armbian-monitoring

#!/bin/bash
#
# armbian-monitoring
#
# This script is used to configure armbianmonitor behaviour.
# It will ask the user whether to activate monitoring or not,
# whether to enable debug monitoring and also how to deal with
# connected disks. In fact it walks through the list of available
# disks, checks them, tries to patch hddtemp.db if necessary
# and provides a proposal for /etc/armbianmonitor/disks.conf
# when a new disk is found.
#
# In case monitoring should be activated the following file
# will be created: /etc/armbianmonitor/start-monitoring. If
# debug output has been chosen, then DEBUG will be written to
# the file.
#
# The script will install smartmontools/gdisk if not already
# installed and patches smartmontools' update-smart-drivedb
# script if necessary. For disks the 'device model' will be
# shown but internally we rely always on the GUID. This is the
# key for entry in /etc/armbianmonitor/disks.conf
#
# When the script exits and the user activated monitoring it
# recommends doing a restart since on the next reboot the
# setup-armbian-monitoring-environment script will configure
# monitoring sources and decides based on the existence and
# contents of /etc/armbianmonitor/start-monitoring whether
# rpimonitord should be started or not.
#
# The format of /etc/armbianmonitor/disks.conf is as follows:
#
# ${GUID}:${smartctl prefix}:${temp call}:${CRC attribute}:${LCC attribute}
#
# Two examples:
#
# A57BF307-7D82-4783-BD1D-B346CA8C195B:::199:193 # WD HDD on SATA
# F8D372DC-63DB-494B-B802-87DC47FAD4E1:sat::199: # SSD in USB enclosure
#
# - GUID is the GUID as determined by gdisk
# - "smartctl prefix" can be empty or should be the the necessary prefix for
#   USB disks, eg. '-d sat' or '-d usbjmicron' and so on -- please have a
#   look at https://www.smartmontools.org/wiki/Supported_USB-Devices
# - "temp call" when being omitted indicates that hddtemp should be used.
#   Otherwise it should contain the complete command line ('DISK' will be
#   dynamically replaced by the device node when the actual monitoring
#   happens), for example:
#   /sbin/hdparm -C DISK | egrep -q "standby|sleeping" || /usr/sbin/smartctl -d sat -a DISK | awk -F" " '/Temperature_Cel/ {printf $10}'
# - 'CRC attribute': The decimal value of the S.M.A.R.T. attribute that
#   is used to store the count of checksum errors between disk and host
#   controller (might be omitted if the drive doesn't support it)
# - 'LCC attribute': The decimal value of the S.M.A.R.T. attribute that
#   should contain the load cycle counter value (might be omitted
#   if the drive doesn't support it)
#
# TODO:
#
# - develop main functionality ;) asking the user regarding monitoring
# - deal with 'SMART overall-health self-assessment test result:'

Main() {
    PreRequisits

    # check whether user runs rpimonitord on his own or we activated it
    if [ -f /etc/armbianmonitor/start-monitoring ]; then
        # we should already provide monitoring, check whether DEBUG
        # is also set
        ArmbianMonitoring=TRUE
        read DebugMode </etc/armbianmonitor/start-monitoring
    fi

    # check whether rpimonitord is running and compare with ${ArmbianMonitoring}
    # In case the user chose to run rpimonitord on his own, we skip the config
    # part and only output disk info
    :

    # check available disk devices
    CheckDisks
} # Main

CheckDisks() {
    # This function walks through all block devices whose name starts with sd* and
    # then gets the name hddtemp expects, the model name from smartctl, looks whether
    # the drive only lists one temperature value and patches hddtemp.db if necessary
    # and also tries to get CRC and LCC S.M.A.R.T. attributes to provide the user
    # with the necessary config file contents for /etc/armbianmonitor/disks.conf:

    ls /sys/block/sd* >/dev/null 2>&1 || exit 0

    for i in /sys/block/sd* ; do
        DeviceNode=/dev/${i##*/}
        # get GUID/UUID for disk and check whether a partition table is existent. If
        # not GUID will always be random
        gdisk -l ${DeviceNode} >"${MyTempDir}/gdisk.txt"
        GUID=$(awk -F" " '/^Disk identifier/ {print $4}' <"${MyTempDir}/gdisk.txt")
        CountOfUnavailablePartitionTables=$(grep ': not present' "${MyTempDir}/gdisk.txt" | wc -l)
        if [ ${CountOfUnavailablePartitionTables} -eq 4 ]; then
            echo -e "\nSkipping ${DeviceNode} due to missing partition table. Use parted to create one."
            break
        else
            echo -e "\nExamining ${DeviceNode} with GUID ${GUID}\c"
        fi

        # get name hddtemp needs
        HddtempName="$(hddtemp --debug ${DeviceNode} | awk -F": " '/^Model: / {print $2}' | \
            cut -c-40 | sed 's/^[ \t]*//;s/[ \t]*$//')"
        # store smartctl output in temporary file
        smartctl -q noserial -s on -a ${DeviceNode} >"${MyTempDir}/smartctl.txt" 2>&1
        DeviceModel="$(awk -F": " '/^Device Model/ {print $2}' <"${MyTempDir}/smartctl.txt" | \
            sed 's/^[ \t]*//;s/[ \t]*$//')"
        if [ "X${DeviceModel}" = "X" ]; then
            # Reading S.M.A.R.T. failed, we try autodetect mode iterating through all
            # known smartctl modes (-d auto|sat|usbcypress|usbjmicron|usbprolific|usbsunplus)
            SMARTPrefix="$(CheckSMARTModes ${DeviceNode} 2>/dev/null)"
            if [ "X${SMARTPrefix}" = "X" ]; then
                # we can't query the disk. Time to give up
                echo -e ". Unable to query the disk through S.M.A.R.T.\nPlease investigate manually using smartctl\n"
                break
            fi
        fi

        # user feedback
        if [ "X${SMARTPrefix}" = "X" ]; then
            echo -e " (accessible through S.M.A.R.T.)"
        else
            echo -e " (can be queried with \"-d ${SMARTPrefix}\" through S.M.A.R.T.)"
        fi

        # check for CRC and LCC attributes
        CRCAttribute=$(awk -F" " '/CRC_Error_Count/ {print $1}' <"${MyTempDir}/smartctl.txt")
        LCCAttribute=$(grep -i "load.cycle" "${MyTempDir}/smartctl.txt" | awk -F" " '{print $1}')

        # check whether /etc/hddtemp.db should be patched
        grep -q "${HddtempName}" /etc/hddtemp.db
        if [ $? -ne 0 ]; then
            # No entry into hddtemp database, we've a look whether there's a 'temperature'
            # attribute available (we take the 1st we find) and if that's the case we use this
            DiskTemp=$(awk -F" " '/Temperature/ {print $1}' <"${MyTempDir}/smartctl.txt" | head -n1)
            if [[ ${DiskTemp} -gt 0 ]]; then
                echo -e "\"${HddtempName}\" ${DiskTemp} C \"${DeviceModel}\"" >>/etc/hddtemp.db
                echo -e "\nAdded disk \"${DeviceModel}\"/\"${HddtempName}\" to /etc/hddtemp.db using S.M.A.R.T. attribute ${DiskTemp}\nbased on the following available thermal values:"
                grep "Temperature" "${MyTempDir}/smartctl.txt"
                # check hddtemp result
                HddtempResult=$(hddtemp -n ${DeviceNode} | awk -F" " '{print $1}')
                if [ "X${HddtempResult}" = "X${DeviceNode}:" ]; then
                    # hddtemp isn't able to query the disk
                    HddtempStatus="does not work. Please check with smartctl and adjust config accordingly"
                    echo -e "\nhddtemp output: $(hddtemp ${DeviceNode})"
                    echo -e "\nIt seems we can not rely on hddtemp to query this disk. Please try smartctl instead\n"
                else
                    HddtempStatus="will work"
                    echo -e "\nhddtemp output: ${HddtempResult})"
                    echo -e "\nIn case this seems not to be correct please adjust /etc/hddtemp.db manually\n"
                fi
            else
                HddtempStatus="does not work. Please check with smartctl and adjust config accordingly"
            fi
        else
            HddtempStatus="will work"
        fi

        # check for firmware updates
        FirmwareUpdate="$(grep "^http" "${MyTempDir}/smartctl.txt")"

        # Check whether the disk (based on GUID) is already configured in our config file
        # /etc/armbianmonitor/disks.conf or not

        grep -q "^${GUID}:" /etc/armbianmonitor/disks.conf >/dev/null 2>/dev/null
        case $? in
            0)
                # already listed, we provide just infos:
                echo -e "Disk is already configured by the following monitoring config:\n$(grep "^${GUID}:" /etc/armbianmonitor/disks.conf)\n"
                ;;
            *)
                # new disk, we recommend an entry for /etc/armbianmonitor/disks.conf
                echo -e "Disk not configured for monitoring. We were able to extract the following information:\nGUID: ${GUID}"
                if [ "X${SMARTPrefix}" != "X" ]; then
                    echo -e "QueryMode: -d ${SMARTPrefix}"
                fi
                echo -e "hddtemp: ${HddtempStatus}\nCRC attribute: ${CRCAttribute}\nLCC Attribute: ${LCCAttribute}"
                case ${HddtempStatus} in
                    "will work")
                        echo -e "If you want to monitor the disk please add to /etc/armbianmonitor/disks.conf:\n${GUID}:${SMARTPrefix}::${CRCAttribute}:${LCCAttribute}"
                        ;;
                    *)
                        echo -e "Proposal for /etc/armbianmonitor/disks.conf:\n${GUID}:${SMARTPrefix}:FIXME:${CRCAttribute}:${LCCAttribute}"
                        echo -e "You have to figure out how to query the disk for its thermal sensor."
                        echo -e "Please check the output of hddtemp --debug ${DeviceNode} and smartctl\n"
                        ;;
                esac
                ;;
        esac
        if [ "X${FirmwareUpdate}" != "X" ]; then
            echo -e "\nWARNING: A firmware update seems to be available:\n${FirmwareUpdate}\n"
        fi
    done
} # CheckDisks

CheckSMARTModes() {
    # This function tries to access USB disks through S.M.A.R.T. and returns the necessary
    # '-d' call as well as fills in ${MyTempDir}/smartctl.txt

    for i in auto sat usbcypress usbjmicron usbprolific usbsunplus ; do
        # user feedback
        echo -n "." >/dev/tty
        # query disk using the specific protocol
        echo -n "" >"${MyTempDir}/smartctl.txt"
        smartctl -q noserial -s on -d ${i} -a ${1} >"${MyTempDir}/smartctl.txt" 2>/dev/null
        DeviceModel="$(awk -F": " '/^Device Model/ {print $2}' <"${MyTempDir}/smartctl.txt" | \
            sed 's/^[ \t]*//;s/[ \t]*$//')"
        if [ "X${DeviceModel}" != "X" ]; then
            echo ${i}
            break
        fi
    done
} # CheckSMARTModes

PreRequisits() {
    # Ensure that we're running as root since otherwise querying SATA/USB disks won't work
    if [ "$(id -u)" != "0" ]; then
        echo "This script must be run as root" >&2
        exit 1
    fi

    export PATH=/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin
    unset LANG
    DISTROCODE=$(lsb_release -s -c)

    # check whether gdisk/smartctl are available and up to date
    echo -e "Check whether necessary software is available\c"
    which gdisk >/dev/null 2>&1 || (echo -e " Installing gdisk\c" ; apt-get -f -qq -y install gdisk)
    which smartctl >/dev/null 2>&1 || (echo -e " Installing smartmontools\c" ; apt-get -f -qq -y install smartmontools)
    echo -e " [done]\nUpdating smartmontools' drivedb\c"
    /usr/sbin/update-smart-drivedb >/dev/null 2>&1
    if [ $? -ne 0 -a "X${DISTROCODE}" = "Xwheezy" ]; then
        sed -i "/^SRCEXPR/{s#=.*#='http://sourceforge.net/p/smartmontools/code/HEAD/tree/\$location/smartmontools/drivedb.h?format=raw'#}" /usr/sbin/update-smart-drivedb
        /usr/sbin/update-smart-drivedb
    fi
    echo -e " [done]"
    CreateTempDir
} # PreRequisits

CreateTempDir() {
    # create a safe temporary dir
    MyTempDir=$(mktemp -d /tmp/${0##*/}.XXXXXX)
    if [ ! -d "${MyTempDir}" ]; then
        MyTempDir=/tmp/${0##*/}.$RANDOM.$RANDOM.$RANDOM.$$
        (umask 066 && mkdir ${MyTempDir}) || (echo "Failed to create temp dir. Aborting" >&2 ; exit 1)
    fi
    chmod 711 "${MyTempDir}"
    trap "rm -rf \"${MyTempDir}\" ; exit 0" 0 1 2 3 15
    for file in smartctl.txt gdisk.txt ; do
        touch "${MyTempDir}/${file}"
        chmod 644 "${MyTempDir}/${file}"
    done
} #CreateTempFiles

InstallRPiMonitor() {
    # Installs rpimonitord based on the official instructions from
    # http://rpi-experiences.blogspot.fr/p/rpi-monitor-installation.html
    apt-get -f -qq -y install apt-transport-https ca-certificates
    wget http://goo.gl/rsel0F -O /etc/apt/sources.list.d/rpimonitor.list
    apt-key adv --recv-keys --keyserver keyserver.ubuntu.com 2C0D3C0F
    apt-get -f -qq -y update
    apt-get -f -qq -y install rpimonitor
    # apt-get -f -qq -y upgrade
    /usr/share/rpimonitor/scripts/updatePackagesStatus.pl
} # InstallRPiMonitor

Main