one or more catalog tables
  6980. ## Convert into a happy hashref and return it
  6981. ## Arguments: three
  6982. ## 1. Type of object
  6983. ## 2. Database number
  6984. ## 3. Version information for the database
  6985. ## Returns: large hashref of information
  6986.  
  6987. ## What type of catalog object this is
  6988. my $type = shift;
  6989.  
  6990. ## We must know about this type
  6991. if (! exists $catalog_info{$type}) {
  6992. die "Unknown type of '$type' sent to find_catalog_info";
  6993. }
  6994. my $ci = $catalog_info{$type};
  6995.  
  6996. ## The final hashref of rows we return
  6997. my $result = {};
  6998.  
  6999. ## Do nothing if we are excluding this type of object entirely
  7000. return $result if $opt{filtered}{"no$type"};
  7001.  
  7002. ## Which database to run this against
  7003. my $dbnum = shift or die;
  7004.  
  7005. ## The version information
  7006. my $dbver = shift or die;
  7007.  
  7008. ## The SQL we use
  7009. my $SQL = $ci->{SQL} or die "No SQL found for type '$type'\n";
  7010.  
  7011. ## Switch to alternate SQL for different versions
  7012. if ($type eq 'language') {
  7013. if (int $dbver->{major} <= 8.2) {
  7014. $SQL = $ci->{SQL2};
  7015. }
  7016. }
  7017.  
  7018. if (exists $ci->{exclude}) {
  7019. if ('temp_schemas' eq $ci->{exclude}) {
  7020. if (! $opt{filtered}{system}) {
  7021. $SQL .= q{ WHERE nspname !~ '^pg_t'};
  7022. }
  7023. }
  7024. elsif ('system' eq $ci->{exclude}) {
  7025. if (! $opt{filtered}{system}) {
  7026. $SQL .= sprintf
  7027. q{ %s n.nspname !~ '^pg' AND n.nspname <> 'information_schema'},
  7028. $SQL =~ /WHERE/ ? 'AND' : 'WHERE';
  7029. }
  7030. }
  7031. else {
  7032. die "Unknown exclude '$ci->{exclude}' called";
  7033. }
  7034. }
  7035.  
  7036. ## Final wrapup
  7037. if (exists $ci->{postSQL}) {
  7038. $SQL .= " $ci->{postSQL}";
  7039. }
  7040.  
  7041. ## Send our SQL to the correct database via psql and grab the results
  7042. my $info = run_command($SQL, { dbnumber => $dbnum });
  7043.  
  7044. ## The row column we use as the main hash key
  7045. my $key = $ci->{keyname} || 'name';
  7046.  
  7047. ## Keep track of the actual column numbers
  7048. my $last_table = '';
  7049. my $colnum = 1;
  7050.  
  7051. ## Only need to pull back the first and only db, so we can say [0] here
  7052. ROW:
  7053. for my $row (@{$info->{db}[0]{slurp}}) {
  7054.  
  7055. ## Remove any information that should be deleted
  7056. for ( @{$info->{deletecols}}) {
  7057. delete $row->{$_};
  7058. }
  7059.  
  7060. ## Determine the name to use. For most things this is simply the passed in key
  7061. my $name = $row->{$key};
  7062.  
  7063. ## For a function, we also want to put the args into the name
  7064. if ($type eq 'function') {
  7065. ## Once per database, grab all mappings
  7066. if (! exists $opt{oid2type}{$dbnum}) {
  7067. $SQL = 'SELECT oid, typname FROM pg_type';
  7068. my $tinfo = run_command($SQL, { dbnumber => $dbnum });
  7069. for my $row (@{ $tinfo->{db}[0]{slurp} }) {
  7070. $opt{oid2type}{$dbnum}{$row->{oid}} = $row->{typname};
  7071. }
  7072. }
  7073. (my $args = $row->{proargtypes}) =~ s/(\d+)/$opt{oid2type}{$dbnum}{$1}||$1/ge;
  7074. $args =~ s/ /,/g;
  7075. $args =~ s/ints/smallint/g;
  7076. $args =~ s/int4/int/g;
  7077. $args =~ s/int8/bigint/g;
  7078. $name .= "($args)";
  7079.  
  7080. }
  7081.  
  7082. ## For columns, reduce the attnum to a simpler canonical form without holes
  7083. if ($type eq 'column') {
  7084. if ($row->{tname} ne $last_table) {
  7085. $last_table = $row->{tname};
  7086. $colnum = 1;
  7087. }
  7088. $row->{column_number} = $colnum++;
  7089. }
  7090.  
  7091. ## Store this row into our result hash, using a good key
  7092. $result->{$name} = $row;
  7093.  
  7094. ## We may want to run additional SQL per row returned
  7095. if (exists $ci->{innerSQL}) {
  7096.  
  7097. if ($type eq 'sequence') {
  7098. ## If this is a sequence, we want to grab them all at once to reduce
  7099. ## the amount of round-trips we do with 'SELECT * FROM seqname'
  7100. if (! exists $opt{seqinfoss}{$dbnum}) {
  7101. $SQL = q{SELECT quote_ident(nspname)||'.'||quote_ident(relname) AS sname }
  7102. . q{FROM pg_class }
  7103. . q{JOIN pg_namespace n ON (n.oid = relnamespace) }
  7104. . q{WHERE relkind = 'S'};
  7105. my $sinfo = run_command($SQL, { dbnumber => $dbnum } );
  7106. $SQL = join "\n UNION ALL\n" =>
  7107. map { "SELECT '$_->{sname}' AS fullname, * FROM $_->{sname}" }
  7108. @{ $sinfo->{db}[0]{slurp}};
  7109. $sinfo = run_command($SQL, { dbnumber => $dbnum } );
  7110.  
  7111. ## Store it back into the global hash
  7112. for my $row (@{ $sinfo->{db}[0]{slurp} }) {
  7113. $opt{seqinfoss}{$dbnum}{$row->{fullname}} = $row;
  7114. }
  7115. }
  7116.  
  7117. ## If it does not exist in the cache, just fall through and do it manually!
  7118. if (exists $opt{seqinfoss}{$dbnum}{$row->{safename}}) {
  7119. $result->{$row->{safename}} = $opt{seqinfoss}{$dbnum}{$row->{safename}};
  7120. next ROW;
  7121. }
  7122. }
  7123.  
  7124. (my $SQL2 = $ci->{innerSQL}) =~ s/ROW(\w+)/$row->{lc $1}/g;
  7125. my $info2 = run_command($SQL2, { dbnumber => $dbnum } );
  7126. for my $row2 (@{ $info2->{db}[0]{slurp} }) {
  7127. for my $inner (keys %{ $row2 }) {
  7128. $result->{$name}{$inner} = $row2->{$inner};
  7129. }
  7130. }
  7131. }
  7132. }
  7133.  
  7134. return $result;
  7135.  
  7136. } ## end of find_catalog_info
  7137.  
  7138.  
  7139. sub check_sequence {
  7140.  
  7141. ## Checks how many values are left in sequences
  7142. ## Supports: Nagios, MRTG
  7143. ## Warning and critical are percentages
  7144. ## Can exclude and include sequences
  7145.  
  7146. my ($warning, $critical) = validate_range
  7147. ({
  7148. type => 'percent',
  7149. default_warning => '85%',
  7150. default_critical => '95%',
  7151. forcemrtg => 1,
  7152. });
  7153.  
  7154. (my $w = $warning) =~ s/\D//;
  7155. (my $c = $critical) =~ s/\D//;
  7156.  
  7157. ## Gather up all sequence names
  7158. ## no critic
  7159. my $SQL = q{
  7160. SELECT DISTINCT ON (nspname, seqname) nspname, seqname,
  7161. quote_ident(nspname) || '.' || quote_ident(seqname) AS safename, typname
  7162. -- sequences by column dependency
  7163. FROM (
  7164. SELECT depnsp.nspname, dep.relname as seqname, typname
  7165. FROM pg_depend
  7166. JOIN pg_class on classid = pg_class.oid
  7167. JOIN pg_class dep on dep.oid = objid
  7168. JOIN pg_namespace depnsp on depnsp.oid= dep.relnamespace
  7169. JOIN pg_class refclass on refclass.oid = refclassid
  7170. JOIN pg_class ref on ref.oid = refobjid
  7171. JOIN pg_namespace refnsp on refnsp.oid = ref.relnamespace
  7172. JOIN pg_attribute refattr ON (refobjid, refobjsubid) = (refattr.attrelid, refattr.attnum)
  7173. JOIN pg_type ON refattr.atttypid = pg_type.oid
  7174. WHERE pg_class.relname = 'pg_class'
  7175. AND refclass.relname = 'pg_class'
  7176. AND dep.relkind in ('S')
  7177. AND ref.relkind in ('r')
  7178. AND typname IN ('int2', 'int4', 'int8')
  7179. UNION ALL
  7180. --sequences by parsing DEFAULT constraints
  7181. SELECT nspname, seq.relname, typname
  7182. FROM pg_attrdef
  7183. JOIN pg_attribute ON (attrelid, attnum) = (adrelid, adnum)
  7184. JOIN pg_type on pg_type.oid = atttypid
  7185. JOIN pg_class rel ON rel.oid = attrelid
  7186. JOIN pg_class seq ON seq.relname = regexp_replace(adsrc, $re$^nextval\('(.+?)'::regclass\)$$re$, $$\1$$)
  7187. AND seq.relnamespace = rel.relnamespace
  7188. JOIN pg_namespace nsp ON nsp.oid = seq.relnamespace
  7189. WHERE adsrc ~ 'nextval' AND seq.relkind = 'S' AND typname IN ('int2', 'int4', 'int8')
  7190. UNION ALL
  7191. -- all sequences, to catch those whose associations are not obviously recorded in pg_catalog
  7192. SELECT nspname, relname, CAST('int8' AS TEXT)
  7193. FROM pg_class
  7194. JOIN pg_namespace nsp ON nsp.oid = relnamespace
  7195. WHERE relkind = 'S'
  7196. ) AS seqs
  7197. WHERE nspname !~ '^pg_temp.*'
  7198. ORDER BY nspname, seqname, typname
  7199. };
  7200. ## use critic
  7201.  
  7202. my $info = run_command($SQL, {regex => qr{\w}, emptyok => 1} );
  7203.  
  7204. my $MAXINT2 = 32767;
  7205. my $MAXINT4 = 2147483647;
  7206. my $MAXINT8 = 9223372036854775807;
  7207.  
  7208. my $limit = 0;
  7209.  
  7210. for $db (@{$info->{db}}) {
  7211. my (@crit,@warn,@ok);
  7212. my $maxp = 0;
  7213. my %seqinfo;
  7214. my %seqperf;
  7215. my $multidb = @{$info->{db}} > 1 ? "$db->{dbname}." : '';
  7216. for my $r (@{$db->{slurp}}) {
  7217. my ($schema, $seq, $seqname, $typename) = @$r{qw/ nspname seqname safename typname /};
  7218. next if skip_item($seq);
  7219. my $maxValue = $typename eq 'int2' ? $MAXINT2 : $typename eq 'int4' ? $MAXINT4 : $MAXINT8;
  7220. $SQL = qq{
  7221. SELECT last_value, slots, used, ROUND(used/slots*100) AS percent,
  7222. CASE WHEN slots < used THEN 0 ELSE slots - used END AS numleft
  7223. FROM (
  7224. SELECT last_value,
  7225. CEIL((LEAST(max_value, $maxValue)-min_value::numeric+1)/increment_by::NUMERIC) AS slots,
  7226. CEIL((last_value-min_value::numeric+1)/increment_by::NUMERIC) AS used
  7227. FROM $seqname) foo
  7228. };
  7229.  
  7230. my $seqinfo = run_command($SQL, { target => $db });
  7231. my $r2 = $seqinfo->{db}[0]{slurp}[0];
  7232. my ($last, $slots, $used, $percent, $left) = @$r2{qw/ last_value slots used percent numleft / };
  7233. if (! defined $last) {
  7234. ndie msg('seq-die', $seqname);
  7235. }
  7236. my $msg = msg('seq-msg', $seqname, $percent, $left);
  7237. my $nicename = perfname("$multidb$seqname");
  7238. $seqperf{$percent}{$seqname} = [$left, " $nicename=$percent%;$w%;$c%"];
  7239. if ($percent >= $maxp) {
  7240. $maxp = $percent;
  7241. if (! exists $opt{perflimit} or $limit++ < $opt{perflimit}) {
  7242. push @{$seqinfo{$percent}} => $MRTG ? [$seqname,$percent,$slots,$used,$left] : $msg;
  7243. }
  7244. }
  7245. next if $MRTG;
  7246.  
  7247. if (length $critical and $percent >= $c) {
  7248. push @crit => $msg;
  7249. }
  7250. elsif (length $warning and $percent >= $w) {
  7251. push @warn => $msg;
  7252. }
  7253. }
  7254. if ($MRTG) {
  7255. my $msg = join ' | ' => map { $_->[0] } @{$seqinfo{$maxp}};
  7256. do_mrtg({one => $maxp, msg => $msg});
  7257. }
  7258. $limit = 0;
  7259. PERF: for my $val (sort { $b <=> $a } keys %seqperf) {
  7260. for my $seq (sort { $seqperf{$val}{$a}->[0] <=> $seqperf{$val}{$b}->[0] or $a cmp $b } keys %{$seqperf{$val}}) {
  7261. last PERF if exists $opt{perflimit} and $limit++ >= $opt{perflimit};
  7262. $db->{perf} .= $seqperf{$val}{$seq}->[1];
  7263. }
  7264. }
  7265.  
  7266. if (@crit) {
  7267. add_critical join ' ' => @crit;
  7268. }
  7269. elsif (@warn) {
  7270. add_warning join ' ' => @warn;
  7271. }
  7272. else {
  7273. if (keys %seqinfo) {
  7274. add_ok join ' ' => @{$seqinfo{$maxp}};
  7275. }
  7276. else {
  7277. add_ok msg('seq-none');
  7278. }
  7279. }
  7280. }
  7281.  
  7282. return;
  7283.  
  7284. } ## end of check_sequence
  7285.  
  7286.  
  7287. sub check_settings_checksum {
  7288.  
  7289. ## Verify the checksum of all settings
  7290. ## Supports: Nagios, MRTG
  7291. ## Not that this will vary from user to user due to ALTER USER
  7292. ## and because superusers see additional settings
  7293. ## One of warning or critical must be given (but not both)
  7294. ## It should run one time to find out the expected checksum
  7295. ## You can use --critical="0" to find out the checksum
  7296. ## You can include or exclude settings as well
  7297. ## Example:
  7298. ## check_postgres_settings_checksum --critical="4e7ba68eb88915d3d1a36b2009da4acd"
  7299.  
  7300. my ($warning, $critical) = validate_range({type => 'checksum', onlyone => 1});
  7301.  
  7302. eval {
  7303. require Digest::MD5;
  7304. };
  7305. if ($@) {
  7306. ndie msg('checksum-nomd');
  7307. }
  7308.  
  7309. $SQL = 'SELECT name, setting FROM pg_settings ORDER BY name';
  7310. my $info = run_command($SQL, { regex => qr[client_encoding] });
  7311.  
  7312. for $db (@{$info->{db}}) {
  7313.  
  7314. my $newstring = '';
  7315. for my $r (@{$db->{slurp}}) {
  7316. next SLURP if skip_item($r->{name});
  7317. $newstring .= "$r->{name} $r->{setting}\n";
  7318. }
  7319. if (! length $newstring) {
  7320. add_unknown msg('no-match-set');
  7321. }
  7322.  
  7323. my $checksum = Digest::MD5::md5_hex($newstring);
  7324.  
  7325. my $msg = msg('checksum-msg', $checksum);
  7326. if ($MRTG) {
  7327. $opt{mrtg} or ndie msg('checksum-nomrtg');
  7328. do_mrtg({one => $opt{mrtg} eq $checksum ? 1 : 0, msg => $checksum});
  7329. }
  7330. if ($critical and $critical ne $checksum) {
  7331. add_critical $msg;
  7332. }
  7333. elsif ($warning and $warning ne $checksum) {
  7334. add_warning $msg;
  7335. }
  7336. elsif (!$critical and !$warning) {
  7337. add_unknown $msg;
  7338. }
  7339. else {
  7340. add_ok $msg;
  7341. }
  7342. }
  7343.  
  7344. return;
  7345.  
  7346. } ## end of check_settings_checksum
  7347.  
  7348.  
  7349. sub check_slony_status {
  7350.  
  7351. ## Checks the sl_status table
  7352. ## Returns unknown if sl_status is not found
  7353. ## Returns critical is status is not "good"
  7354. ## Otherwise, returns based on time-based warning and critical options
  7355. ## Supports: Nagios, MRTG
  7356.  
  7357. my ($warning, $critical) = validate_range
  7358. ({
  7359. type => 'time',
  7360. default_warning => '60',
  7361. default_critical => '300',
  7362. });
  7363.  
  7364. ## If given schemas on the command-line, map back to targetdbs
  7365. if (defined $opt{schema}) {
  7366. my $x = 0;
  7367. for my $db (@targetdb) {
  7368. $db->{schemalist} = $opt{schema}->[$x] || '';
  7369. $x++;
  7370. }
  7371. }
  7372. else {
  7373. ## Otherwise, find all slony schemas and put them in ourselves
  7374. $SQL = q{SELECT quote_ident(nspname) AS nspname FROM pg_namespace WHERE oid IN }.
  7375. q{(SELECT relnamespace FROM pg_class WHERE relkind = 'v' AND relname = 'sl_status')};
  7376. my $info = run_command($SQL);
  7377. for my $db (@{ $info->{db} }) {
  7378. $db->{schemalist} = join ',' => map { $_->{nspname} } @{ $db->{slurp} };
  7379. }
  7380. }
  7381.  
  7382. my $SLSQL =
  7383. q{SELECT
  7384. ROUND(EXTRACT(epoch FROM st_lag_time)) AS lagtime,
  7385. st_origin,
  7386. st_received,
  7387. current_database() AS cd,
  7388. COALESCE(n1.no_comment, '') AS com1,
  7389. COALESCE(n2.no_comment, '') AS com2
  7390. FROM SCHEMA.sl_status
  7391. JOIN SCHEMA.sl_node n1 ON (n1.no_id=st_origin)
  7392. JOIN SCHEMA.sl_node n2 ON (n2.no_id=st_received)};
  7393.  
  7394. my $maxlagtime = -1;
  7395.  
  7396. my $x = 1;
  7397. for $db (@targetdb) {
  7398. next if ! $db->{schemalist};
  7399. $db->{perf} = '';
  7400. my @perf;
  7401. for my $schema (split /,/ => $db->{schemalist}) {
  7402. ## Set for output
  7403. $db->{showschema} = $schema;
  7404.  
  7405. (my $SQL = $SLSQL) =~ s/SCHEMA/$schema/g;
  7406. my $info = run_command($SQL, { dbnumber => $x });
  7407. my $slurp = $info->{db}[0]{slurp}[0];
  7408. if (! defined $slurp->{lagtime}) {
  7409. add_unknown msg('slony-nonumber');
  7410. return;
  7411. }
  7412. my ($lag,$from,$to,$dbname,$fromc,$toc) = @$slurp{qw/ lagtime st_origin st_received cd com1 com2/};
  7413. $maxlagtime = $lag if $lag > $maxlagtime;
  7414. push @perf => [
  7415. $lag,
  7416. $from,
  7417. qq{'$dbname.$schema Node $from($fromc) -> Node $to($toc)'=$lag;$warning;$critical},
  7418. ];
  7419.  
  7420. } ## end each schema in this database
  7421.  
  7422. if ($MRTG) {
  7423. do_mrtg({one => $maxlagtime});
  7424. return;
  7425. }
  7426.  
  7427. $db->{perf} .= join "\n" => map { $_->[2] } sort { $b->[0]<=>$a->[0] or $a->[1] cmp $b->[1] } @perf;
  7428.  
  7429. my $msg = msg('slony-lagtime', $maxlagtime);
  7430. $msg .= sprintf ' (%s)', pretty_time($maxlagtime, $maxlagtime > 500 ? 'S' : '');
  7431. if (length $critical and $maxlagtime >= $critical) {
  7432. add_critical $msg;
  7433. }
  7434. elsif (length $warning and $maxlagtime >= $warning) {
  7435. add_warning $msg;
  7436. }
  7437. else {
  7438. add_ok $msg;
  7439. }
  7440.  
  7441. $x++;
  7442. }
  7443.  
  7444. if ($maxlagtime < 1) { ## No schemas found
  7445. add_unknown msg('slony-noschema');
  7446. }
  7447.  
  7448. return;
  7449.  
  7450. } ## end of check_slony_status
  7451.  
  7452.  
  7453. sub check_timesync {
  7454.  
  7455. ## Compare local time to the database time
  7456. ## Supports: Nagios, MRTG
  7457. ## Warning and critical are given in number of seconds difference
  7458.  
  7459. my ($warning,$critical) = validate_range
  7460. ({
  7461. type => 'seconds',
  7462. default_warning => 2,
  7463. default_critical => 5,
  7464. });
  7465.  
  7466. $SQL = q{SELECT round(extract(epoch FROM now())) AS epok, TO_CHAR(now(),'YYYY-MM-DD HH24:MI:SS') AS pretti};
  7467. my $info = run_command($SQL);
  7468. my $localepoch = time;
  7469. my @l = localtime;
  7470.  
  7471. for $db (@{$info->{db}}) {
  7472. my ($pgepoch,$pgpretty) = @{$db->{slurp}->[0]}{qw/ epok pretti /};
  7473.  
  7474. my $diff = abs($pgepoch - $localepoch);
  7475. if ($MRTG) {
  7476. do_mrtg({one => $diff, msg => "DB: $db->{dbname}"});
  7477. }
  7478. $db->{perf} = sprintf '%s=%ss;%s;%s',
  7479. perfname(msg('timesync-diff')), $diff, $warning, $critical;
  7480.  
  7481. my $localpretty = sprintf '%d-%02d-%02d %02d:%02d:%02d', $l[5]+1900, $l[4]+1, $l[3],$l[2],$l[1],$l[0];
  7482. my $msg = msg('timesync-msg', $diff, $pgpretty, $localpretty);
  7483.  
  7484. if (length $critical and $diff >= $critical) {
  7485. add_critical $msg;
  7486. }
  7487. elsif (length $warning and $diff >= $warning) {
  7488. add_warning $msg;
  7489. }
  7490. else {
  7491. add_ok $msg;
  7492. }
  7493. }
  7494. return;
  7495.  
  7496. } ## end of check_timesync
  7497.  
  7498.  
  7499. sub check_txn_idle {
  7500.  
  7501. ## Check the duration and optionally number of "idle in transaction" processes
  7502. ## Supports: Nagios, MRTG
  7503. ## It makes no sense to run this more than once on the same cluster
  7504. ## Warning and critical are time limits or counts for time limits - default to seconds
  7505. ## Valid time units: s[econd], m[inute], h[our], d[ay]
  7506. ## All above may be written as plural as well (e.g. "2 hours")
  7507. ## Valid counts for time limits: "$int for $time"
  7508. ## Can also ignore databases with exclude and limit with include
  7509. ## Limit to a specific user with the includeuser option
  7510. ## Exclude users with the excludeuser option
  7511.  
  7512. my $type = shift || 'txnidle';
  7513. my $thing = shift || msg('transactions');
  7514. my $perf = shift || msg('txn-time');
  7515. my $start = shift || 'query_start';
  7516. my $clause = shift || q{current_query ~ '^<'};
  7517.  
  7518. ## Extract the warning and critical seconds and counts.
  7519. ## If not given, items will be an empty string
  7520. my ($wcount, $wtime, $ccount, $ctime) = validate_integer_for_time();
  7521.  
  7522. ## We don't GROUP BY because we want details on every connection
  7523. ## Someday we may even break things down by database
  7524. my ($SQL2, $SQL3);
  7525. if ($type ne 'qtime') {
  7526. $SQL = q{SELECT datname, datid, procpid AS pid, usename, client_addr, xact_start, current_query AS current_query, }.
  7527. q{CASE WHEN client_port < 0 THEN 0 ELSE client_port END AS client_port, }.
  7528. qq{COALESCE(ROUND(EXTRACT(epoch FROM now()-$start)),0) AS seconds }.
  7529. qq{FROM pg_stat_activity WHERE $clause$USERWHERECLAUSE }.
  7530. q{ORDER BY xact_start, query_start, procpid DESC};
  7531. ## Craft an alternate version for old servers that do not have the xact_start column:
  7532. ($SQL2 = $SQL) =~ s/xact_start/query_start AS xact_start/;
  7533. $SQL2 =~ s/BY xact_start,/BY/;
  7534. }
  7535. else {
  7536. $SQL = q{SELECT datname, datid, procpid AS pid, usename, client_addr, current_query AS current_query, }.
  7537. q{CASE WHEN client_port < 0 THEN 0 ELSE client_port END AS client_port, }.
  7538. qq{COALESCE(ROUND(EXTRACT(epoch FROM now()-$start)),0) AS seconds }.
  7539. qq{FROM pg_stat_activity WHERE $clause$USERWHERECLAUSE }.
  7540. q{ORDER BY query_start, procpid DESC};
  7541. }
  7542.  
  7543. ## Craft an alternate version for new servers which do not have procpid
  7544. ($SQL3 = $SQL) =~ s/procpid/pid/g;
  7545. $SQL3 =~ s/current_query/query/g;
  7546.  
  7547. my $info = run_command($SQL, { emptyok => 1 , version => [ "<8.3 $SQL2", ">9.1 $SQL3" ] } );
  7548.  
  7549. ## Extract the first entry
  7550. $db = $info->{db}[0];
  7551.  
  7552. ## Store the current longest row
  7553. my $maxr = { seconds => 0 };
  7554.  
  7555. ## How many valid rows did we get?
  7556. my $count = 0;
  7557.  
  7558. ## Info about the top offender
  7559. my $whodunit = '';
  7560. if ($MRTG) {
  7561. if (defined $db->{dbname}) {
  7562. $whodunit = "DB: $db->{dbname}";
  7563. } else {
  7564. $whodunit = sprintf q{DB: %s}, msg('no-db');
  7565. }
  7566. }
  7567.  
  7568. ## Process each returned row
  7569. for my $r (@{ $db->{slurp} }) {
  7570.  
  7571. ## Skip if we don't care about this database
  7572. next if skip_item($r->{datname});
  7573.  
  7574. ## We do a lot of filtering based on the current_query
  7575. my $cq = $r->{current_query};
  7576.  
  7577. ## Return unknown if we cannot see because we are a non-superuser
  7578. if ($cq =~ /insufficient/o) {
  7579. add_unknown msg('psa-nosuper');
  7580. return;
  7581. }
  7582.  
  7583. ## Return unknown if stats_command_string / track_activities is off
  7584. if ($cq =~ /disabled/o or $cq =~ /<command string not enabled>/) {
  7585. add_unknown msg('psa-disabled');
  7586. return;
  7587. }
  7588.  
  7589. ## Detect other cases where pg_stat_activity is not fully populated
  7590. if ($type ne 'qtime' and length $r->{xact_start} and $r->{xact_start} !~ /\d/o) {
  7591. add_unknown msg('psa-noexact');
  7592. return;
  7593. }
  7594.  
  7595. ## Filter out based on the action
  7596. next if $action eq 'txn_idle' and $cq ne '<IDLE> in transaction';
  7597.  
  7598. ## Keep track of the longest overall time
  7599. $maxr = $r if $r->{seconds} >= $maxr->{seconds};
  7600.  
  7601. $count++;
  7602. }
  7603.  
  7604. ## If there were no matches, then there were no rows, or no non-excluded rows
  7605. ## We don't care which at the moment, and return the same message
  7606. if (! $count) {
  7607. $MRTG and do_mrtg({one => 0, msg => $whodunit});
  7608. $db->{perf} = "$perf=0;$wtime;$ctime";
  7609.  
  7610. add_ok msg("$type-none");
  7611. return;
  7612. }
  7613.  
  7614. ## Extract the seconds to avoid typing out the hash each time
  7615. my $max = $maxr->{seconds};
  7616.  
  7617. ## See if we have a minimum number of matches
  7618. my $base_count = $wcount || $ccount;
  7619. if ($base_count and $count < $base_count) {
  7620. $db->{perf} = "$perf=$count;$wcount;$ccount";
  7621. add_ok msg("$type-count-none", $base_count);
  7622. return;
  7623. }
  7624.  
  7625. ## Details on who the top offender was
  7626. if ($max > 0) {
  7627. $whodunit = sprintf q{%s:%s %s:%s %s:%s%s%s},
  7628. msg('PID'), $maxr->{pid},
  7629. msg('database'), $maxr->{datname},
  7630. msg('username'), $maxr->{usename},
  7631. $maxr->{client_addr} eq '' ? '' : (sprintf ' %s:%s', msg('address'), $maxr->{client_addr}),
  7632. ($maxr->{client_port} eq '' or $maxr->{client_port} < 1)
  7633. ? '' : (sprintf ' %s:%s', msg('port'), $maxr->{client_port});
  7634. }
  7635.  
  7636. ## For MRTG, we can simply exit right now
  7637. if ($MRTG) {
  7638. do_mrtg({one => $max, msg => $whodunit});
  7639. exit;
  7640. }
  7641.  
  7642. ## If the number of seconds is high, show an alternate form
  7643. my $ptime = $max > 300 ? ' (' . pretty_time($max) . ')' : '';
  7644.  
  7645. ## Show the maximum number of seconds in the perf section
  7646. $db->{perf} .= sprintf q{%s=%ss;%s;%s},
  7647. $perf,
  7648. $max,
  7649. $wtime,
  7650. $ctime;
  7651.  
  7652. if (length $ctime and length $ccount) {
  7653. if ($max >= $ctime and $count >= $ccount) {
  7654. add_critical msg("$type-for-msg", $count, $ctime, $max, $ptime, $whodunit);
  7655. return;
  7656. }
  7657. }
  7658. elsif (length $ctime) {
  7659. if ($max >= $ctime) {
  7660. add_critical msg("$type-msg", $max, $ptime, $whodunit);
  7661. return;
  7662. }
  7663. }
  7664. elsif (length $ccount) {
  7665. if ($count >= $ccount) {
  7666. add_critical msg("$type-count-msg", $count);
  7667. return;
  7668. }
  7669. }
  7670.  
  7671. if (length $wtime and length $wcount) {
  7672. if ($max >= $wtime and $count >= $wcount) {
  7673. add_warning msg("$type-for-msg", $count, $wtime, $max, $ptime, $whodunit);
  7674. return;
  7675. }
  7676. }
  7677. elsif (length $wtime) {
  7678. if ($max >= $wtime) {
  7679. add_warning msg("$type-msg", $max, $ptime, $whodunit);
  7680. return;
  7681. }
  7682. }
  7683. elsif (length $wcount) {
  7684. if ($count >= $wcount) {
  7685. add_warning msg("$type-count-msg", $count);
  7686. return;
  7687. }
  7688. }
  7689.  
  7690. add_ok msg("$type-msg", $max, $ptime, $whodunit);
  7691.  
  7692. return;
  7693.  
  7694. } ## end of check_txn_idle
  7695.  
  7696.  
  7697. sub check_txn_time {
  7698.  
  7699. ## This is the same as check_txn_idle, but we want where the
  7700. ## transaction start time is not null
  7701.  
  7702. check_txn_idle('txntime',
  7703. '',
  7704. '',
  7705. 'xact_start',
  7706. q{xact_start IS NOT NULL});
  7707.  
  7708. return;
  7709.  
  7710. } ## end of check_txn_time
  7711.  
  7712.  
  7713. sub check_txn_wraparound {
  7714.  
  7715. ## Check how close to transaction wraparound we are on all databases
  7716. ## Supports: Nagios, MRTG
  7717. ## Warning and critical are the number of transactions performed
  7718. ## Thus, anything *over* that number will trip the alert
  7719. ## See: http://www.postgresql.org/docs/current/static/routine-vacuuming.html#VACUUM-FOR-WRAPAROUND
  7720. ## It makes no sense to run this more than once on the same cluster
  7721.  
  7722. my ($warning, $critical) = validate_range
  7723. ({
  7724. type => 'positive integer',
  7725. default_warning => 1_300_000_000,
  7726. default_critical => 1_400_000_000,
  7727. });
  7728.  
  7729. if ($warning and $warning >= 2_000_000_000) {
  7730. ndie msg('txnwrap-wbig');
  7731. }
  7732. if ($critical and $critical >= 2_000_000_000) {
  7733. ndie msg('txnwrap-cbig');
  7734. }
  7735.  
  7736. $SQL = q{SELECT datname, age(datfrozenxid) AS age FROM pg_database WHERE datallowconn ORDER BY 1, 2};
  7737. my $info = run_command($SQL, { regex => qr[\w+\s+\|\s+\d+] } );
  7738.  
  7739. my ($mrtgmax,$mrtgmsg) = (0,'?');
  7740. for $db (@{$info->{db}}) {
  7741. my ($max,$msg) = (0,'?');
  7742. for my $r (@{$db->{slurp}}) {
  7743. my ($dbname,$dbtxns) = ($r->{datname},$r->{age});
  7744. $db->{perf} .= sprintf ' %s=%s;%s;%s;%s;%s',
  7745. perfname($dbname), $dbtxns, $warning, $critical, 0, 2000000000;
  7746. next SLURP if skip_item($dbname);
  7747. if ($dbtxns > $max) {
  7748. $max = $dbtxns;
  7749. $msg = qq{$dbname: $dbtxns};
  7750. if ($dbtxns > $mrtgmax) {
  7751. $mrtgmax = $dbtxns;
  7752. $mrtgmsg = "DB: $dbname";
  7753. }
  7754. }
  7755. }
  7756. if (length $critical and $max >= $critical) {
  7757. add_critical $msg;
  7758. }
  7759. elsif (length $warning and $max >= $warning) {
  7760. add_warning $msg;
  7761. }
  7762. else {
  7763. add_ok $msg;
  7764. }
  7765. }
  7766. $MRTG and do_mrtg({one => $mrtgmax, msg => $mrtgmsg});
  7767.  
  7768. return;
  7769.  
  7770. } ## end of check_txn_wraparound
  7771.  
  7772.  
  7773. sub check_version {
  7774.  
  7775. ## Compare version with what we think it should be
  7776. ## Supports: Nagios, MRTG
  7777. ## Warning and critical are the major and minor (e.g. 8.3)
  7778. ## or the major, minor, and revision (e.g. 8.2.4 or even 8.3beta4)
  7779.  
  7780. if ($MRTG) {
  7781. if (!exists $opt{mrtg} or $opt{mrtg} !~ /^\d+\.\d+/) {
  7782. ndie msg('version-badmrtg');
  7783. }
  7784. if ($opt{mrtg} =~ /^\d+\.\d+$/) {
  7785. $opt{critical} = $opt{mrtg};
  7786. }
  7787. else {
  7788. $opt{warning} = $opt{mrtg};
  7789. }
  7790. }
  7791.  
  7792. my ($warning, $critical) = validate_range({type => 'version', forcemrtg => 1});
  7793.  
  7794. my ($warnfull, $critfull) = (($warning =~ /^\d+\.\d+$/ ? 0 : 1),($critical =~ /^\d+\.\d+$/ ? 0 : 1));
  7795.  
  7796. my $info = run_command('SELECT version() AS version');
  7797.  
  7798. for $db (@{$info->{db}}) {
  7799. my $row = $db->{slurp}[0];
  7800. if ($row->{version} !~ /((\d+\.\d+)(\w+|\.\d+))/o) {
  7801. add_unknown msg('invalid-query', $row->{version});
  7802. next;
  7803. }
  7804. my ($full,$version,$revision) = ($1,$2,$3||'?');
  7805. $revision =~ s/^\.//;
  7806.  
  7807. my $ok = 1;
  7808.  
  7809. if (length $critical) {
  7810. if (($critfull and $critical ne $full)
  7811. or (!$critfull and $critical ne $version)) {
  7812. $MRTG and do_mrtg({one => 0, msg => $full});
  7813. add_critical msg('version-fail', $full, $critical);
  7814. $ok = 0;
  7815. }
  7816. }
  7817. elsif (length $warning) {
  7818. if (($warnfull and $warning ne $full)
  7819. or (!$warnfull and $warning ne $version)) {
  7820. $MRTG and do_mrtg({one => 0, msg => $full});
  7821. add_warning msg('version-fail', $full, $warning);
  7822. $ok = 0;
  7823. }
  7824. }
  7825. if ($ok) {
  7826. $MRTG and do_mrtg({one => 1, msg => $full});
  7827. add_ok msg('version-ok', $full);
  7828. }
  7829. }
  7830.  
  7831. return;
  7832.  
  7833. } ## end of check_version
  7834.  
  7835.  
  7836. sub check_wal_files {
  7837.  
  7838. ## Check on the number of WAL, or WAL "ready", files in use
  7839. ## Supports: Nagios, MRTG
  7840. ## Must run as a superuser
  7841. ## Critical and warning are the number of files
  7842. ## Example: --critical=40
  7843.  
  7844. my $subdir = shift || '';
  7845. my $extrabit = shift || '';
  7846.  
  7847. my $default_warning = shift || 10;
  7848. my $default_critical = shift || 15;
  7849.  
  7850. my $arg = {type => 'positive integer', leastone => 1};
  7851. if ($default_warning) {
  7852. $arg->{default_warning} = $default_warning;
  7853. }
  7854. if ($default_critical) {
  7855. $arg->{default_critical} = $default_critical;
  7856. }
  7857.  
  7858. my ($warning, $critical) = validate_range($arg);
  7859.  
  7860. ## Figure out where the pg_xlog directory is
  7861. $SQL = qq{SELECT count(*) AS count FROM pg_ls_dir('pg_xlog$subdir') WHERE pg_ls_dir ~ E'^[0-9A-F]{24}$extrabit\$'}; ## no critic (RequireInterpolationOfMetachars)
  7862.  
  7863. my $info = run_command($SQL, {regex => qr[\d] });
  7864.  
  7865. my $found = 0;
  7866. for $db (@{$info->{db}}) {
  7867. my $r = $db->{slurp}[0];
  7868. my $numfiles = $r->{count};
  7869. if ($MRTG) {
  7870. do_mrtg({one => $numfiles});
  7871. }
  7872. my $msg = $extrabit ? msg('wal-numfound2', $numfiles, $extrabit)
  7873. : msg('wal-numfound', $numfiles);
  7874. $db->{perf} .= sprintf '%s=%s;%s;%s',
  7875. perfname(msg('files')), $numfiles, $warning, $critical;
  7876. if (length $critical and $numfiles > $critical) {
  7877. add_critical $msg;
  7878. }
  7879. elsif (length $warning and $numfiles > $warning) {
  7880. add_warning $msg;
  7881. }
  7882. else {
  7883. add_ok $msg;
  7884. }
  7885. }
  7886.  
  7887. return;
  7888.  
  7889. } ## end of check_wal_files
  7890.  
  7891.  
  7892.  
  7893. =pod
  7894.  
  7895. =head1 NAME
  7896.  
  7897. B<check_postgres.pl> - a Postgres monitoring script for Nagios, MRTG, Cacti, and others
  7898.  
  7899. This documents describes check_postgres.pl version 2.20.0
  7900.  
  7901. =head1 SYNOPSIS
  7902.  
  7903. ## Create all symlinks
  7904. check_postgres.pl --symlinks
  7905.  
  7906. ## Check connection to Postgres database 'pluto':
  7907. check_postgres.pl --action=connection --db=pluto
  7908.  
  7909. ## Same things, but using the symlink
  7910. check_postgres_connection --db=pluto
  7911.  
  7912. ## Warn if > 100 locks, critical if > 200, or > 20 exclusive
  7913. check_postgres_locks --warning=100 --critical="total=200:exclusive=20"
  7914.  
  7915. ## Show the current number of idle connections on port 6543:
  7916. check_postgres_txn_idle --port=6543 --output=simple
  7917.  
  7918. ## There are many other actions and options, please keep reading.
  7919.  
  7920. The latest news and documentation can always be found at:
  7921. http://bucardo.org/check_postgres/
  7922.  
  7923. =head1 DESCRIPTION
  7924.  
  7925. check_postgres.pl is a Perl script that runs many different tests against
  7926. one or more Postgres databases. It uses the psql program to gather the
  7927. information, and outputs the results in one of three formats: Nagios, MRTG,
  7928. or simple.
  7929.  
  7930. =head2 Output Modes
  7931.  
  7932. The output can be changed by use of the C<--output> option. The default output
  7933. is nagios, although this can be changed at the top of the script if you wish. The
  7934. current option choices are B<nagios>, B<mrtg>, and B<simple>. To avoid having to
  7935. enter the output argument each time, the type of output is automatically set
  7936. if no --output argument is given, and if the current directory has one of the
  7937. output options in its name. For example, creating a directory named mrtg and
  7938. populating it with symlinks via the I<--symlinks> argument would ensure that
  7939. any actions run from that directory will always default to an output of "mrtg"
  7940. As a shortcut for --output=simple, you can enter --simple, which also overrides
  7941. the directory naming trick.
  7942.  
  7943.  
  7944. =head3 Nagios output
  7945.  
  7946. The default output format is for Nagios, which is a single line of information, along
  7947. with four specific exit codes:
  7948.  
  7949. =over 2
  7950.  
  7951. =item 0 (OK)
  7952.  
  7953. =item 1 (WARNING)
  7954.  
  7955. =item 2 (CRITICAL)
  7956.  
  7957. =item 3 (UNKNOWN)
  7958.  
  7959. =back
  7960.  
  7961. The output line is one of the words above, a colon, and then a short description of what
  7962. was measured. Additional statistics information, as well as the total time the command
  7963. took, can be output as well: see the documentation on the arguments
  7964. I<L<--showperf|/--showperf=VAL>>,
  7965. I<L<--perflimit|/--perflimit=i>>, and
  7966. I<L<--showtime|/--showtime=VAL>>.
  7967.  
  7968. =head3 MRTG output
  7969.  
  7970. The MRTG output is four lines, with the first line always giving a single number of importance.
  7971. When possible, this number represents an actual value such as a number of bytes, but it
  7972. may also be a 1 or a 0 for actions that only return "true" or "false", such as check_postgres_version.
  7973. The second line is an additional stat and is only used for some actions. The third line indicates
  7974. an "uptime" and is not used. The fourth line is a description and usually indicates the name of
  7975. the database the stat from the first line was pulled from, but may be different depending on the
  7976. action.
  7977.  
  7978. Some actions accept an optional I<--mrtg> argument to further control the output.
  7979.  
  7980. See the documentation on each action for details on the exact MRTG output for each one.
  7981.  
  7982. =head3 Simple output
  7983.  
  7984. The simple output is simply a truncated version of the MRTG one, and simply returns the first number
  7985. and nothing else. This is very useful when you just want to check the state of something, regardless
  7986. of any threshold. You can transform the numeric output by appending KB, MB, GB, TB, or EB to the output
  7987. argument, for example:
  7988.  
  7989. --output=simple,MB
  7990.  
  7991. =head3 Cacti output
  7992.  
  7993. The Cacti output consists of one or more items on the same line, with a simple name, a colon, and
  7994. then a number. At the moment, the only action with explicit Cacti output is 'dbstats', and using
  7995. the --output option is not needed in this case, as Cacti is the only output for this action. For many
  7996. other actions, using --simple is enough to make Cacti happy.
  7997.  
  7998. =head1 DATABASE CONNECTION OPTIONS
  7999.  
  8000. All actions accept a common set of database options.
  8001.  
  8002. =over 4
  8003.  
  8004. =item B<-H NAME> or B<--host=NAME>
  8005.  
  8006. Connect to the host indicated by NAME. Can be a comma-separated list of names. Multiple host arguments
  8007. are allowed. If no host is given, defaults to the C<PGHOST> environment variable or no host at all
  8008. (which indicates using a local Unix socket). You may also use "--dbhost".
  8009.  
  8010. =item B<-p PORT> or B<--port=PORT>
  8011.  
  8012. Connects using the specified PORT number. Can be a comma-separated list of port numbers, and multiple
  8013. port arguments are allowed. If no port number is given, defaults to the C<PGPORT> environment variable. If
  8014. that is not set, it defaults to 5432. You may also use "--dbport"
  8015.  
  8016. =item B<-db NAME> or B<--dbname=NAME>
  8017.  
  8018. Specifies which database to connect to. Can be a comma-separated list of names, and multiple dbname
  8019. arguments are allowed. If no dbname option is provided, defaults to the C<PGDATABASE> environment variable.
  8020. If that is not set, it defaults to 'postgres' if psql is version 8 or greater, and 'template1' otherwise.
  8021.  
  8022. =item B<-u USERNAME> or B<--dbuser=USERNAME>
  8023.  
  8024. The name of the database user to connect as. Can be a comma-separated list of usernames, and multiple
  8025. dbuser arguments are allowed. If this is not provided, it defaults to the C<PGUSER> environment variable, otherwise
  8026. it defaults to 'postgres'.
  8027.  
  8028. =item B<--dbpass=PASSWORD>
  8029.  
  8030. Provides the password to connect to the database with. Use of this option is highly discouraged.
  8031. Instead, one should use a .pgpass or pg_service.conf file.
  8032.  
  8033. =item B<--dbservice=NAME>
  8034.  
  8035. The name of a service inside of the pg_service.conf file. Before version 9.0 of Postgres, this is
  8036. a global file, usually found in /etc/pg_service.conf. If you are using version 9.0 or higher of
  8037. Postgres, you can use the file ".pg_service.conf" in the home directory of the user running
  8038. the script, e.g. nagios.
  8039.  
  8040. This file contains a simple list of connection options. You can also pass additional information
  8041. when using this option such as --dbservice="maindatabase sslmode=require"
  8042.  
  8043. The documentation for this file can be found at
  8044. http://www.postgresql.org/docs/current/static/libpq-pgservice.html
  8045.  
  8046. =back
  8047.  
  8048. The database connection options can be grouped: I<--host=a,b --host=c --port=1234 --port=3344>
  8049. would connect to a-1234, b-1234, and c-3344. Note that once set, an option
  8050. carries over until it is changed again.
  8051.  
  8052. Examples:
  8053.  
  8054. --host=a,b --port=5433 --db=c
  8055. Connects twice to port 5433, using database c, to hosts a and b: a-5433-c b-5433-c
  8056.  
  8057. --host=a,b --port=5433 --db=c,d
  8058. Connects four times: a-5433-c a-5433-d b-5433-c b-5433-d
  8059.  
  8060. --host=a,b --host=foo --port=1234 --port=5433 --db=e,f
  8061. Connects six times: a-1234-e a-1234-f b-1234-e b-1234-f foo-5433-e foo-5433-f
  8062.  
  8063. --host=a,b --host=x --port=5432,5433 --dbuser=alice --dbuser=bob -db=baz
  8064. Connects three times: a-5432-alice-baz b-5433-alice-baz x-5433-bob-baz
  8065.  
  8066. --dbservice="foo" --port=5433
  8067. Connects using the named service 'foo' in the pg_service.conf file, but overrides the port
  8068.  
  8069. =head1 OTHER OPTIONS
  8070.  
  8071. Other options include:
  8072.  
  8073. =over 4
  8074.  
  8075. =item B<--action=NAME>
  8076.  
  8077. States what action we are running. Required unless using a symlinked file,
  8078. in which case the name of the file is used to figure out the action.
  8079.  
  8080. =item B<--warning=VAL or -w VAL>
  8081.  
  8082. Sets the threshold at which a warning alert is fired. The valid options for this
  8083. option depends on the action used.
  8084.  
  8085. =item B<--critical=VAL or -c VAL>
  8086.  
  8087. Sets the threshold at which a critical alert is fired. The valid options for this
  8088. option depends on the action used.
  8089.  
  8090. =item B<-t VAL> or B<--timeout=VAL>
  8091.  
  8092. Sets the timeout in seconds after which the script will abort whatever it is doing
  8093. and return an UNKNOWN status. The timeout is per Postgres cluster, not for the entire
  8094. script. The default value is 10; the units are always in seconds.
  8095.  
  8096. =item B<--assume-standby-mode>
  8097.  
  8098. If specified, first the check if server in standby mode will be performed
  8099. (--datadir is required), if so, all checks that require SQL queries will be
  8100. ignored and "Server in standby mode" with OK status will be returned instead.
  8101.  
  8102. Example:
  8103.  
  8104. postgres@db$./check_postgres.pl --action=version --warning=8.1 --datadir /var/lib/postgresql/8.3/main/ --assume-standby-mode
  8105. POSTGRES_VERSION OK: Server in standby mode | time=0.00
  8106.  
  8107. =item B<--assume-prod>
  8108.  
  8109. If specified, check if server in production mode is performed (--datadir is required).
  8110. The option is only relevant for (C<symlink: check_postgres_checkpoint>).
  8111.  
  8112. Example:
  8113.  
  8114. postgres@db$./check_postgres.pl --action=checkpoint --datadir /var/lib/postgresql/8.3/main/ --assume-prod
  8115. POSTGRES_CHECKPOINT OK: Last checkpoint was 72 seconds ago | age=72;;300 mode=MASTER
  8116.  
  8117. =item B<-h> or B<--help>
  8118.  
  8119. Displays a help screen with a summary of all actions and options.
  8120.  
  8121. =item B<--man>
  8122.  
  8123. Displays the entire manual.
  8124.  
  8125. =item B<-V> or B<--version>
  8126.  
  8127. Shows the current version.
  8128.  
  8129. =item B<-v> or B<--verbose>
  8130.  
  8131. Set the verbosity level. Can call more than once to boost the level. Setting it to three
  8132. or higher (in other words, issuing C<-v -v -v>) turns on debugging information for this
  8133. program which is sent to stderr.
  8134.  
  8135. =item B<--showperf=VAL>
  8136.  
  8137. Determines if we output additional performance data in standard Nagios format
  8138. (at end of string, after a pipe symbol, using name=value).
  8139. VAL should be 0 or 1. The default is 1. Only takes effect if using Nagios output mode.
  8140.  
  8141. =item B<--perflimit=i>
  8142.  
  8143. Sets a limit as to how many items of interest are reported back when using the
  8144. I<showperf> option. This only has an effect for actions that return a large
  8145. number of items, such as B<table_size>. The default is 0, or no limit. Be
  8146. careful when using this with the I<--include> or I<--exclude> options, as
  8147. those restrictions are done I<after> the query has been run, and thus your
  8148. limit may not include the items you want. Only takes effect if using Nagios output mode.
  8149.  
  8150. =item B<--showtime=VAL>
  8151.  
  8152. Determines if the time taken to run each query is shown in the output. VAL
  8153. should be 0 or 1. The default is 1. No effect unless I<showperf> is on.
  8154. Only takes effect if using Nagios output mode.
  8155.  
  8156. =item B<--test>
  8157.  
  8158. Enables test mode. See the L</"TEST MODE"> section below.
  8159.  
  8160. =item B<--PGBINDIR=PATH>
  8161.  
  8162. Tells the script where to find the psql binaries. Useful if you have more than
  8163. one version of the PostgreSQL executables on your system, or if there are not
  8164. in your path. Note that this option is in all uppercase. By default, this option
  8165. is I<not allowed>. To enable it, you must change the C<$NO_PSQL_OPTION> near the
  8166. top of the script to 0. Avoid using this option if you can, and instead use
  8167. environment variable c<PGBINDIR> or hard-coded C<$PGBINDIR> variable, also near
  8168. the top of the script, to set the path to the PostgreSQL to use.
  8169.  
  8170. =item B<--PSQL=PATH>
  8171.  
  8172. I<(deprecated, this option may be removed in a future release!)>
  8173. Tells the script where to find the psql program. Useful if you have more than
  8174. one version of the psql executable on your system, or if there is no psql program
  8175. in your path. Note that this option is in all uppercase. By default, this option
  8176. is I<not allowed>. To enable it, you must change the C<$NO_PSQL_OPTION> near the
  8177. top of the script to 0. Avoid using this option if you can, and instead hard-code
  8178. your psql location into the C<$PSQL> variable, also near the top of the script.
  8179.  
  8180. =item B<--symlinks>
  8181.  
  8182. Creates symlinks to the main program for each action.
  8183.  
  8184. =item B<--output=VAL>
  8185.  
  8186. Determines the format of the output, for use in various programs. The
  8187. default is 'nagios'. Available options are 'nagios', 'mrtg', 'simple'
  8188. and 'cacti'.
  8189.  
  8190. =item B<--mrtg=VAL>
  8191.  
  8192. Used only for the MRTG or simple output, for a few specific actions.
  8193.  
  8194. =item B<--debugoutput=VAL>
  8195.  
  8196. Outputs the exact string returned by psql, for use in debugging. The value is one or more letters,
  8197. which determine if the output is displayed or not, where 'a' = all, 'c' = critical, 'w' = warning,
  8198. 'o' = ok, and 'u' = unknown. Letters can be combined.
  8199.  
  8200. =item B<--get_method=VAL>
  8201.  
  8202. Allows specification of the method used to fetch information for the C<new_version_cp>,
  8203. C<new_version_pg>, C<new_version_bc>, C<new_version_box>, and C<new_version_tnm> checks.
  8204. The following programs are tried, in order, to grab the information from the web:
  8205. GET, wget, fetch, curl, lynx, links. To force the use of just one (and thus remove the
  8206. overhead of trying all the others until one of those works), enter one of the names as
  8207. the argument to get_method. For example, a BSD box might enter the following line in
  8208. their C<.check_postgresrc> file:
  8209.  
  8210. get_method=fetch
  8211.  
  8212. =item B<--language=VAL>
  8213.  
  8214. Set the language to use for all output messages. Normally, this is detected by examining
  8215. the environment variables LC_ALL, LC_MESSAGES, and LANG, but setting this option
  8216. will override any such detection.
  8217.  
  8218. =back
  8219.  
  8220.  
  8221. =head1 ACTIONS
  8222.  
  8223. The script runs one or more actions. This can either be done with the --action
  8224. flag, or by using a symlink to the main file that contains the name of the action
  8225. inside of it. For example, to run the action "timesync", you may either issue:
  8226.  
  8227. check_postgres.pl --action=timesync
  8228.  
  8229. or use a program named:
  8230.  
  8231. check_postgres_timesync
  8232.  
  8233. All the symlinks are created for you in the current directory
  8234. if use the option --symlinks
  8235.  
  8236. perl check_postgres.pl --symlinks
  8237.  
  8238. If the file name already exists, it will not be overwritten. If the file exists
  8239. and is a symlink, you can force it to overwrite by using "--action=build_symlinks_force"
  8240.  
  8241. Most actions take a I<--warning> and a I<--critical> option, indicating at what
  8242. point we change from OK to WARNING, and what point we go to CRITICAL. Note that
  8243. because criticals are always checked first, setting the warning equal to the
  8244. critical is an effective way to turn warnings off and always give a critical.
  8245.  
  8246. The current supported actions are:
  8247.  
  8248. =head2 B<archive_ready>
  8249.  
  8250. (C<symlink: check_postgres_archive_ready>) Checks how many WAL files with extension F<.ready>
  8251. exist in the F<pg_xlog/archive_status> directory, which is found
  8252. off of your B<data_directory>. This action must be run as a superuser, in order to access the
  8253. contents of the F<pg_xlog/archive_status> directory. The minimum version to use this action is
  8254. Postgres 8.1. The I<--warning> and I<--critical> options are simply the number of
  8255. F<.ready> files in the F<pg_xlog/archive_status> directory.
  8256. Usually, these values should be low, turning on the archive mechanism, we usually want it to
  8257. archive WAL files as fast as possible.
  8258.  
  8259. If the archive command fail, number of WAL in your F<pg_xlog> directory will grow until
  8260. exhausting all the disk space and force PostgreSQL to stop immediately.
  8261.  
  8262. Example 1: Check that the number of ready WAL files is 10 or less on host "pluto"
  8263.  
  8264. check_postgres_archive_ready --host=pluto --critical=10
  8265.  
  8266. For MRTG output, reports the number of ready WAL files on line 1.
  8267.  
  8268. =head2 B<autovac_freeze>
  8269.  
  8270. (C<symlink: check_postgres_autovac_freeze>) Checks how close each database is to the Postgres B<autovacuum_freeze_max_age> setting. This
  8271. action will only work for databases version 8.2 or higher. The I<--warning> and
  8272. I<--critical> options should be expressed as percentages. The 'age' of the transactions
  8273. in each database is compared to the autovacuum_freeze_max_age setting (200 million by default)
  8274. to generate a rounded percentage. The default values are B<90%> for the warning and B<95%> for
  8275. the critical. Databases can be filtered by use of the I<--include> and I<--exclude> options.
  8276. See the L</"BASIC FILTERING"> section for more details.
  8277.  
  8278. Example 1: Give a warning when any databases on port 5432 are above 97%
  8279.  
  8280. check_postgres_autovac_freeze --port=5432 --warning="97%"
  8281.  
  8282. For MRTG output, the highest overall percentage is reported on the first line, and the highest age is
  8283. reported on the second line. All databases which have the percentage from the first line are reported
  8284. on the fourth line, separated by a pipe symbol.
  8285.  
  8286. =head2 B<backends>
  8287.  
  8288. (C<symlink: check_postgres_backends>) Checks the current number of connections for one or more databases, and optionally
  8289. compares it to the maximum allowed, which is determined by the
  8290. Postgres configuration variable B<max_connections>. The I<--warning> and
  8291. I<--critical> options can take one of three forms. First, a simple number can be
  8292. given, which represents the number of connections at which the alert will be
  8293. given. This choice does not use the B<max_connections> setting. Second, the
  8294. percentage of available connections can be given. Third, a negative number can
  8295. be given which represents the number of connections left until B<max_connections>
  8296. is reached. The default values for I<--warning> and I<--critical> are '90%' and '95%'.
  8297. You can also filter the databases by use of the I<--include> and I<--exclude> options.
  8298. See the L</"BASIC FILTERING"> section for more details.
  8299.  
  8300. To view only non-idle processes, you can use the I<--noidle> argument. Note that the
  8301. user you are connecting as must be a superuser for this to work properly.
  8302.  
  8303. Example 1: Give a warning when the number of connections on host quirm reaches 120, and a critical if it reaches 150.
  8304.  
  8305. check_postgres_backends --host=quirm --warning=120 --critical=150
  8306.  
  8307. Example 2: Give a critical when we reach 75% of our max_connections setting on hosts lancre or lancre2.
  8308.  
  8309. check_postgres_backends --warning='75%' --critical='75%' --host=lancre,lancre2
  8310.  
  8311. Example 3: Give a warning when there are only 10 more connection slots left on host plasmid, and a critical
  8312. when we have only 5 left.
  8313.  
  8314. check_postgres_backends --warning=-10 --critical=-5 --host=plasmid
  8315.  
  8316. Example 4: Check all databases except those with "test" in their name, but allow ones that are named "pg_greatest". Connect as port 5432 on the first two hosts, and as port 5433 on the third one. We want to always throw a critical when we reach 30 or more connections.
  8317.  
  8318. check_postgres_backends --dbhost=hong,kong --dbhost=fooey --dbport=5432 --dbport=5433 --warning=30 --critical=30 --exclude="~test" --include="pg_greatest,~prod"
  8319.  
  8320. For MRTG output, the number of connections is reported on the first line, and the fourth line gives the name of the database,
  8321. plus the current maximum_connections. If more than one database has been queried, the one with the highest number of
  8322. connections is output.
  8323.  
  8324. =head2 B<bloat>
  8325.  
  8326. (C<symlink: check_postgres_bloat>) Checks the amount of bloat in tables and indexes. (Bloat is generally the amount
  8327. of dead unused space taken up in a table or index. This space is usually reclaimed
  8328. by use of the VACUUM command.) This action requires that stats collection be
  8329. enabled on the target databases, and requires that ANALYZE is run frequently.
  8330. The I<--include> and I<--exclude> options can be used to filter out which tables
  8331. to look at. See the L</"BASIC FILTERING"> section for more details.
  8332.  
  8333. The I<--warning> and I<--critical> options can be specified as sizes, percents, or both.
  8334. Valid size units are bytes, kilobytes, megabytes, gigabytes, terabytes, exabytes,
  8335. petabytes, and zettabytes. You can abbreviate all of those with the first letter. Items
  8336. without units are assumed to be 'bytes'. The default values are '1 GB' and '5 GB'. The value
  8337. represents the number of "wasted bytes", or the difference between what is actually
  8338. used by the table and index, and what we compute that it should be.
  8339.  
  8340. Note that this action has two hard-coded values to avoid false alarms on
  8341. smaller relations. Tables must have at least 10 pages, and indexes at least 15,
  8342. before they can be considered by this test. If you really want to adjust these
  8343. values, you can look for the variables I<$MINPAGES> and I<$MINIPAGES> at the top of the
  8344. C<check_bloat> subroutine. These values are ignored if either I<--exclude> or
  8345. I<--include> is used.
  8346.  
  8347. Only the top 10 most bloated relations are shown. You can change this number by
  8348. using the I<--perflimit> option to set your own limit.
  8349.  
  8350. The schema named 'information_schema' is excluded from this test, as the only tables
  8351. it contains are small and do not change.
  8352.  
  8353. Please note that the values computed by this action are not precise, and
  8354. should be used as a guideline only. Great effort was made to estimate the
  8355. correct size of a table, but in the end it is only an estimate. The correct
  8356. index size is even more of a guess than the correct table size, but both
  8357. should give a rough idea of how bloated things are.
  8358.  
  8359. Example 1: Warn if any table on port 5432 is over 100 MB bloated, and critical if over 200 MB
  8360.  
  8361. check_postgres_bloat --port=5432 --warning='100 M' --critical='200 M'
  8362.  
  8363. Example 2: Give a critical if table 'orders' on host 'sami' has more than 10 megs of bloat
  8364.  
  8365. check_postgres_bloat --host=sami --include=orders --critical='10 MB'
  8366.  
  8367. Example 3: Give a critical if table 'q4' on database 'sales' is over 50% bloated
  8368.  
  8369. check_postgres_bloat --db=sales --include=q4 --critical='50%'
  8370.  
  8371. Example 4: Give a critical any table is over 20% bloated I<and> has over 150
  8372. MB of bloat:
  8373.  
  8374. check_postgres_bloat --port=5432 --critical='20% and 150 M'
  8375.  
  8376. Example 5: Give a critical any table is over 40% bloated I<or> has over 500 MB
  8377. of bloat:
  8378.  
  8379. check_postgres_bloat --port=5432 --warning='500 M or 40%'
  8380.  
  8381. For MRTG output, the first line gives the highest number of wasted bytes for the tables, and the
  8382. second line gives the highest number of wasted bytes for the indexes. The fourth line gives the database
  8383. name, table name, and index name information. If you want to output the bloat ratio instead (how many
  8384. times larger the relation is compared to how large it should be), just pass in C<--mrtg=ratio>.
  8385.  
  8386. =head2 B<checkpoint>
  8387.  
  8388. (C<symlink: check_postgres_checkpoint>) Determines how long since the last checkpoint has
  8389. been run. This must run on the same server as the database that is being checked (e.g. the -h
  8390. flag will not work). This check is meant to run on a "warm standby" server that is actively
  8391. processing shipped WAL files, and is meant to check that your warm standby is truly 'warm'.
  8392. The data directory must be set, either by the environment variable C<PGDATA>, or passing
  8393. the C<--datadir> argument. It returns the number of seconds since the last checkpoint
  8394. was run, as determined by parsing the call to C<pg_controldata>. Because of this, the
  8395. pg_controldata executable must be available in the current path. Alternatively,
  8396. you can specify C<PGBINDIR> as the directory that it lives in.
  8397. It is also possible to use the special options I<--assume-prod> or
  8398. I<--assume-standby-mode>, if the mode found is not the one expected, a CRITICAL is emitted.
  8399.  
  8400. At least one warning or critical argument must be set.
  8401.  
  8402. This action requires the Date::Parse module.
  8403.  
  8404. For MRTG or simple output, returns the number of seconds.
  8405.  
  8406. =head2 B<cluster_id>
  8407.  
  8408. (C<symlink: check_postgres_cluster-id>) Checks that the Database System Identifier
  8409. provided by pg_controldata is the same as last time you checked. This must run on the same
  8410. server as the database that is being checked (e.g. the -h flag will not work).
  8411. Either the I<--warning> or the I<--critical> option should be given, but not both. The value
  8412. of each one is the cluster identifier, an integer value. You can run with the special C<--critical=0> option
  8413. to find out an existing cluster identifier.
  8414.  
  8415. Example 1: Find the initial identifier
  8416.  
  8417. check_postgres_cluster_id --critical=0 --datadir=/var//lib/postgresql/9.0/main
  8418.  
  8419. Example 2: Make sure the cluster is the same and warn if not, using the result from above.
  8420.  
  8421. check_postgres_cluster_id --critical=5633695740047915135
  8422.  
  8423. For MRTG output, returns a 1 or 0 indicating success of failure of the identifier to match. A
  8424. identifier must be provided as the C<--mrtg> argument. The fourth line always gives the
  8425. current identifier.
  8426.  
  8427. =head2 B<commitratio>
  8428.  
  8429. (C<symlink: check_postgres_commitratio>) Checks the commit ratio of all databases and complains when they are too low.
  8430. There is no need to run this command more than once per database cluster.
  8431. Databases can be filtered with
  8432. the I<--include> and I<--exclude> options. See the L</"BASIC FILTERING"> section
  8433. for more details.
  8434. They can also be filtered by the owner of the database with the
  8435. I<--includeuser> and I<--excludeuser> options.
  8436. See the L</"USER NAME FILTERING"> section for more details.
  8437.  
  8438. The warning and critical options should be specified as percentages. There are not
  8439. defaults for this action: the warning and critical must be specified. The warning value
  8440. cannot be greater than the critical value. The output returns all databases sorted by
  8441. commitratio, smallest first.
  8442.  
  8443. Example: Warn if any database on host flagg is less than 90% in commitratio, and critical if less then 80%.
  8444.  
  8445. check_postgres_database_commitratio --host=flagg --warning='90%' --critical='80%'
  8446.  
  8447. For MRTG output, returns the percentage of the database with the smallest commitratio on the first line,
  8448. and the name of the database on the fourth line.
  8449.  
  8450. =head2 B<connection>
  8451.  
  8452. (C<symlink: check_postgres_connection>) Simply connects, issues a 'SELECT version()', and leaves.
  8453. Takes no I<--warning> or I<--critical> options.
  8454.  
  8455. For MRTG output, simply outputs a 1 (good connection) or a 0 (bad connection) on the first line.
  8456.  
  8457. =head2 B<custom_query>
  8458.  
  8459. (C<symlink: check_postgres_custom_query>) Runs a custom query of your choosing, and parses the results.
  8460. The query itself is passed in through the C<query> argument, and should be kept as simple as possible.
  8461. If at all possible, wrap it in a view or a function to keep things easier to manage. The query should
  8462. return one or two columns. It is required that one of the columns be named "result" and is the item
  8463. that will be checked against your warning and critical values. The second column is for the performance
  8464. data and any name can be used: this will be the 'value' inside the performance data section.
  8465.  
  8466. At least one warning or critical argument must be specified. What these are set to depends on the type of
  8467. query you are running. There are four types of custom_queries that can be run, specified by the C<valtype>
  8468. argument. If none is specified, this action defaults to 'integer'. The four types are:
  8469.  
  8470. B<integer>:
  8471. Does a simple integer comparison. The first column should be a simple integer, and the warning and
  8472. critical values should be the same.
  8473.  
  8474. B<string>:
  8475. The warning and critical are strings, and are triggered only if the value in the first column matches
  8476. it exactly. This is case-sensitive.
  8477.  
  8478. B<time>:
  8479. The warning and the critical are times, and can have units of seconds, minutes, hours, or days.
  8480. Each may be written singular or abbreviated to just the first letter. If no units are given,
  8481. seconds are assumed. The first column should be an integer representing the number of seconds
  8482. to check.
  8483.  
  8484. B<size>:
  8485. The warning and the critical are sizes, and can have units of bytes, kilobytes, megabytes, gigabytes,
  8486. terabytes, or exabytes. Each may be abbreviated to the first letter. If no units are given,
  8487. bytes are assumed. The first column should be an integer representing the number of bytes to check.
  8488.  
  8489. Normally, an alert is triggered if the values returned are B<greater than> or equal to the critical or warning
  8490. value. However, an option of I<--reverse> will trigger the alert if the returned value is
  8491. B<lower than> or equal to the critical or warning value.
  8492.  
  8493. Example 1: Warn if any relation over 100 pages is named "rad", put the number of pages
  8494. inside the performance data section.
  8495.  
  8496. check_postgres_custom_query --valtype=string -w "rad" --query=
  8497. "SELECT relname AS result, relpages AS pages FROM pg_class WHERE relpages > 100"
  8498.  
  8499. Example 2: Give a critical if the "foobar" function returns a number over 5MB:
  8500.  
  8501. check_postgres_custom_query --critical='5MB'--valtype=size --query="SELECT foobar() AS result"
  8502.  
  8503. Example 2: Warn if the function "snazzo" returns less than 42:
  8504.  
  8505. check_postgres_custom_query --critical=42 --query="SELECT snazzo() AS result" --reverse
  8506.  
  8507. If you come up with a useful custom_query, consider sending in a patch to this program
  8508. to make it into a standard action that other people can use.
  8509.  
  8510. This action does not support MRTG or simple output yet.
  8511.  
  8512. =head2 B<database_size>
  8513.  
  8514. (C<symlink: check_postgres_database_size>) Checks the size of all databases and complains when they are too big.
  8515. There is no need to run this command more than once per database cluster.
  8516. Databases can be filtered with
  8517. the I<--include> and I<--exclude> options. See the L</"BASIC FILTERING"> section
  8518. for more details.
  8519. They can also be filtered by the owner of the database with the
  8520. I<--includeuser> and I<--excludeuser> options.
  8521. See the L</"USER NAME FILTERING"> section for more details.
  8522.  
  8523. The warning and critical options can be specified as bytes, kilobytes, megabytes,
  8524. gigabytes, terabytes, or exabytes. Each may be abbreviated to the first letter as well.
  8525. If no unit is given, the units are assumed to be bytes. There are not defaults for this
  8526. action: the warning and critical must be specified. The warning value cannot be greater
  8527. than the critical value. The output returns all databases sorted by size largest first,
  8528. showing both raw bytes and a "pretty" version of the size.
  8529.  
  8530. Example 1: Warn if any database on host flagg is over 1 TB in size, and critical if over 1.1 TB.
  8531.  
  8532. check_postgres_database_size --host=flagg --warning='1 TB' --critical='1.1 t'
  8533.  
  8534. Example 2: Give a critical if the database template1 on port 5432 is over 10 MB.
  8535.  
  8536. check_postgres_database_size --port=5432 --include=template1 --warning='10MB' --critical='10MB'
  8537.  
  8538. Example 3: Give a warning if any database on host 'tardis' owned by the user 'tom' is over 5 GB
  8539.  
  8540. check_postgres_database_size --host=tardis --includeuser=tom --warning='5 GB' --critical='10 GB'
  8541.  
  8542. For MRTG output, returns the size in bytes of the largest database on the first line,
  8543. and the name of the database on the fourth line.
  8544.  
  8545. =head2 B<dbstats>
  8546.  
  8547. (C<symlink: check_postgres_dbstats>) Reports information from the pg_stat_database view,
  8548. and outputs it in a Cacti-friendly manner. No other output is supported, as the output
  8549. is informational and does not lend itself to alerts, such as used with Nagios. If no
  8550. options are given, all databases are returned, one per line. You can include a specific
  8551. database by use of the C<--include> option, or you can use the C<--dbname> option.
  8552.  
  8553. Eleven items are returned on each line, in the format name:value, separated by a single
  8554. space. The items are:
  8555.  
  8556. =over 4
  8557.  
  8558. =item backends
  8559.  
  8560. The number of currently running backends for this database.
  8561.  
  8562. =item commits
  8563.  
  8564. The total number of commits for this database since it was created or reset.
  8565.  
  8566. =item rollbacks
  8567.  
  8568. The total number of rollbacks for this database since it was created or reset.
  8569.  
  8570. =item read
  8571.  
  8572. The total number of disk blocks read.
  8573.  
  8574. =item hit
  8575.  
  8576. The total number of buffer hits.
  8577.  
  8578. =item ret
  8579.  
  8580. The total number of rows returned.
  8581.  
  8582. =item fetch
  8583.  
  8584. The total number of rows fetched.
  8585.  
  8586. =item ins
  8587.  
  8588. The total number of rows inserted.
  8589.  
  8590. =item upd
  8591.  
  8592. The total number of rows updated.
  8593.  
  8594. =item del
  8595.  
  8596. The total number of rows deleted.
  8597.  
  8598. =item dbname
  8599.  
  8600. The name of the database.
  8601.  
  8602. =back
  8603.  
  8604. Note that ret, fetch, ins, upd, and del items will always be 0 if Postgres is version 8.2 or lower, as those stats were
  8605. not available in those versions.
  8606.  
  8607. If the dbname argument is given, seven additional items are returned:
  8608.  
  8609. =over 4
  8610.  
  8611. =item idxscan
  8612.  
  8613. Total number of user index scans.
  8614.  
  8615. =item idxtupread
  8616.  
  8617. Total number of user index entries returned.
  8618.  
  8619. =item idxtupfetch
  8620.  
  8621. Total number of rows fetched by simple user index scans.
  8622.  
  8623. =item idxblksread
  8624.  
  8625. Total number of disk blocks read for all user indexes.
  8626.  
  8627. =item idxblkshit
  8628.  
  8629. Total number of buffer hits for all user indexes.
  8630.  
  8631. =item seqscan
  8632.  
  8633. Total number of sequential scans against all user tables.
  8634.  
  8635. =item seqtupread
  8636.  
  8637. Total number of tuples returned from all user tables.
  8638.  
  8639. =back
  8640.  
  8641. Example 1: Grab the stats for a database named "products" on host "willow":
  8642.  
  8643. check_postgres_dbstats --dbhost willow --dbname products
  8644.  
  8645. The output returned will be like this (all on one line, not wrapped):
  8646.  
  8647. backends:82 commits:58374408 rollbacks:1651 read:268435543 hit:2920381758 idxscan:310931294 idxtupread:2777040927
  8648. idxtupfetch:1840241349 idxblksread:62860110 idxblkshit:1107812216 seqscan:5085305 seqtupread:5370500520
  8649. ret:0 fetch:0 ins:0 upd:0 del:0 dbname:willow
  8650.  
  8651. =head2 B<disabled_triggers>
  8652.  
  8653. (C<symlink: check_postgres_disabled_triggers>) Checks on the number of disabled triggers inside the database.
  8654. The I<--warning> and I<--critical> options are the number of such triggers found, and both
  8655. default to "1", as in normal usage having disabled triggers is a dangerous event. If the
  8656. database being checked is 8.3 or higher, the check is for the number of triggers that are
  8657. in a 'disabled' status (as opposed to being 'always' or 'replica'). The output will show
  8658. the name of the table and the name of the trigger for each disabled trigger.
  8659.  
  8660. Example 1: Make sure that there are no disabled triggers
  8661.  
  8662. check_postgres_disabled_triggers
  8663.  
  8664. For MRTG output, returns the number of disabled triggers on the first line.
  8665.  
  8666. =head2 B<disk_space>
  8667.  
  8668. (C<symlink: check_postgres_disk_space>) Checks on the available physical disk space used by Postgres. This action requires
  8669. that you have the executable "/bin/df" available to report on disk sizes, and it
  8670. also needs to be run as a superuser, so it can examine the B<data_directory>
  8671. setting inside of Postgres. The I<--warning> and I<--critical> options are
  8672. given in either sizes or percentages or both. If using sizes, the standard unit types
  8673. are allowed: bytes, kilobytes, gigabytes, megabytes, gigabytes, terabytes, or
  8674. exabytes. Each may be abbreviated to the first letter only; no units at all
  8675. indicates 'bytes'. The default values are '90%' and '95%'.
  8676.  
  8677. This command checks the following things to determine all of the different
  8678. physical disks being used by Postgres.
  8679.  
  8680. B<data_directory> - The disk that the main data directory is on.
  8681.  
  8682. B<log directory> - The disk that the log files are on.
  8683.  
  8684. B<WAL file directory> - The disk that the write-ahead logs are on (e.g. symlinked pg_xlog)
  8685.  
  8686. B<tablespaces> - Each tablespace that is on a separate disk.
  8687.  
  8688. The output shows the total size used and available on each disk, as well as
  8689. the percentage, ordered by highest to lowest percentage used. Each item above
  8690. maps to a file system: these can be included or excluded. See the
  8691. L</"BASIC FILTERING"> section for more details.
  8692.  
  8693. Example 1: Make sure that no file system is over 90% for the database on port 5432.
  8694.  
  8695. check_postgres_disk_space --port=5432 --warning='90%' --critical='90%'
  8696.  
  8697. Example 2: Check that all file systems starting with /dev/sda are smaller than 10 GB and 11 GB (warning and critical)
  8698.  
  8699. check_postgres_disk_space --port=5432 --warning='10 GB' --critical='11 GB' --include="~^/dev/sda"
  8700.  
  8701. Example 4: Make sure that no file system is both over 50% I<and> has over 15 GB
  8702.  
  8703. check_postgres_disk_space --critical='50% and 15 GB'
  8704.  
  8705. Example 5: Issue a warning if any file system is either over 70% full I<or> has
  8706. more than 1T
  8707.  
  8708. check_postgres_disk_space --warning='1T or 75'
  8709.  
  8710. For MRTG output, returns the size in bytes of the file system on the first line,
  8711. and the name of the file system on the fourth line.
  8712.  
  8713. =head2 B<fsm_pages>
  8714.  
  8715. (C<symlink: check_postgres_fsm_pages>) Checks how close a cluster is to the Postgres B<max_fsm_pages> setting.
  8716. This action will only work for databases of 8.2 or higher, and it requires the contrib
  8717. module B<pg_freespacemap> be installed. The I<--warning> and I<--critical> options should be expressed
  8718. as percentages. The number of used pages in the free-space-map is determined by looking in the
  8719. pg_freespacemap_relations view, and running a formula based on the formula used for
  8720. outputting free-space-map pageslots in the vacuum verbose command. The default values are B<85%> for the
  8721. warning and B<95%> for the critical.
  8722.  
  8723. Example 1: Give a warning when our cluster has used up 76% of the free-space pageslots, with pg_freespacemap installed in database robert
  8724.  
  8725. check_postgres_fsm_pages --dbname=robert --warning="76%"
  8726.  
  8727. While you need to pass in the name of the database where pg_freespacemap is installed, you only need to run this check once per cluster. Also, checking this information does require obtaining special locks on the free-space-map, so it is recommend you do not run this check with short intervals.
  8728.  
  8729. For MRTG output, returns the percent of free-space-map on the first line, and the number of pages currently used on
  8730. the second line.
  8731.  
  8732. =head2 B<fsm_relations>
  8733.  
  8734. (C<symlink: check_postgres_fsm_relations>) Checks how close a cluster is to the Postgres B<max_fsm_relations> setting.
  8735. This action will only work for databases of 8.2 or higher, and it requires the contrib module B<pg_freespacemap> be
  8736. installed. The I<--warning> and I<--critical> options should be expressed as percentages. The number of used relations
  8737. in the free-space-map is determined by looking in the pg_freespacemap_relations view. The default values are B<85%> for
  8738. the warning and B<95%> for the critical.
  8739.  
  8740. Example 1: Give a warning when our cluster has used up 80% of the free-space relations, with pg_freespacemap installed in database dylan
  8741.  
  8742. check_postgres_fsm_relations --dbname=dylan --warning="75%"
  8743.  
  8744. While you need to pass in the name of the database where pg_freespacemap is installed, you only need to run this check
  8745. once per cluster. Also,
  8746. checking this information does require obtaining special locks on the free-space-map, so it is recommend you do not
  8747. run this check with short intervals.
  8748.  
  8749. For MRTG output, returns the percent of free-space-map on the first line, the number of relations currently used on
  8750. the second line.
  8751.  
  8752. =head2 B<hitratio>
  8753.  
  8754. (C<symlink: check_postgres_hitratio>) Checks the hit ratio of all databases and complains when they are too low.
  8755. There is no need to run this command more than once per database cluster.
  8756. Databases can be filtered with
  8757. the I<--include> and I<--exclude> options. See the L</"BASIC FILTERING"> section
  8758. for more details.
  8759. They can also be filtered by the owner of the database with the
  8760. I<--includeuser> and I<--excludeuser> options.
  8761. See the L</"USER NAME FILTERING"> section for more details.
  8762.  
  8763. The warning and critical options should be specified as percentages. There are not
  8764. defaults for this action: the warning and critical must be specified. The warning value
  8765. cannot be greater than the critical value. The output returns all databases sorted by
  8766. hitratio, smallest first.
  8767.  
  8768. Example: Warn if any database on host flagg is less than 90% in hitratio, and critical if less then 80%.
  8769.  
  8770. check_postgres_hitratio --host=flagg --warning='90%' --critical='80%'
  8771.  
  8772. For MRTG output, returns the percentage of the database with the smallest hitratio on the first line,
  8773. and the name of the database on the fourth line.
  8774.  
  8775. =head2 B<hot_standby_delay>
  8776.  
  8777. (C<symlink: check_hot_standby_delay>) Checks the streaming replication lag by computing the delta
  8778. between the xlog position of a master server and the one of the slaves connected to it. The slave_
  8779. server must be in hot_standby (e.g. read only) mode, therefore the minimum version to use this_
  8780. action is Postgres 9.0. The I<--warning> and I<--critical> options are the delta between xlog
  8781. location. These values should match the volume of transactions needed to have the streaming
  8782. replication disconnect from the master because of too much lag.
  8783.  
  8784. You must provide information on how to reach the second database by a connection
  8785. parameter ending in the number 2, such as "--dbport2=5543". If if it not given,
  8786. the action fails.
  8787.  
  8788. =head2 B<index_size>
  8789.  
  8790. =head2 B<table_size>
  8791.  
  8792. =head2 B<relation_size>
  8793.  
  8794. (symlinks: C<check_postgres_index_size>, C<check_postgres_table_size>, and C<check_postgres_relation_size>)
  8795. The actions B<table_size> and B<index_size> are simply variations of the
  8796. B<relation_size> action, which checks for a relation that has grown too big.
  8797. Relations (in other words, tables and indexes) can be filtered with the
  8798. I<--include> and I<--exclude> options. See the L</"BASIC FILTERING"> section
  8799. for more details. Relations can also be filtered by the user that owns them,
  8800. by using the I<--includeuser> and I<--excludeuser> options.
  8801. See the L</"USER NAME FILTERING"> section for more details.
  8802.  
  8803. The values for the I<--warning> and I<--critical> options are file sizes, and
  8804. may have units of bytes, kilobytes, megabytes, gigabytes, terabytes, or exabytes.
  8805. Each can be abbreviated to the first letter. If no units are given, bytes are
  8806. assumed. There are no default values: both the warning and the critical option
  8807. must be given. The return text shows the size of the largest relation found.
  8808.  
  8809. If the I<--showperf> option is enabled, I<all> of the relations with their sizes
  8810. will be given. To prevent this, it is recommended that you set the
  8811. I<--perflimit> option, which will cause the query to do a
  8812. C<ORDER BY size DESC LIMIT (perflimit)>.
  8813.  
  8814. Example 1: Give a critical if any table is larger than 600MB on host burrick.
  8815.  
  8816. check_postgres_table_size --critical='600 MB' --warning='600 MB' --host=burrick
  8817.  
  8818. Example 2: Warn if the table products is over 4 GB in size, and give a critical at 4.5 GB.
  8819.  
  8820. check_postgres_table_size --host=burrick --warning='4 GB' --critical='4.5 GB' --include=products
  8821.  
  8822. Example 3: Warn if any index not owned by postgres goes over 500 MB.
  8823.  
  8824. check_postgres_index_size --port=5432 --excludeuser=postgres -w 500MB -c 600MB
  8825.  
  8826. For MRTG output, returns the size in bytes of the largest relation, and the name of the database
  8827. and relation as the fourth line.
  8828.  
  8829. =head2 B<last_analyze>
  8830.  
  8831. =head2 B<last_vacuum>
  8832.  
  8833. =head2 B<last_autoanalyze>
  8834.  
  8835. =head2 B<last_autovacuum>
  8836.  
  8837. (symlinks: C<check_postgres_last_analyze>, C<check_postgres_last_vacuum>,
  8838. C<check_postgres_last_autoanalyze>, and C<check_postgres_last_autovacuum>)
  8839. Checks how long it has been since vacuum (or analyze) was last run on each
  8840. table in one or more databases. Use of these actions requires that the target
  8841. database is version 8.3 or greater, or that the version is 8.2 and the
  8842. configuration variable B<stats_row_level> has been enabled. Tables can be filtered with the
  8843. I<--include> and I<--exclude> options. See the L</"BASIC FILTERING"> section
  8844. for more details.
  8845. Tables can also be filtered by their owner by use of the
  8846. I<--includeuser> and I<--excludeuser> options.
  8847. See the L</"USER NAME FILTERING"> section for more details.
  8848.  
  8849. The units for I<--warning> and I<--critical> are specified as times.
  8850. Valid units are seconds, minutes, hours, and days; all can be abbreviated
  8851. to the first letter. If no units are given, 'seconds' are assumed. The
  8852. default values are '1 day' and '2 days'. Please note that there are cases
  8853. in which this field does not get automatically populated. If certain tables
  8854. are giving you problems, make sure that they have dead rows to vacuum,
  8855. or just exclude them from the test.
  8856.  
  8857. The schema named 'information_schema' is excluded from this test, as the only tables
  8858. it contains are small and do not change.
  8859.  
  8860. Note that the non-'auto' versions will also check on the auto versions as well. In other words,
  8861. using last_vacuum will report on the last vacuum, whether it was a normal vacuum, or
  8862. one run by the autovacuum daemon.
  8863.  
  8864. Example 1: Warn if any table has not been vacuumed in 3 days, and give a
  8865. critical at a week, for host wormwood
  8866.  
  8867. check_postgres_last_vacuum --host=wormwood --warning='3d' --critical='7d'
  8868.  
  8869. Example 2: Same as above, but skip tables belonging to the users 'eve' or 'mallory'
  8870.  
  8871. check_postgres_last_vacuum --host=wormwood --warning='3d' --critical='7d' --excludeusers=eve,mallory
  8872.  
  8873. For MRTG output, returns (on the first line) the LEAST amount of time in seconds since a table was
  8874. last vacuumed or analyzed. The fourth line returns the name of the database and name of the table.
  8875.  
  8876. =head2 B<listener>
  8877.  
  8878. (C<symlink: check_postgres_listener>) Confirm that someone is listening for one or more
  8879. specific strings (using the LISTEN/NOTIFY system), by looking at the pg_listener table.
  8880. Only one of warning or critical is needed. The format is a simple string representing the
  8881. LISTEN target, or a tilde character followed by a string for a regular expression check.
  8882. Note that this check will not work on versions of Postgres 9.0 or higher.
  8883.  
  8884. Example 1: Give a warning if nobody is listening for the string bucardo_mcp_ping on ports 5555 and 5556
  8885.  
  8886. check_postgres_listener --port=5555,5556 --warning=bucardo_mcp_ping
  8887.  
  8888. Example 2: Give a critical if there are no active LISTEN requests matching 'grimm' on database oskar
  8889.  
  8890. check_postgres_listener --db oskar --critical=~grimm
  8891.  
  8892. For MRTG output, returns a 1 or a 0 on the first, indicating success or failure. The name of the notice must
  8893. be provided via the I<--mrtg> option.
  8894.  
  8895. =head2 B<locks>
  8896.  
  8897. (C<symlink: check_postgres_locks>) Check the total number of locks on one or more databases. There is no
  8898. need to run this more than once per database cluster. Databases can be filtered
  8899. with the I<--include> and I<--exclude> options. See the L</"BASIC FILTERING"> section
  8900. for more details.
  8901.  
  8902. The I<--warning> and I<--critical> options can be specified as simple numbers,
  8903. which represent the total number of locks, or they can be broken down by type of lock.
  8904. Valid lock names are C<'total'>, C<'waiting'>, or the name of a lock type used by Postgres.
  8905. These names are case-insensitive and do not need the "lock" part on the end,
  8906. so B<exclusive> will match 'ExclusiveLock'. The format is name=number, with different
  8907. items separated by colons or semicolons (or any other symbol).
  8908.  
  8909. Example 1: Warn if the number of locks is 100 or more, and critical if 200 or more, on host garrett
  8910.  
  8911. check_postgres_locks --host=garrett --warning=100 --critical=200
  8912.  
  8913. Example 2: On the host artemus, warn if 200 or more locks exist, and give a critical if over 250 total locks exist, or if over 20 exclusive locks exist, or if over 5 connections are waiting for a lock.
  8914.  
  8915. check_postgres_locks --host=artemus --warning=200 --critical="total=250:waiting=5:exclusive=20"
  8916.  
  8917. For MRTG output, returns the number of locks on the first line, and the name of the database on the fourth line.
  8918.  
  8919. =head2 B<logfile>
  8920.  
  8921. (C<symlink: check_postgres_logfile>) Ensures that the logfile is in the expected location and is being logged to.
  8922. This action issues a command that throws an error on each database it is
  8923. checking, and ensures that the message shows up in the logs. It scans the
  8924. various log_* settings inside of Postgres to figure out where the logs should be.
  8925. If you are using syslog, it does a rough (but not foolproof) scan of
  8926. F</etc/syslog.conf>. Alternatively, you can provide the name of the logfile
  8927. with the I<--logfile> option. This is especially useful if the logs have a
  8928. custom rotation scheme driven be an external program. The B<--logfile> option
  8929. supports the following escape characters: C<%Y %m %d %H>, which represent
  8930. the current year, month, date, and hour respectively. An error is always
  8931. reported as critical unless the warning option has been passed in as a non-zero
  8932. value. Other than that specific usage, the C<--warning> and C<--critical>
  8933. options should I<not> be used.
  8934.  
  8935. Example 1: On port 5432, ensure the logfile is being written to the file /home/greg/pg8.2.log
  8936.  
  8937. check_postgres_logfile --port=5432 --logfile=/home/greg/pg8.2.log
  8938.  
  8939. Example 2: Same as above, but raise a warning, not a critical
  8940.  
  8941. check_postgres_logfile --port=5432 --logfile=/home/greg/pg8.2.log -w 1
  8942.  
  8943. For MRTG output, returns a 1 or 0 on the first line, indicating success or failure. In case of a
  8944. failure, the fourth line will provide more detail on the failure encountered.
  8945.  
  8946. =head2 B<new_version_bc>
  8947.  
  8948. (C<symlink: check_postgres_new_version_bc>) Checks if a newer version of the Bucardo
  8949. program is available. The current version is obtained by running C<bucardo_ctl --version>.
  8950. If a major upgrade is available, a warning is returned. If a revision upgrade is
  8951. available, a critical is returned. (Bucardo is a master to slave, and master to master
  8952. replication system for Postgres: see http://bucardo.org for more information).
  8953. See also the information on the C<--get_method> option.
  8954.  
  8955. =head2 B<new_version_box>
  8956.  
  8957. (C<symlink: check_postgres_new_version_box>) Checks if a newer version of the boxinfo
  8958. program is available. The current version is obtained by running C<boxinfo.pl --version>.
  8959. If a major upgrade is available, a warning is returned. If a revision upgrade is
  8960. available, a critical is returned. (boxinfo is a program for grabbing important
  8961. information from a server and putting it into a HTML format: see
  8962. http://bucardo.org/wiki/boxinfo for more information). See also the information on
  8963. the C<--get_method> option.
  8964.  
  8965. =head2 B<new_version_cp>
  8966.  
  8967. (C<symlink: check_postgres_new_version_cp>) Checks if a newer version of this program
  8968. (check_postgres.pl) is available, by grabbing the version from a small text file
  8969. on the main page of the home page for the project. Returns a warning if the returned
  8970. version does not match the one you are running. Recommended interval to check is
  8971. once a day. See also the information on the C<--get_method> option.
  8972.  
  8973. =head2 B<new_version_pg>
  8974.  
  8975. (C<symlink: check_postgres_new_version_pg>) Checks if a newer revision of Postgres
  8976. exists for each database connected to. Note that this only checks for revision, e.g.
  8977. going from 8.3.6 to 8.3.7. Revisions are always 100% binary compatible and involve no
  8978. dump and restore to upgrade. Revisions are made to address bugs, so upgrading as soon
  8979. as possible is always recommended. Returns a warning if you do not have the latest revision.
  8980. It is recommended this check is run at least once a day. See also the information on
  8981. the C<--get_method> option.
  8982.  
  8983.  
  8984. =head2 B<new_version_tnm>
  8985.  
  8986. (C<symlink: check_postgres_new_version_tnm>) Checks if a newer version of the
  8987. tail_n_mail program is available. The current version is obtained by running
  8988. C<tail_n_mail --version>. If a major upgrade is available, a warning is returned. If a
  8989. revision upgrade is available, a critical is returned. (tail_n_mail is a log monitoring
  8990. tool that can send mail when interesting events appear in your Postgres logs.
  8991. See: http://bucardo.org/wiki/Tail_n_mail for more information).
  8992. See also the information on the C<--get_method> option.
  8993.  
  8994. =head2 B<pgb_pool_cl_active>
  8995.  
  8996. =head2 B<pgb_pool_cl_waiting>
  8997.  
  8998. =head2 B<pgb_pool_sv_active>
  8999.  
  9000. =head2 B<pgb_pool_sv_idle>
  9001.  
  9002. =head2 B<pgb_pool_sv_used>
  9003.  
  9004. =head2 B<pgb_pool_sv_tested>
  9005.  
  9006. =head2 B<pgb_pool_sv_login>
  9007.  
  9008. =head2 B<pgb_pool_maxwait>
  9009.  
  9010. (symlinks: C<check_postgres_pgb_pool_cl_active>, C<check_postgres_pgb_pool_cl_waiting>,
  9011. C<check_postgres_pgb_pool_sv_active>, C<check_postgres_pgb_pool_sv_idle>,
  9012. C<check_postgres_pgb_pool_sv_used>, C<check_postgres_pgb_pool_sv_tested>,
  9013. C<check_postgres_pgb_pool_sv_login>, and C<check_postgres_pgb_pool_maxwait>)
  9014.  
  9015. Examines pgbouncer's pool statistics. Each pool has a set of "client"
  9016. connections, referring to connections from external clients, and "server"
  9017. connections, referring to connections to PostgreSQL itself. The related
  9018. check_postgres actions are prefixed by "cl_" and "sv_", respectively. Active
  9019. client connections are those connections currently linked with an active server
  9020. connection. Client connections may also be "waiting", meaning they have not yet
  9021. been allocated a server connection. Server connections are "active" (linked to
  9022. a client), "idle" (standing by for a client connection to link with), "used"
  9023. (just unlinked from a client, and not yet returned to the idle pool), "tested"
  9024. (currently being tested) and "login" (in the process of logging in). The
  9025. maxwait value shows how long in seconds the oldest waiting client connection
  9026. has been waiting.
  9027.  
  9028. =head2 B<pgbouncer_backends>
  9029.  
  9030. (C<symlink: check_postgres_pgbouncer_backends>) Checks the current number of
  9031. connections for one or more databases through pgbouncer, and optionally
  9032. compares it to the maximum allowed, which is determined by the pgbouncer
  9033. configuration variable B<max_client_conn>. The I<--warning> and I<--critical>
  9034. options can take one of three forms. First, a simple number can be given,
  9035. which represents the number of connections at which the alert will be given.
  9036. This choice does not use the B<max_connections> setting. Second, the
  9037. percentage of available connections can be given. Third, a negative number can
  9038. be given which represents the number of connections left until
  9039. B<max_connections> is reached. The default values for I<--warning> and
  9040. I<--critical> are '90%' and '95%'. You can also filter the databases by use
  9041. of the I<--include> and I<--exclude> options. See the L</"BASIC FILTERING">
  9042. section for more details.
  9043.  
  9044. To view only non-idle processes, you can use the I<--noidle> argument. Note
  9045. that the user you are connecting as must be a superuser for this to work
  9046. properly.
  9047.  
  9048. Example 1: Give a warning when the number of connections on host quirm reaches
  9049. 120, and a critical if it reaches 150.
  9050.  
  9051. check_postgres_pgbouncer_backends --host=quirm --warning=120 --critical=150 -p 6432 -u pgbouncer
  9052.  
  9053. Example 2: Give a critical when we reach 75% of our max_connections setting on
  9054. hosts lancre or lancre2.
  9055.  
  9056. check_postgres_pgbouncer_backends --warning='75%' --critical='75%' --host=lancre,lancre2 -p 6432 -u pgbouncer
  9057.  
  9058. Example 3: Give a warning when there are only 10 more connection slots left on
  9059. host plasmid, and a critical when we have only 5 left.
  9060.  
  9061. check_postgres_pgbouncer_backends --warning=-10 --critical=-5 --host=plasmid -p 6432 -u pgbouncer
  9062.  
  9063. For MRTG output, the number of connections is reported on the first line, and
  9064. the fourth line gives the name of the database, plus the current
  9065. max_client_conn. If more than one database has been queried, the one with the
  9066. highest number of connections is output.
  9067.  
  9068. =head2 B<pgbouncer_checksum>
  9069.  
  9070. (C<symlink: check_postgres_pgbouncer_checksum>) Checks that all the
  9071. pgBouncer settings are the same as last time you checked.
  9072. This is done by generating a checksum of a sorted list of setting names and
  9073. their values. Note that you shouldn't specify the database name, it will
  9074. automatically default to pgbouncer. Either the I<--warning> or the I<--critical> option
  9075. should be given, but not both. The value of each one is the checksum, a
  9076. 32-character hexadecimal value. You can run with the special C<--critical=0> option
  9077. to find out an existing checksum.
  9078.  
  9079. This action requires the Digest::MD5 module.
  9080.  
  9081. Example 1: Find the initial checksum for pgbouncer configuration on port 6432 using the default user (usually postgres)
  9082.  
  9083. check_postgres_pgbouncer_checksum --port=6432 --critical=0
  9084.  
  9085. Example 2: Make sure no settings have changed and warn if so, using the checksum from above.
  9086.  
  9087. check_postgres_pgbouncer_checksum --port=6432 --warning=cd2f3b5e129dc2b4f5c0f6d8d2e64231
  9088.  
  9089. For MRTG output, returns a 1 or 0 indicating success of failure of the checksum to match. A
  9090. checksum must be provided as the C<--mrtg> argument. The fourth line always gives the
  9091. current checksum.
  9092.  
  9093. =head2 B<pgagent_jobs>
  9094.  
  9095. (C<symlink: check_postgres_pgagent_jobs>) Checks that all the pgAgent jobs
  9096. that have executed in the preceding interval of time have succeeded. This is
  9097. done by checking for any steps that have a non-zero result.
  9098.  
  9099. Either C<--warning> or C<--critical>, or both, may be specified as times, and
  9100. jobs will be checked for failures withing the specified periods of time before
  9101. the current time. Valid units are seconds, minutes, hours, and days; all can
  9102. be abbreviated to the first letter. If no units are given, 'seconds' are
  9103. assumed.
  9104.  
  9105. Example 1: Give a critical when any jobs executed in the last day have failed.
  9106.  
  9107. check_postgres_pgagent_jobs --critical=1d
  9108.  
  9109. Example 2: Give a warning when any jobs executed in the last week have failed.
  9110.  
  9111. check_postgres_pgagent_jobs --warning=7d
  9112.  
  9113. Example 3: Give a critical for jobs that have failed in the last 2 hours and a
  9114. warning for jobs that have failed in the last 4 hours:
  9115.  
  9116. check_postgres_pgagent_jobs --critical=2h --warning=4h
  9117.  
  9118. =head2 B<prepared_txns>
  9119.  
  9120. (C<symlink: check_postgres_prepared_txns>) Check on the age of any existing prepared transactions.
  9121. Note that most people will NOT use prepared transactions, as they are part of two-part commit
  9122. and complicated to maintain. They should also not be confused with prepared STATEMENTS, which is
  9123. what most people think of when they hear prepare. The default value for a warning is 1 second, to
  9124. detect any use of prepared transactions, which is probably a mistake on most systems. Warning and
  9125. critical are the number of seconds a prepared transaction has been open before an alert is given.
  9126.  
  9127. Example 1: Give a warning on detecting any prepared transactions:
  9128.  
  9129. check_postgres_prepared_txns -w 0
  9130.  
  9131. Example 2: Give a critical if any prepared transaction has been open longer than 10 seconds, but allow
  9132. up to 360 seconds for the database 'shrike':
  9133.  
  9134. check_postgres_prepared_txns --critical=10 --exclude=shrike
  9135. check_postgres_prepared_txns --critical=360 --include=shrike
  9136.  
  9137. For MRTG output, returns the number of seconds the oldest transaction has been open as the first line,
  9138. and which database is came from as the final line.
  9139.  
  9140. =head2 B<query_runtime>
  9141.  
  9142. (C<symlink: check_postgres_query_runtime>) Checks how long a specific query takes to run, by executing a "EXPLAIN ANALYZE"
  9143. against it. The I<--warning> and I<--critical> options are the maximum amount of
  9144. time the query should take. Valid units are seconds, minutes, and hours; any can be
  9145. abbreviated to the first letter. If no units are given, 'seconds' are assumed.
  9146. Both the warning and the critical option must be given. The name of the view or
  9147. function to be run must be passed in to the I<--queryname> option. It must consist
  9148. of a single word (or schema.word), with optional parens at the end.
  9149.  
  9150. Example 1: Give a critical if the function named "speedtest" fails to run in 10 seconds or less.
  9151.  
  9152. check_postgres_query_runtime --queryname='speedtest()' --critical=10 --warning=10
  9153.  
  9154. For MRTG output, reports the time in seconds for the query to complete on the first line. The fourth
  9155. line lists the database.
  9156.  
  9157. =head2 B<query_time>
  9158.  
  9159. (C<symlink: check_postgres_query_time>) Checks the length of running queries on one or more databases.
  9160. There is no need to run this more than once on the same database cluster. Note that
  9161. this already excludes queries that are "idle in transaction". Databases can be filtered
  9162. by using the I<--include> and I<--exclude> options. See the L</"BASIC FILTERING">
  9163. section for more details. You can also filter on the user running the
  9164. query with the I<--includeuser> and I<--excludeuser> options.
  9165. See the L</"USER NAME FILTERING"> section for more details.
  9166.  
  9167. The values for the I<--warning> and I<--critical> options are amounts of
  9168. time, and default to '2 minutes' and '5 minutes' respectively. Valid units
  9169. are 'seconds', 'minutes', 'hours', or 'days'. Each may be written singular or
  9170. abbreviated to just the first letter. If no units are given, the unit is
  9171. assumed to be seconds.
  9172.  
  9173. This action requires Postgres 8.1 or better.
  9174.  
  9175. Example 1: Give a warning if any query has been running longer than 3 minutes, and a critical if longer than 5 minutes.
  9176.  
  9177. check_postgres_query_time --port=5432 --warning='3 minutes' --critical='5 minutes'
  9178.  
  9179. Example 2: Using default values (2 and 5 minutes), check all databases except those starting with 'template'.
  9180.  
  9181. check_postgres_query_time --port=5432 --exclude=~^template
  9182.  
  9183. Example 3: Warn if user 'don' has a query running over 20 seconds
  9184.  
  9185. check_postgres_query_time --port=5432 --includeuser=don --warning=20s
  9186.  
  9187. For MRTG output, returns the length in seconds of the longest running query on the first line. The fourth
  9188. line gives the name of the database.
  9189.  
  9190. =head2 B<replicate_row>
  9191.  
  9192. (C<symlink: check_postgres_replicate_row>) Checks that master-slave replication is working to one or more slaves.
  9193. The slaves are specified the same as the normal databases, except with
  9194. the number 2 at the end of them, so "--port2" instead of "--port", etc.
  9195. The values or the I<--warning> and I<--critical> options are units of time, and
  9196. at least one must be provided (no defaults). Valid units are 'seconds', 'minutes', 'hours',
  9197. or 'days'. Each may be written singular or abbreviated to just the first letter.
  9198. If no units are given, the units are assumed to be seconds.
  9199.  
  9200. This check updates a single row on the master, and then measures how long it
  9201. takes to be applied to the slaves. To do this, you need to pick a table that
  9202. is being replicated, then find a row that can be changed, and is not going
  9203. to be changed by any other process. A specific column of this row will be changed
  9204. from one value to another. All of this is fed to the C<repinfo> option, and should
  9205. contain the following options, separated by commas: table name, primary key, key id,
  9206. column, first value, second value.
  9207.  
  9208. Example 1: Slony is replicating a table named 'orders' from host 'alpha' to
  9209. host 'beta', in the database 'sales'. The primary key of the table is named
  9210. id, and we are going to test the row with an id of 3 (which is historical and
  9211. never changed). There is a column named 'salesrep' that we are going to toggle
  9212. from a value of 'slon' to 'nols' to check on the replication. We want to throw
  9213. a warning if the replication does not happen within 10 seconds.
  9214.  
  9215. check_postgres_replicate_row --host=alpha --dbname=sales --host2=beta
  9216. --dbname2=sales --warning=10 --repinfo=orders,id,3,salesrep,slon,nols
  9217.  
  9218. Example 2: Bucardo is replicating a table named 'receipt' from host 'green'
  9219. to hosts 'red', 'blue', and 'yellow'. The database for both sides is 'public'.
  9220. The slave databases are running on port 5455. The primary key is named 'receipt_id',
  9221. the row we want to use has a value of 9, and the column we want to change for the
  9222. test is called 'zone'. We'll toggle between 'north' and 'south' for the value of
  9223. this column, and throw a critical if the change is not on all three slaves within 5 seconds.
  9224.  
  9225. check_postgres_replicate_row --host=green --port2=5455 --host2=red,blue,yellow
  9226. --critical=5 --repinfo=receipt,receipt_id,9,zone,north,south
  9227.  
  9228. For MRTG output, returns on the first line the time in seconds the replication takes to finish.
  9229. The maximum time is set to 4 minutes 30 seconds: if no replication has taken place in that long
  9230. a time, an error is thrown.
  9231.  
  9232. =head2 B<same_schema>
  9233.  
  9234. (C<symlink: check_postgres_same_schema>) Verifies that two or more databases are identical as far as their
  9235. schema (but not the data within). This is particularly handy for making sure your slaves have not
  9236. been modified or corrupted in any way when using master to slave replication. Unlike most other
  9237. actions, this has no warning or critical criteria - the databases are either in sync, or are not.
  9238. If they are different, a detailed list of the differences is presented.
  9239.  
  9240. You may want to exclude or filter out certain differences. The way to do this is to add strings
  9241. to the C<--filter> option. To exclude a type of object, use "noname", where 'name' is the type of
  9242. object, for example, "noschema". To exclude objects of a certain type by a regular expression against
  9243. their name, use "noname=regex". See the examples below for a better understanding.
  9244.  
  9245. The types of objects that can be filtered include:
  9246.  
  9247. =over 4
  9248.  
  9249. =item user
  9250.  
  9251. =item schema
  9252.  
  9253. =item table
  9254.  
  9255. =item view
  9256.  
  9257. =item index
  9258.  
  9259. =item sequence
  9260.  
  9261. =item constraint
  9262.  
  9263. =item trigger
  9264.  
  9265. =item function
  9266.  
  9267. =back
  9268.  
  9269. The filter option "noposition" prevents verification of the position of
  9270. columns within a table.
  9271.  
  9272. The filter option "nofuncbody" prevents comparison of the bodies of all
  9273. functions.
  9274.  
  9275. The filter option "noperm" prevents comparison of object permissions.
  9276.  
  9277. To provide the second database, just append the differences to the first one
  9278. by a call to the appropriate connection argument. For example, to compare
  9279. databases on hosts alpha and bravo, use "--dbhost=alpha,bravo". Also see the
  9280. examples below.
  9281.  
  9282. If only a single host is given, it is assumed we are doing a "time-based" report.
  9283. The first time this is run a snapshot of all the items in the database is
  9284. saved to a local file. When you run it again, that snapshot is read in and
  9285. becomes "database #2" and is compared to the current database.
  9286.  
  9287. To replace the old stored file with the new version, use the --replace argument.
  9288.  
  9289. To enable snapshots at various points in time, you can use the "--suffix"
  9290. argument to make the filenames unique to each run. See the examples below.
  9291.  
  9292. Example 1: Verify that two databases on hosts star and line are the same:
  9293.  
  9294. check_postgres_same_schema --dbhost=star,line
  9295.  
  9296. Example 2: Same as before, but exclude any triggers with "slony" in their name
  9297.  
  9298. check_postgres_same_schema --dbhost=star,line --filter="notrigger=slony"
  9299.  
  9300. Example 3: Same as before, but also exclude all indexes
  9301.  
  9302. check_postgres_same_schema --dbhost=star,line --filter="notrigger=slony noindexes"
  9303.  
  9304. Example 4: Check differences for the database "battlestar" on different ports
  9305.  
  9306. check_postgres_same_schema --dbname=battlestar --dbport=5432,5544
  9307.  
  9308. Example 5: Create a daily and weekly snapshot file
  9309.  
  9310. check_postgres_same_schema --dbname=cylon --suffix=daily
  9311. check_postgres_same_schema --dbname=cylon --suffix=weekly
  9312.  
  9313. Example 6: Run a historical comparison, then replace the file
  9314.  
  9315. check_postgres_same_schema --dbname=cylon --suffix=daily --replace
  9316.  
  9317. =head2 B<sequence>
  9318.  
  9319. (C<symlink: check_postgres_sequence>) Checks how much room is left on all sequences in the database.
  9320. This is measured as the percent of total possible values that have been used for each sequence.
  9321. The I<--warning> and I<--critical> options should be expressed as percentages. The default values
  9322. are B<85%> for the warning and B<95%> for the critical. You may use --include and --exclude to
  9323. control which sequences are to be checked. Note that this check does account for unusual B<minvalue>
  9324. and B<increment by> values, but does not care if the sequence is set to cycle or not.
  9325.  
  9326. The output for Nagios gives the name of the sequence, the percentage used, and the number of 'calls'
  9327. left, indicating how many more times nextval can be called on that sequence before running into
  9328. the maximum value.
  9329.  
  9330. The output for MRTG returns the highest percentage across all sequences on the first line, and
  9331. the name of each sequence with that percentage on the fourth line, separated by a "|" (pipe)
  9332. if there are more than one sequence at that percentage.
  9333.  
  9334. Example 1: Give a warning if any sequences are approaching 95% full.
  9335.  
  9336. check_postgres_sequence --dbport=5432 --warning=95%
  9337.  
  9338. Example 2: Check that the sequence named "orders_id_seq" is not more than half full.
  9339.  
  9340. check_postgres_sequence --dbport=5432 --critical=50% --include=orders_id_seq
  9341.  
  9342. =head2 B<settings_checksum>
  9343.  
  9344. (C<symlink: check_postgres_settings_checksum>) Checks that all the Postgres settings are the same as last time you checked.
  9345. This is done by generating a checksum of a sorted list of setting names and
  9346. their values. Note that different users in the same database may have different
  9347. checksums, due to ALTER USER usage, and due to the fact that superusers see more
  9348. settings than ordinary users. Either the I<--warning> or the I<--critical> option
  9349. should be given, but not both. The value of each one is the checksum, a
  9350. 32-character hexadecimal value. You can run with the special C<--critical=0> option
  9351. to find out an existing checksum.
  9352.  
  9353. This action requires the Digest::MD5 module.
  9354.  
  9355. Example 1: Find the initial checksum for the database on port 5555 using the default user (usually postgres)
  9356.  
  9357. check_postgres_settings_checksum --port=5555 --critical=0
  9358.  
  9359. Example 2: Make sure no settings have changed and warn if so, using the checksum from above.
  9360.  
  9361. check_postgres_settings_checksum --port=5555 --warning=cd2f3b5e129dc2b4f5c0f6d8d2e64231
  9362.  
  9363. For MRTG output, returns a 1 or 0 indicating success of failure of the checksum to match. A
  9364. checksum must be provided as the C<--mrtg> argument. The fourth line always gives the
  9365. current checksum.
  9366.  
  9367. =head2 B<slony_status>
  9368.  
  9369. (C<symlink: check_postgres_slony_status>) Checks in the status of a Slony cluster by looking
  9370. at the results of Slony's sl_status view. This is returned as the number of seconds of "lag time".
  9371. The I<--warning> and I<--critical> options should be expressed as times. The default values
  9372. are B<60 seconds> for the warning and B<300 seconds> for the critical.
  9373.  
  9374. The optional argument I<--schema> indicated the schema that Slony is installed under. If it is
  9375. not given, the schema will be determined automatically each time this check is run.
  9376.  
  9377. Example 1: Give a warning if any Slony is lagged by more than 20 seconds
  9378.  
  9379. check_postgres_slony_status --warning 20
  9380.  
  9381. Example 2: Give a critical if Slony, installed under the schema "_slony", is over 10 minutes lagged
  9382.  
  9383. check_postgres_slony_status --schema=_slony --critical=600
  9384.  
  9385. =head2 B<timesync>
  9386.  
  9387. (C<symlink: check_postgres_timesync>) Compares the local system time with the time reported by one or more databases.
  9388. The I<--warning> and I<--critical> options represent the number of seconds between
  9389. the two systems before an alert is given. If neither is specified, the default values
  9390. are used, which are '2' and '5'. The warning value cannot be greater than the critical
  9391. value. Due to the non-exact nature of this test, values of '0' or '1' are not recommended.
  9392.  
  9393. The string returned shows the time difference as well as the time on each side written out.
  9394.  
  9395. Example 1: Check that databases on hosts ankh, morpork, and klatch are no more than 3 seconds off from the local time:
  9396.  
  9397. check_postgres_timesync --host=ankh,morpork,klatch --critical=3
  9398.  
  9399. For MRTG output, returns one the first line the number of seconds difference between the local
  9400. time and the database time. The fourth line returns the name of the database.
  9401.  
  9402. =head2 B<txn_idle>
  9403.  
  9404. (C<symlink: check_postgres_txn_idle>) Checks the number and duration of "idle
  9405. in transaction" queries on one or more databases. There is no need to run this
  9406. more than once on the same database cluster. Databases can be filtered by
  9407. using the I<--include> and I<--exclude> options. See the L</"BASIC FILTERING">
  9408. section below for more details.
  9409.  
  9410. The I<--warning> and I<--critical> options are given as units of time, signed
  9411. integers, or integers for units of time, and both must be provided (there are
  9412. no defaults). Valid units are 'seconds', 'minutes', 'hours', or 'days'. Each
  9413. may be written singular or abbreviated to just the first letter. If no units
  9414. are given and the numbers are unsigned, the units are assumed to be seconds.
  9415.  
  9416. This action requires Postgres 8.3 or better.
  9417.  
  9418. Example 1: Give a warning if any connection has been idle in transaction for more than 15 seconds:
  9419.  
  9420. check_postgres_txn_idle --port=5432 --warning='15 seconds'
  9421.  
  9422. Example 2: Give a warning if there are 50 or more transactions
  9423.  
  9424. check_postgres_txn_idle --port=5432 --warning='+50'
  9425.  
  9426. Example 3: Give a critical if 5 or more connections have been idle in
  9427. transaction for more than 10 seconds:
  9428.  
  9429. check_postgres_txn_idle --port=5432 --critical='5 for 10 seconds'
  9430.  
  9431. For MRTG output, returns the time in seconds the longest idle transaction has been running. The fourth
  9432. line returns the name of the database and other information about the longest transaction.
  9433.  
  9434. =head2 B<txn_time>
  9435.  
  9436. (C<symlink: check_postgres_txn_time>) Checks the length of open transactions on one or more databases.
  9437. There is no need to run this command more than once per database cluster.
  9438. Databases can be filtered by use of the
  9439. I<--include> and I<--exclude> options. See the L</"BASIC FILTERING"> section
  9440. for more details. The owner of the transaction can also be filtered, by use of
  9441. the I<--includeuser> and I<--excludeuser> options.
  9442. See the L</"USER NAME FILTERING"> section for more details.
  9443.  
  9444. The values or the I<--warning> and I<--critical> options are units of time, and
  9445. must be provided (no default). Valid units are 'seconds', 'minutes', 'hours',
  9446. or 'days'. Each may be written singular or abbreviated to just the first letter.
  9447. If no units are given, the units are assumed to be seconds.
  9448.  
  9449. This action requires Postgres 8.3 or better.
  9450.  
  9451. Example 1: Give a critical if any transaction has been open for more than 10 minutes:
  9452.  
  9453. check_postgres_txn_time --port=5432 --critical='10 minutes'
  9454.  
  9455. Example 1: Warn if user 'warehouse' has a transaction open over 30 seconds
  9456.  
  9457. check_postgres_txn_time --port-5432 --warning=30s --includeuser=warehouse
  9458.  
  9459. For MRTG output, returns the maximum time in seconds a transaction has been open on the
  9460. first line. The fourth line gives the name of the database.
  9461.  
  9462. =head2 B<txn_wraparound>
  9463.  
  9464. (C<symlink: check_postgres_txn_wraparound>) Checks how close to transaction wraparound one or more databases are getting.
  9465. The I<--warning> and I<--critical> options indicate the number of transactions done, and must be a positive integer.
  9466. If either option is not given, the default values of 1.3 and 1.4 billion are used. There is no need to run this command
  9467. more than once per database cluster. For a more detailed discussion of what this number represents and what to do about
  9468. it, please visit the page
  9469. L<http://www.postgresql.org/docs/current/static/routine-vacuuming.html#VACUUM-FOR-WRAPAROUND>
  9470.  
  9471. The warning and critical values can have underscores in the number for legibility, as Perl does.
  9472.  
  9473. Example 1: Check the default values for the localhost database
  9474.  
  9475. check_postgres_txn_wraparound --host=localhost
  9476.  
  9477. Example 2: Check port 6000 and give a critical when 1.7 billion transactions are hit:
  9478.  
  9479. check_postgres_txn_wraparound --port=6000 --critical=1_700_000_000
  9480.  
  9481. For MRTG output, returns the highest number of transactions for all databases on line one,
  9482. while line 4 indicates which database it is.
  9483.  
  9484. =head2 B<version>
  9485.  
  9486. (C<symlink: check_postgres_version>) Checks that the required version of Postgres is running. The
  9487. I<--warning> and I<--critical> options (only one is required) must be of
  9488. the format B<X.Y> or B<X.Y.Z> where B<X> is the major version number,
  9489. B<Y> is the minor version number, and B<Z> is the revision.
  9490.  
  9491. Example 1: Give a warning if the database on port 5678 is not version 8.4.10:
  9492.  
  9493. check_postgres_version --port=5678 -w=8.4.10
  9494.  
  9495. Example 2: Give a warning if any databases on hosts valley,grain, or sunshine is not 8.3:
  9496.  
  9497. check_postgres_version -H valley,grain,sunshine --critical=8.3
  9498.  
  9499. For MRTG output, reports a 1 or a 0 indicating success or failure on the first line. The
  9500. fourth line indicates the current version. The version must be provided via the C<--mrtg> option.
  9501.  
  9502. =head2 B<wal_files>
  9503.  
  9504. (C<symlink: check_postgres_wal_files>) Checks how many WAL files exist in the F<pg_xlog> directory, which is found
  9505. off of your B<data_directory>, sometimes as a symlink to another physical disk for
  9506. performance reasons. This action must be run as a superuser, in order to access the
  9507. contents of the F<pg_xlog> directory. The minimum version to use this action is
  9508. Postgres 8.1. The I<--warning> and I<--critical> options are simply the number of
  9509. files in the F<pg_xlog> directory. What number to set this to will vary, but a general
  9510. guideline is to put a number slightly higher than what is normally there, to catch
  9511. problems early.
  9512.  
  9513. Normally, WAL files are closed and then re-used, but a long-running open
  9514. transaction, or a faulty B<archive_command> script, may cause Postgres to
  9515. create too many files. Ultimately, this will cause the disk they are on to run
  9516. out of space, at which point Postgres will shut down.
  9517.  
  9518. Example 1: Check that the number of WAL files is 20 or less on host "pluto"
  9519.  
  9520. check_postgres_wal_files --host=pluto --critical=20
  9521.  
  9522. For MRTG output, reports the number of WAL files on line 1.
  9523.  
  9524. =head2 B<rebuild_symlinks>
  9525.  
  9526. =head2 B<rebuild_symlinks_force>
  9527.  
  9528. This action requires no other arguments, and does not connect to any databases,
  9529. but simply creates symlinks in the current directory for each action, in the form
  9530. B<check_postgres_E<lt>action_nameE<gt>>.
  9531. If the file already exists, it will not be overwritten. If the action is rebuild_symlinks_force,
  9532. then symlinks will be overwritten. The option --symlinks is a shorter way of saying
  9533. --action=rebuild_symlinks
  9534.  
  9535. =head1 BASIC FILTERING
  9536.  
  9537. The options I<--include> and I<--exclude> can be combined to limit which
  9538. things are checked, depending on the action. The name of the database can
  9539. be filtered when using the following actions:
  9540. backends, database_size, locks, query_time, txn_idle, and txn_time.
  9541. The name of a relation can be filtered when using the following actions:
  9542. bloat, index_size, table_size, relation_size, last_vacuum, last_autovacuum,
  9543. last_analyze, and last_autoanalyze.
  9544. The name of a setting can be filtered when using the settings_checksum action.
  9545. The name of a file system can be filtered when using the disk_space action.
  9546.  
  9547. If only an include option is given, then ONLY those entries that match will be
  9548. checked. However, if given both exclude and include, the exclusion is done first,
  9549. and the inclusion after, to reinstate things that may have been excluded. Both
  9550. I<--include> and I<--exclude> can be given multiple times,
  9551. and/or as comma-separated lists. A leading tilde will match the following word
  9552. as a regular expression.
  9553.  
  9554. To match a schema, end the search term with a single period. Leading tildes can
  9555. be used for schemas as well.
  9556.  
  9557. Be careful when using filtering: an inclusion rule on the backends, for example,
  9558. may report no problems not only because the matching database had no backends,
  9559. but because you misspelled the name of the database!
  9560.  
  9561. Examples:
  9562.  
  9563. Only checks items named pg_class:
  9564.  
  9565. --include=pg_class
  9566.  
  9567. Only checks items containing the letters 'pg_':
  9568.  
  9569. --include=~pg_
  9570.  
  9571. Only check items beginning with 'pg_':
  9572.  
  9573. --include=~^pg_
  9574.  
  9575. Exclude the item named 'test':
  9576.  
  9577. --exclude=test
  9578.  
  9579. Exclude all items containing the letters 'test:
  9580.  
  9581. --exclude=~test
  9582.  
  9583. Exclude all items in the schema 'pg_catalog':
  9584.  
  9585. --exclude='pg_catalog.'
  9586.  
  9587. Exclude all items containing the letters 'ace', but allow the item 'faceoff':
  9588.  
  9589. --exclude=~ace --include=faceoff
  9590.  
  9591. Exclude all items which start with the letters 'pg_', which contain the letters 'slon',
  9592. or which are named 'sql_settings' or 'green'. Specifically check items with the letters 'prod' in their names, and always check the item named 'pg_relname':
  9593.  
  9594. --exclude=~^pg_,~slon,sql_settings --exclude=green --include=~prod,pg_relname
  9595.  
  9596. =head1 USER NAME FILTERING
  9597.  
  9598. The options I<--includeuser> and I<--excludeuser> can be used on some actions
  9599. to only examine database objects owned by (or not owned by) one or more users.
  9600. An I<--includeuser> option always trumps an I<--excludeuser> option. You can
  9601. give each option more than once for multiple users, or you can give a
  9602. comma-separated list. The actions that currently use these options are:
  9603.  
  9604. =over 4
  9605.  
  9606. =item database_size
  9607.  
  9608. =item last_analyze
  9609.  
  9610. =item last_autoanalyze
  9611.  
  9612. =item last_vacuum
  9613.  
  9614. =item last_autovacuum
  9615.  
  9616. =item query_time
  9617.  
  9618. =item relation_size
  9619.  
  9620. =item txn_time
  9621.  
  9622. =back
  9623.  
  9624. Examples:
  9625.  
  9626. Only check items owned by the user named greg:
  9627.  
  9628. --includeuser=greg
  9629.  
  9630. Only check items owned by either watson or crick:
  9631.  
  9632. --includeuser=watson,crick
  9633.  
  9634. Only check items owned by crick,franklin, watson, or wilkins:
  9635.  
  9636. --includeuser=watson --includeuser=franklin --includeuser=crick,wilkins
  9637.  
  9638. Check all items except for those belonging to the user scott:
  9639.  
  9640. --excludeuser=scott
  9641.  
  9642. =head1 TEST MODE
  9643.  
  9644. To help in setting things up, this program can be run in a "test mode" by
  9645. specifying the I<--test> option. This will perform some basic tests to
  9646. make sure that the databases can be contacted, and that certain per-action
  9647. prerequisites are met, such as whether the user is a superuser, if the version
  9648. of Postgres is new enough, and if stats_row_level is enabled.
  9649.  
  9650. =head1 FILES
  9651.  
  9652. In addition to command-line configurations, you can put any options inside of a file. The file
  9653. F<.check_postgresrc> in the current directory will be used if found. If not found, then the file
  9654. F<~/.check_postgresrc> will be used. Finally, the file /etc/check_postgresrc will be used if available.
  9655. The format of the file is option = value, one per line. Any line starting with a '#' will be skipped.
  9656. Any values loaded from a check_postgresrc file will be overwritten by command-line options. All
  9657. check_postgresrc files can be ignored by supplying a C<--no-checkpostgresrc> argument.
  9658.  
  9659. =head1 ENVIRONMENT VARIABLES
  9660.  
  9661. The environment variable I<$ENV{HOME}> is used to look for a F<.check_postgresrc> file.
  9662. The environment variable I<$ENV{PGBINDIR}> is used to look for PostgreSQL binaries.
  9663.  
  9664. =head1 TIPS AND TRICKS
  9665.  
  9666. Since this program uses the B<psql> program, make sure it is accessible to the
  9667. user running the script. If run as a cronjob, this often means modifying the
  9668. B<PATH> environment variable.
  9669.  
  9670. If you are using Nagios in embedded Perl mode, use the C<--action> argument
  9671. instead of symlinks, so that the plugin only gets compiled one time.
  9672.  
  9673. =head1 DEPENDENCIES
  9674.  
  9675. Access to a working version of psql, and the following very standard Perl modules:
  9676.  
  9677. =over 4
  9678.  
  9679. =item B<Cwd>
  9680.  
  9681. =item B<Getopt::Long>
  9682.  
  9683. =item B<File::Basename>
  9684.  
  9685. =item B<File::Temp>
  9686.  
  9687. =item B<Time::HiRes> (if C<$opt{showtime}> is set to true, which is the default)
  9688.  
  9689. =back
  9690.  
  9691. The L</settings_checksum> action requires the B<Digest::MD5> module.
  9692.  
  9693. The L</checkpoint> action requires the B<Date::Parse> module.
  9694.  
  9695. Some actions require access to external programs. If psql is not explicitly
  9696. specified, the command B<C<which>> is used to find it. The program B<C</bin/df>>
  9697. is needed by the L</disk_space> action.
  9698.  
  9699. =head1 DEVELOPMENT
  9700.  
  9701. Development happens using the git system. You can clone the latest version by doing:
  9702.  
  9703. git clone git://bucardo.org/check_postgres.git
  9704.  
  9705. =head1 MAILING LIST
  9706.  
  9707. Three mailing lists are available. For discussions about the program, bug reports,
  9708. feature requests, and commit notices, send email to check_postgres@bucardo.org
  9709.  
  9710. https://mail.endcrypt.com/mailman/listinfo/check_postgres
  9711.  
  9712. A low-volume list for announcement of new versions and important notices is the
  9713. 'check_postgres-announce' list:
  9714.  
  9715. https://mail.endcrypt.com/mailman/listinfo/check_postgres-announce
  9716.  
  9717. Source code changes (via git-commit) are sent to the
  9718. 'check_postgres-commit' list:
  9719.  
  9720. https://mail.endcrypt.com/mailman/listinfo/check_postgres-commit
  9721.  
  9722. =head1 HISTORY
  9723.  
  9724. Items not specifically attributed are by GSM (Greg Sabino Mullane).
  9725.  
  9726. =over 4
  9727.  
  9728. =item B<Version 2.20.0>
  9729.  
  9730. Add check for pgagent jobs (David E. Wheeler)
  9731.  
  9732. Force STDOUT to use utf8 for proper output
  9733. (Greg Sabino Mullane; reported by Emmanuel Lesouef)
  9734.  
  9735. Allow for spaces in item lists when doing same_schema.
  9736.  
  9737. Allow txn_idle to work again for < 8.3 servers by switching to query_time.
  9738.  
  9739. Cache sequence information to speed up same_schema runs.
  9740.  
  9741. Adjust docs to show colon as a better separator inside args for locks
  9742. (Charles Sprickman)
  9743.  
  9744. =item B<Version 2.19.0> January 17, 2012
  9745.  
  9746. Add the --assume-prod option (Cédric Villemain)
  9747.  
  9748. Add the cluster_id check (Cédric Villemain)
  9749.  
  9750. Improve settings_checksum and checkpoint tests (Cédric Villemain)
  9751.  
  9752. Do no do an inner join to pg_user when checking database size
  9753. (Greg Sabino Mullane; reported by Emmanuel Lesouef)
  9754.  
  9755. Use the full path when getting sequence information for same_schema.
  9756. (Greg Sabino Mullane; reported by Cindy Wise)
  9757.  
  9758. Fix the formula for calculating xlog positions (Euler Taveira de Oliveira)
  9759.  
  9760. Better ordering of output for bloat check - make indexes as important
  9761. as tables (Greg Sabino Mullane; reported by Jens Wilke)
  9762.  
  9763. Show the dbservice if it was used at top of same_schema output
  9764. (Mike Blackwell)
  9765.  
  9766. Better installation paths (Greg Sabino Mullane, per bug 53)
  9767.  
  9768. =item B<Version 2.18.0> October 2, 2011
  9769.  
  9770. Redo the same_schema action. Use new --filter argument for all filtering.
  9771. Allow comparisons between any number of databases.
  9772. Remove the dbname2, dbport2, etc. arguments.
  9773. Allow comparison of the same db over time.
  9774.  
  9775. Swap db1 and db2 if the slave is 1 for the hot standby check (David E. Wheeler)
  9776.  
  9777. Allow multiple --schema arguments for the slony_status action (GSM and Jehan-Guillaume de Rorthais)
  9778.  
  9779. Fix ORDER BY in the last vacuum/analyze action (Nicolas Thauvin)
  9780.  
  9781. Fix check_hot_standby_delay perfdata output (Nicolas Thauvin)
  9782.  
  9783. Look in the correct place for the .ready files with the archive_ready action (Nicolas Thauvin)
  9784.  
  9785. New action: commitratio (Guillaume Lelarge)
  9786.  
  9787. New action: hitratio (Guillaume Lelarge)
  9788.  
  9789. Make sure --action overrides the symlink naming trick.
  9790.  
  9791. Set defaults for archive_ready and wal_files (Thomas Guettler, GSM)
  9792.  
  9793. Better output for wal_files and archive_ready (GSM)
  9794.  
  9795. Fix warning when client_port set to empty string (bug #79)
  9796.  
  9797. Account for "empty row" in -x output (i.e. source of functions).
  9798.  
  9799. Fix some incorrectly named data fields (Andy Lester)
  9800.  
  9801. Expand the number of pgbouncer actions (Ruslan Kabalin)
  9802.  
  9803. Give detailed information and refactor txn_idle, txn_time, and query_time
  9804. (Per request from bug #61)
  9805.  
  9806. Set maxalign to 8 in the bloat check if box identified as '64-bit'
  9807. (Michel Sijmons, bug #66)
  9808.  
  9809. Support non-standard version strings in the bloat check.
  9810. (Michel Sijmons and Gurjeet Singh, bug #66)
  9811.  
  9812. Do not show excluded databases in some output (Ruslan Kabalin)
  9813.  
  9814. Allow "and", "or" inside arguments (David E. Wheeler)
  9815.  
  9816. Add the "new_version_box" action.
  9817.  
  9818. Fix psql version regex (Peter Eisentraut, bug #69)
  9819.  
  9820. Add the --assume-standby-mode option (Ruslan Kabalin)
  9821.  
  9822. Note that txn_idle and query_time require 8.3 (Thomas Guettler)
  9823.  
  9824. Standardize and clean up all perfdata output (bug #52)
  9825.  
  9826. Exclude "idle in transaction" from the query_time check (bug #43)
  9827.  
  9828. Fix the perflimit for the bloat action (bug #50)
  9829.  
  9830. Clean up the custom_query action a bit.
  9831.  
  9832. Fix space in perfdata for hot_standby_delay action (Nicolas Thauvin)
  9833.  
  9834. Handle undef percents in check_fsm_relations (Andy Lester)
  9835.  
  9836. Fix typo in dbstats action (Stas Vitkovsky)
  9837.  
  9838. Fix MRTG for last vacuum and last_analyze actions.
  9839.  
  9840. =item B<Version 2.17.0> no public release
  9841.  
  9842. =item B<Version 2.16.0> January 20, 2011
  9843.  
  9844. Add new action 'hot_standby_delay' (Nicolas Thauvin)
  9845. Add cache-busting for the version-grabbing utilities.
  9846. Fix problem with going to next method for new_version_pg
  9847. (Greg Sabino Mullane, reported by Hywel Mallett in bug #65)
  9848. Allow /usr/local/etc as an alternative location for the
  9849. check_postgresrc file (Hywel Mallett)
  9850. Do not use tgisconstraint in same_schema if Postgres >= 9
  9851. (Guillaume Lelarge)
  9852.  
  9853. =item B<Version 2.15.4> January 3, 2011
  9854.  
  9855. Fix warning when using symlinks
  9856. (Greg Sabino Mullane, reported by Peter Eisentraut in bug #63)
  9857.  
  9858. =item B<Version 2.15.3> December 30, 2010
  9859.  
  9860. Show OK for no matching txn_idle entries.
  9861.  
  9862. =item B<Version 2.15.2> December 28, 2010
  9863.  
  9864. Better formatting of sizes in the bloat action output.
  9865.  
  9866. Remove duplicate perfs in bloat action output.
  9867.  
  9868. =item B<Version 2.15.1> December 27, 2010
  9869.  
  9870. Fix problem when examining items in pg_settings (Greg Sabino Mullane)
  9871.  
  9872. For connection test, return critical, not unknown, on FATAL errors
  9873. (Greg Sabino Mullane, reported by Peter Eisentraut in bug #62)
  9874.  
  9875. =item B<Version 2.15.0> November 8, 2010
  9876.  
  9877. Add --quiet argument to suppress output on OK Nagios results
  9878. Add index comparison for same_schema (Norman Yamada and Greg Sabino Mullane)
  9879. Use $ENV{PGSERVICE} instead of "service=" to prevent problems (Guillaume Lelarge)
  9880. Add --man option to show the entire manual. (Andy Lester)
  9881. Redo the internal run_command() sub to use -x and hashes instead of regexes.
  9882. Fix error in custom logic (Andreas Mager)
  9883. Add the "pgbouncer_checksum" action (Guillaume Lelarge)
  9884. Fix regex to work on WIN32 for check_fsm_relations and check_fsm_pages (Luke Koops)
  9885. Don't apply a LIMIT when using --exclude on the bloat action (Marti Raudsepp)
  9886. Change the output of query_time to show pid,user,port, and address (Giles Westwood)
  9887. Fix to show database properly when using slony_status (Guillaume Lelarge)
  9888. Allow warning items for same_schema to be comma-separated (Guillaume Lelarge)
  9889. Constraint definitions across Postgres versions match better in same_schema.
  9890. Work against "EnterpriseDB" databases (Sivakumar Krishnamurthy and Greg Sabino Mullane)
  9891. Separate perfdata with spaces (Jehan-Guillaume (ioguix) de Rorthais)
  9892. Add new action "archive_ready" (Jehan-Guillaume (ioguix) de Rorthais)
  9893.  
  9894. =item B<Version 2.14.3> (March 1, 2010)
  9895.  
  9896. Allow slony_status action to handle more than one slave.
  9897. Use commas to separate function args in same_schema output (Robert Treat)
  9898.  
  9899. =item B<Version 2.14.2> (February 18, 2010)
  9900.  
  9901. Change autovac_freeze default warn/critical back to 90%/95% (Robert Treat)
  9902. Put all items one-per-line for relation size actions if --verbose=1
  9903.  
  9904. =item B<Version 2.14.1> (February 17, 2010)
  9905.  
  9906. Don't use $^T in logfile check, as script may be long-running
  9907. Change the error string for the logfile action for easier exclusion
  9908. by programs like tail_n_mail
  9909.  
  9910. =item B<Version 2.14.0> (February 11, 2010)
  9911.  
  9912. Added the 'slony_status' action.
  9913. Changed the logfile sleep from 0.5 to 1, as 0.5 gets rounded to 0 on some boxes!
  9914.  
  9915. =item B<Version 2.13.2> (February 4, 2010)
  9916.  
  9917. Allow timeout option to be used for logtime 'sleep' time.
  9918.  
  9919. =item B<Version 2.13.2> (February 4, 2010)
  9920.  
  9921. Show offending database for query_time action.
  9922. Apply perflimit to main output for sequence action.
  9923. Add 'noowner' option to same_schema action.
  9924. Raise sleep timeout for logfile check to 15 seconds.
  9925.  
  9926. =item B<Version 2.13.1> (February 2, 2010)
  9927.  
  9928. Fix bug preventing column constraint differences from 2 > 1 for same_schema from being shown.
  9929. Allow aliases 'dbname1', 'dbhost1', 'dbport1',etc.
  9930. Added "nolanguage" as a filter for the same_schema option.
  9931. Don't track "generic" table constraints (e.. $1, $2) using same_schema
  9932.  
  9933. =item B<Version 2.13.0> (January 29, 2010)
  9934.  
  9935. Allow "nofunctions" as a filter for the same_schema option.
  9936. Added "noperm" as a filter for the same_schema option.
  9937. Ignore dropped columns when considered positions for same_schema (Guillaume Lelarge)
  9938.  
  9939. =item B<Version 2.12.1> (December 3, 2009)
  9940.  
  9941. Change autovac_freeze default warn/critical from 90%/95% to 105%/120% (Marti Raudsepp)
  9942.  
  9943. =item B<Version 2.12.0> (December 3, 2009)
  9944.  
  9945. Allow the temporary directory to be specified via the "tempdir" argument,
  9946. for systems that need it (e.g. /tmp is not owned by root).
  9947. Fix so old versions of Postgres (< 8.0) use the correct default database (Giles Westwood)
  9948. For "same_schema" trigger mismatches, show the attached table.
  9949. Add the new_version_bc check for Bucardo version checking.
  9950. Add database name to perf output for last_vacuum|analyze (Guillaume Lelarge)
  9951. Fix for bloat action against old versions of Postgres without the 'block_size' param.
  9952.  
  9953. =item B<Version 2.11.1> (August 27, 2009)
  9954.  
  9955. Proper Nagios output for last_vacuum|analyze actions. (Cédric Villemain)
  9956. Proper Nagios output for locks action. (Cédric Villemain)
  9957. Proper Nagios output for txn_wraparound action. (Cédric Villemain)
  9958. Fix for constraints with embedded newlines for same_schema.
  9959. Allow --exclude for all items when using same_schema.
  9960.  
  9961. =item B<Version 2.11.0> (August 23, 2009)
  9962.  
  9963. Add Nagios perf output to the wal_files check (Cédric Villemain)
  9964. Add support for .check_postgresrc, per request from Albe Laurenz.
  9965. Allow list of web fetch methods to be changed with the --get_method option.
  9966. Add support for the --language argument, which overrides any ENV.
  9967. Add the --no-check_postgresrc flag.
  9968. Ensure check_postgresrc options are completely overridden by command-line options.
  9969. Fix incorrect warning > critical logic in replicate_rows (Glyn Astill)
  9970.  
  9971. =item B<Version 2.10.0> (August 3, 2009)
  9972.  
  9973. For same_schema, compare view definitions, and compare languages.
  9974. Make script into a global executable via the Makefile.PL file.
  9975. Better output when comparing two databases.
  9976. Proper Nagios output syntax for autovac_freeze and backends checks (Cédric Villemain)
  9977.  
  9978. =item B<Version 2.9.5> (July 24, 2009)
  9979.  
  9980. Don't use a LIMIT in check_bloat if --include is used. Per complaint from Jeff Frost.
  9981.  
  9982. =item B<Version 2.9.4> (July 21, 2009)
  9983.  
  9984. More French translations (Guillaume Lelarge)
  9985.  
  9986. =item B<Version 2.9.3> (July 14, 2009)
  9987.  
  9988. Quote dbname in perf output for the backends check. (Davide Abrigo)
  9989. Add 'fetch' as an alternative method for new_version checks, as this
  9990. comes by default with FreeBSD. (Hywel Mallett)
  9991.  
  9992. =item B<Version 2.9.2> (July 12, 2009)
  9993.  
  9994. Allow dots and dashes in database name for the backends check (Davide Abrigo)
  9995. Check and display the database for each match in the bloat check (Cédric Villemain)
  9996. Handle 'too many connections' FATAL error in the backends check with a critical,
  9997. rather than a generic error (Greg, idea by Jürgen Schulz-Brüssel)
  9998. Do not allow perflimit to interfere with exclusion rules in the vacuum and
  9999. analyze tests. (Greg, bug reported by Jeff Frost)
  10000.  
  10001. =item B<Version 2.9.1> (June 12, 2009)
  10002.  
  10003. Fix for multiple databases with the check_bloat action (Mark Kirkwood)
  10004. Fixes and improvements to the same_schema action (Jeff Boes)
  10005. Write tests for same_schema, other minor test fixes (Jeff Boes)
  10006.  
  10007. =item B<Version 2.9.0> (May 28, 2009)
  10008.  
  10009. Added the same_schema action (Greg)
  10010.  
  10011. =item B<Version 2.8.1> (May 15, 2009)
  10012.  
  10013. Added timeout via statement_timeout in addition to perl alarm (Greg)
  10014.  
  10015. =item B<Version 2.8.0> (May 4, 2009)
  10016.  
  10017. Added internationalization support (Greg)
  10018. Added the 'disabled_triggers' check (Greg)
  10019. Added the 'prepared_txns' check (Greg)
  10020. Added the 'new_version_cp' and 'new_version_pg' checks (Greg)
  10021. French translations (Guillaume Lelarge)
  10022. Make the backends search return ok if no matches due to inclusion rules,
  10023. per report by Guillaume Lelarge (Greg)
  10024. Added comprehensive unit tests (Greg, Jeff Boes, Selena Deckelmann)
  10025. Make fsm_pages and fsm_relations handle 8.4 servers smoothly. (Greg)
  10026. Fix missing 'upd' field in show_dbstats (Andras Fabian)
  10027. Allow ENV{PGCONTROLDATA} and ENV{PGBINDIR}. (Greg)
  10028. Add various Perl module infrastructure (e.g. Makefile.PL) (Greg)
  10029. Fix incorrect regex in txn_wraparound (Greg)
  10030. For txn_wraparound: consistent ordering and fix duplicates in perf output (Andras Fabian)
  10031. Add in missing exabyte regex check (Selena Deckelmann)
  10032. Set stats to zero if we bail early due to USERWHERECLAUSE (Andras Fabian)
  10033. Add additional items to dbstats output (Andras Fabian)
  10034. Remove --schema option from the fsm_ checks. (Greg Mullane and Robert Treat)
  10035. Handle case when ENV{PGUSER} is set. (Andy Lester)
  10036. Many various fixes. (Jeff Boes)
  10037. Fix --dbservice: check version and use ENV{PGSERVICE} for old versions (Cédric Villemain)
  10038.  
  10039. =item B<Version 2.7.3> (February 10, 2009)
  10040.  
  10041. Make the sequence action check if sequence being used for a int4 column and
  10042. react appropriately. (Michael Glaesemann)
  10043.  
  10044. =item B<Version 2.7.2> (February 9, 2009)
  10045.  
  10046. Fix to prevent multiple groupings if db arguments given.
  10047.  
  10048. =item B<Version 2.7.1> (February 6, 2009)
  10049.  
  10050. Allow the -p argument for port to work again.
  10051.  
  10052. =item B<Version 2.7.0> (February 4, 2009)
  10053.  
  10054. Do not require a connection argument, but use defaults and ENV variables when
  10055. possible: PGHOST, PGPORT, PGUSER, PGDATABASE.
  10056.  
  10057. =item B<Version 2.6.1> (February 4, 2009)
  10058.  
  10059. Only require Date::Parse to be loaded if using the checkpoint action.
  10060.  
  10061. =item B<Version 2.6.0> (January 26, 2009)
  10062.  
  10063. Add the 'checkpoint' action.
  10064.  
  10065. =item B<Version 2.5.4> (January 7, 2009)
  10066.  
  10067. Better checking of $opt{dbservice} structure (Cédric Villemain)
  10068. Fix time display in timesync action output (Selena Deckelmann)
  10069. Fix documentation typos (Josh Tolley)
  10070.  
  10071. =item B<Version 2.5.3> (December 17, 2008)
  10072.  
  10073. Minor fix to regex in verify_version (Lee Jensen)
  10074.  
  10075. =item B<Version 2.5.2> (December 16, 2008)
  10076.  
  10077. Minor documentation tweak.
  10078.  
  10079. =item B<Version 2.5.1> (December 11, 2008)
  10080.  
  10081. Add support for --noidle flag to prevent backends action from counting idle processes.
  10082. Patch by Selena Deckelmann.
  10083.  
  10084. Fix small undefined warning when not using --dbservice.
  10085.  
  10086. =item B<Version 2.5.0> (December 4, 2008)
  10087.  
  10088. Add support for the pg_Service.conf file with the --dbservice option.
  10089.  
  10090. =item B<Version 2.4.3> (November 7, 2008)
  10091.  
  10092. Fix options for replicate_row action, per report from Jason Gordon.
  10093.  
  10094. =item B<Version 2.4.2> (November 6, 2008)
  10095.  
  10096. Wrap File::Temp::cleanup() calls in eval, in case File::Temp is an older version.
  10097. Patch by Chris Butler.
  10098.  
  10099. =item B<Version 2.4.1> (November 5, 2008)
  10100.  
  10101. Cast numbers to numeric to support sequences ranges > bigint in check_sequence action.
  10102. Thanks to Scott Marlowe for reporting this.
  10103.  
  10104. =item B<Version 2.4.0> (October 26, 2008)
  10105.  
  10106. Add Cacti support with the dbstats action.
  10107. Pretty up the time output for last vacuum and analyze actions.
  10108. Show the percentage of backends on the check_backends action.
  10109.  
  10110. =item B<Version 2.3.10> (October 23, 2008)
  10111.  
  10112. Fix minor warning in action check_bloat with multiple databases.
  10113. Allow warning to be greater than critical when using the --reverse option.
  10114. Support the --perflimit option for the check_sequence action.
  10115.  
  10116. =item B<Version 2.3.9> (October 23, 2008)
  10117.  
  10118. Minor tweak to way we store the default port.
  10119.  
  10120. =item B<Version 2.3.8> (October 21, 2008)
  10121.  
  10122. Allow the default port to be changed easily.
  10123. Allow transform of simple output by MB, GB, etc.
  10124.  
  10125. =item B<Version 2.3.7> (October 14, 2008)
  10126.  
  10127. Allow multiple databases in 'sequence' action. Reported by Christoph Zwerschke.
  10128.  
  10129. =item B<Version 2.3.6> (October 13, 2008)
  10130.  
  10131. Add missing $schema to check_fsm_pages. (Robert Treat)
  10132.  
  10133. =item B<Version 2.3.5> (October 9, 2008)
  10134.  
  10135. Change option 'checktype' to 'valtype' to prevent collisions with -c[ritical]
  10136. Better handling of errors.
  10137.  
  10138. =item B<Version 2.3.4> (October 9, 2008)
  10139.  
  10140. Do explicit cleanups of the temp directory, per problems reported by sb@nnx.com.
  10141.  
  10142. =item B<Version 2.3.3> (October 8, 2008)
  10143.  
  10144. Account for cases where some rounding queries give -0 instead of 0.
  10145. Thanks to Glyn Astill for helping to track this down.
  10146.  
  10147. =item B<Version 2.3.2> (October 8, 2008)
  10148.  
  10149. Always quote identifiers in check_replicate_row action.
  10150.  
  10151. =item B<Version 2.3.1> (October 7, 2008)
  10152.  
  10153. Give a better error if one of the databases cannot be reached.
  10154.  
  10155. =item B<Version 2.3.0> (October 4, 2008)
  10156.  
  10157. Add the "sequence" action, thanks to Gavin M. Roy for the idea.
  10158. Fix minor problem with autovac_freeze action when using MRTG output.
  10159. Allow output argument to be case-insensitive.
  10160. Documentation fixes.
  10161.  
  10162. =item B<Version 2.2.4> (October 3, 2008)
  10163.  
  10164. Fix some minor typos
  10165.  
  10166. =item B<Version 2.2.3> (October 1, 2008)
  10167.  
  10168. Expand range of allowed names for --repinfo argument (Glyn Astill)
  10169. Documentation tweaks.
  10170.  
  10171. =item B<Version 2.2.2> (September 30, 2008)
  10172.  
  10173. Fixes for minor output and scoping problems.
  10174.  
  10175. =item B<Version 2.2.1> (September 28, 2008)
  10176.  
  10177. Add MRTG output to fsm_pages and fsm_relations.
  10178. Force error messages to one-line for proper Nagios output.
  10179. Check for invalid prereqs on failed command. From conversations with Euler Taveira de Oliveira.
  10180. Tweak the fsm_pages formula a little.
  10181.  
  10182. =item B<Version 2.2.0> (September 25, 2008)
  10183.  
  10184. Add fsm_pages and fsm_relations actions. (Robert Treat)
  10185.  
  10186. =item B<Version 2.1.4> (September 22, 2008)
  10187.  
  10188. Fix for race condition in txn_time action.
  10189. Add --debugoutput option.
  10190.  
  10191. =item B<Version 2.1.3> (September 22, 2008)
  10192.  
  10193. Allow alternate arguments "dbhost" for "host" and "dbport" for "port".
  10194. Output a zero as default value for second line of MRTG output.
  10195.  
  10196. =item B<Version 2.1.2> (July 28, 2008)
  10197.  
  10198. Fix sorting error in the "disk_space" action for non-Nagios output.
  10199. Allow --simple as a shortcut for --output=simple.
  10200.  
  10201. =item B<Version 2.1.1> (July 22, 2008)
  10202.  
  10203. Don't check databases with datallowconn false for the "autovac_freeze" action.
  10204.  
  10205. =item B<Version 2.1.0> (July 18, 2008)
  10206.  
  10207. Add the "autovac_freeze" action, thanks to Robert Treat for the idea and design.
  10208. Put an ORDER BY on the "txn_wraparound" action.
  10209.  
  10210. =item B<Version 2.0.1> (July 16, 2008)
  10211.  
  10212. Optimizations to speed up the "bloat" action quite a bit.
  10213. Fix "version" action to not always output in mrtg mode.
  10214.  
  10215. =item B<Version 2.0.0> (July 15, 2008)
  10216.  
  10217. Add support for MRTG and "simple" output options.
  10218. Many small improvements to nearly all actions.
  10219.  
  10220. =item B<Version 1.9.1> (June 24, 2008)
  10221.  
  10222. Fix an error in the bloat SQL in 1.9.0
  10223. Allow percentage arguments to be over 99%
  10224. Allow percentages in the bloat --warning and --critical (thanks to Robert Treat for the idea)
  10225.  
  10226. =item B<Version 1.9.0> (June 22, 2008)
  10227.  
  10228. Don't include information_schema in certain checks. (Jeff Frost)
  10229. Allow --include and --exclude to use schemas by using a trailing period.
  10230.  
  10231. =item B<Version 1.8.5> (June 22, 2008)
  10232.  
  10233. Output schema name before table name where appropriate.
  10234. Thanks to Jeff Frost.
  10235.  
  10236. =item B<Version 1.8.4> (June 19, 2008)
  10237.  
  10238. Better detection of problems in --replicate_row.
  10239.  
  10240. =item B<Version 1.8.3> (June 18, 2008)
  10241.  
  10242. Fix 'backends' action: there may be no rows in pg_stat_activity, so run a second
  10243. query if needed to find the max_connections setting.
  10244. Thanks to Jeff Frost for the bug report.
  10245.  
  10246. =item B<Version 1.8.2> (June 10, 2008)
  10247.  
  10248. Changes to allow working under Nagios' embedded Perl mode. (Ioannis Tambouras)
  10249.  
  10250. =item B<Version 1.8.1> (June 9, 2008)
  10251.  
  10252. Allow 'bloat' action to work on Postgres version 8.0.
  10253. Allow for different commands to be run for each action depending on the server version.
  10254. Give better warnings when running actions not available on older Postgres servers.
  10255.  
  10256. =item B<Version 1.8.0> (June 3, 2008)
  10257.  
  10258. Add the --reverse option to the custom_query action.
  10259.  
  10260. =item B<Version 1.7.1> (June 2, 2008)
  10261.  
  10262. Fix 'query_time' action: account for race condition in which zero rows appear in pg_stat_activity.
  10263. Thanks to Dustin Black for the bug report.
  10264.  
  10265. =item B<Version 1.7.0> (May 11, 2008)
  10266.  
  10267. Add --replicate_row action
  10268.  
  10269. =item B<Version 1.6.1> (May 11, 2008)
  10270.  
  10271. Add --symlinks option as a shortcut to --action=rebuild_symlinks
  10272.  
  10273. =item B<Version 1.6.0> (May 11, 2008)
  10274.  
  10275. Add the custom_query action.
  10276.  
  10277. =item B<Version 1.5.2> (May 2, 2008)
  10278.  
  10279. Fix problem with too eager creation of custom pgpass file.
  10280.  
  10281. =item B<Version 1.5.1> (April 17, 2008)
  10282.  
  10283. Add example Nagios configuration settings (Brian A. Seklecki)
  10284.  
  10285. =item B<Version 1.5.0> (April 16, 2008)
  10286.  
  10287. Add the --includeuser and --excludeuser options. Documentation cleanup.
  10288.  
  10289. =item B<Version 1.4.3> (April 16, 2008)
  10290.  
  10291. Add in the 'output' concept for future support of non-Nagios programs.
  10292.  
  10293. =item B<Version 1.4.2> (April 8, 2008)
  10294.  
  10295. Fix bug preventing --dbpass argument from working (Robert Treat).
  10296.  
  10297. =item B<Version 1.4.1> (April 4, 2008)
  10298.  
  10299. Minor documentation fixes.
  10300.  
  10301. =item B<Version 1.4.0> (April 2, 2008)
  10302.  
  10303. Have 'wal_files' action use pg_ls_dir (idea by Robert Treat).
  10304. For last_vacuum and last_analyze, respect autovacuum effects, add separate
  10305. autovacuum checks (ideas by Robert Treat).
  10306.  
  10307. =item B<Version 1.3.1> (April 2, 2008)
  10308.  
  10309. Have txn_idle use query_start, not xact_start.
  10310.  
  10311. =item B<Version 1.3.0> (March 23, 2008)
  10312.  
  10313. Add in txn_idle and txn_time actions.
  10314.  
  10315. =item B<Version 1.2.0> (February 21, 2008)
  10316.  
  10317. Add the 'wal_files' action, which counts the number of WAL files
  10318. in your pg_xlog directory.
  10319. Fix some typos in the docs.
  10320. Explicitly allow -v as an argument.
  10321. Allow for a null syslog_facility in the 'logfile' action.
  10322.  
  10323. =item B<Version 1.1.2> (February 5, 2008)
  10324.  
  10325. Fix error preventing --action=rebuild_symlinks from working.
  10326.  
  10327. =item B<Version 1.1.1> (February 3, 2008)
  10328.  
  10329. Switch vacuum and analyze date output to use 'DD', not 'D'. (Glyn Astill)
  10330.  
  10331. =item B<Version 1.1.0> (December 16, 2008)
  10332.  
  10333. Fixes, enhancements, and performance tracking.
  10334. Add performance data tracking via --showperf and --perflimit
  10335. Lots of refactoring and cleanup of how actions handle arguments.
  10336. Do basic checks to figure out syslog file for 'logfile' action.
  10337. Allow for exact matching of beta versions with 'version' action.
  10338. Redo the default arguments to only populate when neither 'warning' nor 'critical' is provided.
  10339. Allow just warning OR critical to be given for the 'timesync' action.
  10340. Remove 'redirect_stderr' requirement from 'logfile' due to 8.3 changes.
  10341. Actions 'last_vacuum' and 'last_analyze' are 8.2 only (Robert Treat)
  10342.  
  10343. =item B<Version 1.0.16> (December 7, 2007)
  10344.  
  10345. First public release, December 2007
  10346.  
  10347. =back
  10348.  
  10349. =head1 BUGS AND LIMITATIONS
  10350.  
  10351. The index bloat size optimization is rough.
  10352.  
  10353. Some actions may not work on older versions of Postgres (before 8.0).
  10354.  
  10355. Please report any problems to check_postgres@bucardo.org
  10356.  
  10357. =head1 AUTHOR
  10358.  
  10359. Greg Sabino Mullane <greg@endpoint.com>
  10360.  
  10361.  
  10362. =head1 NAGIOS EXAMPLES
  10363.  
  10364. Some example Nagios configuration settings using this script:
  10365.  
  10366. define command {
  10367. command_name check_postgres_size
  10368. command_line $USER2$/check_postgres.pl -H $HOSTADDRESS$ -u pgsql -db postgres --action database_size -w $ARG1$ -c $ARG2$
  10369. }
  10370.  
  10371. define command {
  10372. command_name check_postgres_locks
  10373. command_line $USER2$/check_postgres.pl -H $HOSTADDRESS$ -u pgsql -db postgres --action locks -w $ARG1$ -c $ARG2$
  10374. }
  10375.  
  10376.  
  10377. define service {
  10378. use generic-other
  10379. host_name dbhost.gtld
  10380. service_description dbhost PostgreSQL Service Database Usage Size
  10381. check_command check_postgres_size!256000000!512000000
  10382. }
  10383.  
  10384. define service {
  10385. use generic-other
  10386. host_name dbhost.gtld
  10387. service_description dbhost PostgreSQL Service Database Locks
  10388. check_command check_postgres_locks!2!3
  10389. }
  10390.  
  10391. =head1 LICENSE AND COPYRIGHT
  10392.  
  10393. Copyright (c) 2007-2012 Greg Sabino Mullane <greg@endpoint.com>.
  10394.  
  10395. Redistribution and use in source and binary forms, with or without
  10396. modification, are permitted provided that the following conditions are met:
  10397.  
  10398. 1. Redistributions of source code must retain the above copyright notice,
  10399. this list of conditions and the following disclaimer.
  10400. 2. Redistributions in binary form must reproduce the above copyright notice,
  10401. this list of conditions and the following disclaimer in the documentation
  10402. and/or other materials provided with the distribution.
  10403.  
  10404. THIS SOFTWARE IS PROVIDED BY THE AUTHOR "AS IS" AND ANY EXPRESS OR IMPLIED
  10405. WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
  10406. MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
  10407. EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
  10408. EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
  10409. OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  10410. INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
  10411. CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
  10412. IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY
  10413. OF SUCH DAMAGE.
  10414.  
  10415. =cut
  10416.  
  10417. # vi: tabstop=4 shiftwidth=4 expandtab
Add Comment
Please, Sign In to add comment
Public Pastes
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!