Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- 18:34 -!- tc424 [~steved@2001:470:1f1d:e3b::34] has joined #winswitch
- 18:34 -!- Topic for #winswitch: winswitch 0.12.23 | xpra 2.3.2 | please try updating to the newest version before reporting bugs
- 18:34 -!- Topic set by totaam [~antoine@users.devloop.org.uk] [Tue Jun 26 19:01:38 2018]
- 18:34 [Users #winswitch]
- 18:34 [@totaam ] [ Benjojo ] [ davidl_ ] [ icasdri ] [ qyliss ] [ tc424 ]
- 18:34 [ [Neurotic]] [ bolt ] [ docmax ] [ idupree7 ] [ raso2 ] [ tribut ]
- 18:34 [ `brain` ] [ Boobuigi] [ ds_shadof ] [ izibi ] [ rootman ] [ tsto ]
- 18:34 [ akaWolf ] [ bug2000 ] [ early ] [ jbrett ] [ rudi_s ] [ wodencafe]
- 18:34 [ Akronym ] [ CcxWrk ] [ elibrokeit] [ kugel ] [ shadoxx ] [ yomii ]
- 18:34 [ amir ] [ Chaz6 ] [ EliDupree ] [ luke-jr ] [ smo ]
- 18:34 [ aterius ] [ copec ] [ femme ] [ mtgx` ] [ Some_Person]
- 18:34 [ atrus ] [ darvon ] [ grazzolini] [ ohama ] [ stux ]
- 18:34 [ Avengence ] [ davidl ] [ hosenknopf] [ pfallenop] [ tbr ]
- 18:34 -!- Irssi: #winswitch: Total of 50 nicks [1 ops, 0 halfops, 0 voices, 49 normal]
- 18:34 -!- Channel #winswitch created Wed Nov 2 07:11:40 2011
- 18:34 -!- Irssi: Join to #winswitch was synced in 1 secs
- 18:36 < tc424> hi all, new to xpra, have installed latest version on server (ubuntu 16.04) and client (18.04), but a bit surprised at the latency versus direct X forwarding, given this is 1Gb/s connection
- 18:37 < tc424> forcing rgb/png helps a bit, but not as much as I'd hoped, and I don't know why it's not picked automatically given link speed. anything obvious I may have overlooked?
- 18:39 <@totaam> tc424: set --speed=100 instead
- 18:39 <@totaam> or force just rgb
- 18:39 <@totaam> there should be zero latency on a gigabit link
- 18:40 <@totaam> except for link latency itself, obviously
- 18:41 < tc424> ah, will try that, thanks - was a bit shocked when moving from flaky wifi to ethernet didn't seem to help!
- 18:43 <@totaam> wifi is usually terrible, but better in 2.4
- 18:43 <@totaam> latency jitter is hard to deal with
- 18:43 <@totaam> on ethernet gigabit, you can forward 4k windows as long as they don't update too quickly
- 18:43 <@totaam> for 4k video, you need hardware encoding
- 18:44 < tc424> ok, got --speed=100, was briefly ok when I first connected, now seem frame delay around ~250ms!
- 18:44 < tc424> something is obviously really wrong :(
- 18:44 <@totaam> that's definitely not right
- 18:45 <@totaam> tc424: see http://xpra.org/trac/wiki/ReportingBugs
- 18:46 <@totaam> those are the kinds of details we need to investigate anything
- 18:46 < tc424> thanks, it could be something local, my machines tend to end up a bit hacked up, but the client machine at least should be pretty clean
- 18:46 < tc424> it's basically an Intel SoC box being used as an X terminal
- 18:47 <@totaam> SoC - as in low power?
- 18:48 < tc424> yeah, fairly
- 18:48 < tc424> but fully accelerated gfx
- 18:48 < tc424> seeing may 10-20% cpu utilization
- 18:49 < tc424> straight X using XDMCP is fine, pretty much indistinguishable from the "server" console
- 18:50 <@totaam> tc424: make sure opengl rendering is enabled
- 18:50 < tc424> I /think/ it is
- 18:50 <@totaam> it should say in the client output
- 18:50 < tc424> 2018-07-07 18:42:28,299 OpenGL enabled with Mesa DRI Intel(R) HD Graphics 500 (Broxton 2x6)
- 18:50 < tc424> 2018-07-07 18:42:27,684 No OpenGL_accelerate module loaded: No module named OpenGL_accelerate
- 18:50 < tc424> 2018-07-07 18:42:28,241 Warning: vendor 'Intel Open Source Technology Center' is greylisted,
- 18:50 <@totaam> OK, then it is
- 18:50 <@totaam> I'm out of ideas
- 18:51 <@totaam> 250ms is way too much
- 18:51 < tc424> now at 875ms!
- 18:51 < tc424> the batch delay keeps climbing
- 18:51 <@totaam> please file a ticket
- 18:51 <@totaam> with at least "xpra info"
- 18:52 < tc424> OK, will do, is there a way to monitor batch delay computations in real time as they happen?
- 18:52 <@totaam> "-d stats" will show more details
- 18:52 <@totaam> (server side)
- 18:52 <@totaam> it's not very readable
- 18:52 <@totaam> coule be bandwidth limits triggering it
- 18:53 <@totaam> try adding --bandwidth-limit=0
- 18:53 < tc424> will do, it's not set according to the GUI, but just in case...
- 18:57 -!- docmax1 [~docmax@x4e3376e8.dyn.telefonica.de] has joined #winswitch
- 19:00 -!- docmax [~docmax@x55b26367.dyn.telefonica.de] has quit [Ping timeout: 256 seconds]
- 19:00 -!- docmax1 is now known as docmax
- 19:05 < tc424> OK I see the same thing just running the client + server on the same machine ..
- 19:06 < tc424> network 1-3ms, damage ~90ms, batch ~110ms
- 19:25 <@totaam> tc424: running on the same machine should use mmap, which means all delays should be near zero there
- 20:54 -!- Boobuigi [~weechat@unaffiliated/boobuigi] has quit [Ping timeout: 240 seconds]
- 21:31 -!- Boobuigi [~weechat@unaffiliated/boobuigi] has joined #winswitch
- 21:33 < tc424> totaam: yeah, that's what I thought :(
- 21:51 < tc424> totaam: will open proper ticket tomorrow hopefully when brain is less broiled, but in the meantime https://pastebin.com/nuZjFBH0 shows the escalating batch delay
- 21:53 < tc424> 2018-07-07 21:38:29,099 record_latency: took 2098.1 ms round trip, 2098.1 for echo, 285.0 for decoding of 273 pixels, 1145 bytes sent over the network in 1813.0 ms, 1813.0 ms for echo
- 21:53 < tc424> ..that doesn't look healthy..
- 22:11 -!- davidl_ [~usrname@fsf/member/usrname] has quit [Ping timeout: 265 seconds]
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement