18:14:54 <h01ger> #startmeeting
18:14:54 <MeetBot> Meeting started Wed Apr 26 18:14:54 2017 UTC.  The chair is h01ger. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:14:54 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
18:15:13 <h01ger> #topic say hi / agenda planning
18:15:38 <h01ger> the reoocuring agenda is
18:15:48 <h01ger> * short intro, why are we here (aka: say hi)
18:15:48 <h01ger> * jenkins.d.n status
18:15:48 <h01ger> * jenkins.d.o migration next steps (see below for details)
18:15:48 <h01ger> * jenkins sprint in 2017?
18:15:49 <h01ger> * AOB
18:15:49 <h01ger> * thanks to profitbricks for hosting
18:16:21 <tnnn> hi
18:17:24 <tnnn> I'm here because I like meetings and maybe I can be useful to qa team :)
18:17:32 <h01ger> hi tnnn :)
18:18:07 <tnnn> ...and I'm afraid that's all I can report for now ;>
18:18:14 * h01ger is here to form a team to keep jenkins running in future
18:18:49 <h01ger> #topic jenkins.d.n status
18:19:42 <h01ger> hopefully much better now that we moved 120 out of 162 constantly running reproducible builds jobs away from jenkins (the service) to being started by reproducible_build_service.sh started by systemd
18:20:11 <h01ger> the UI feels much more responsive to me, login is faster, triggered builds are immediatly run again
18:21:15 <mapreri> o/ (sorry I am a bit late)
18:21:19 <mapreri> I also noticed it is a tad more responsive, fwiw! :)
18:21:29 <h01ger> those remaining 42 jobs (so that 120 out of 162 become 162 out of 162) will be moved soon
18:21:38 <h01ger> hey mapreri
18:21:51 <tnnn> Yup, seems nice h01ger - thanks for working on that
18:22:39 <h01ger> unrelatedy i made the chroot-installation test jobs complain about transitional packages in sid now,but made it ignore some (to get better results for edu tests), which i will reenable once buster development has begun
18:22:54 <h01ger> plus i added chroot-installation tests for the Debian Parl blends
18:24:01 <h01ger> anything else about jenkins.d.n?
18:26:23 <h01ger> #topic status jenkins.d.org migration
18:26:30 <h01ger> "nothing new", afaik
18:28:16 <mapreri> I keep updating the plugins every so often, but yeah, nothing's happening right now
18:28:49 <tnnn> ...I guess that in this case no news is (rather) good news
18:28:57 <mapreri> I should try to set some hard deadline to do something myself, perhaps.
18:29:01 <mapreri> tnnn: not really "good", no.
18:29:43 <h01ger> it's getting a bit more "urgency" in the back of my mind as i would like to run a lot more chroot-installationtests (maybe even on i386 and arm*) and scaling them to a multinode setup would be needed for that (and then if i really touch them again, i can also migrate them while at it)
18:32:32 <h01ger> oh, /me has a simple unix question: pmap $pid_of_a_bash_process tells me bash is using 23mb while dash uses 4mb… considering "ps fax | grep -c bash" results in 600 bash processes, i'm considering at least moving _worker.sh to dash…
18:32:54 <tnnn> mapreri: yeah, you are right. I'm in damage control mode atm which means that no breakeages is good news to me.
18:33:03 <mapreri> tnnn: heh
18:33:05 <h01ger> or is that not worth bothering on a 64gb ram machine?
18:33:30 <mapreri> 23 MB vs 4 MB… are you considering the same script?
18:33:56 <danielsh> how much of those 23mb are shared among bash instances?
18:35:48 <h01ger> thats measured by starting bash in an xterm, then starting dash from that bash and then running pmap on those idling processes
18:35:58 <h01ger> danielsh: thats exactly my question…
18:37:00 <h01ger> anyhow… next topic, we dont need to finish this discussion in this in the meeting…
18:37:26 <h01ger> #info & see the TODO in git for a more detailed status
18:37:33 <h01ger> #topic jenkins sprint in 2017
18:37:59 <mapreri> october? *blink*
18:38:06 <h01ger> https://wiki.debian.org/ReproducibleBuilds/HamburgHackathon2017 - not exactly a jenkins sprint but you are much invited anyway to come around and hack on jenkins ;)
18:38:59 <KGB-2> 03Holger Levsen 05master c8042a4 06jenkins.debian.net 10TODO add todays IRC meeting and next months * 14https://deb.li/ix4t7
18:38:59 <KGB-2> 03Holger Levsen 05master 78a73e1 06jenkins.debian.net 10TODO suggest october for the jenkins sprint * 14https://deb.li/4Lcl
18:39:01 <mapreri> I suppose that if I were to organize something, that something might even happen.
18:39:07 <h01ger> mapreri: /me likes
18:39:12 <tnnn> I won't be able to make it to Hamburg. Still, I'd be happy to attend.
18:39:13 <h01ger> *very much*
18:39:34 <h01ger> mapreri: october in italy is also much better than hamburg in october
18:39:36 <tnnn> to attend something if that something will get organised ;)
18:39:59 <mapreri> h01ger: that's surely true
18:40:09 <h01ger> #topic Any Other Business
18:40:29 <h01ger> (not that i want to stop discussing sprint plans…!)
18:40:31 <tnnn> Just about that munin plugin
18:40:39 <h01ger> which?
18:40:46 <mapreri> jstat, i suppose
18:40:50 <tnnn> https://jenkins.debian.net/munin/debian.net/jenkins.debian.net/jenkins_jstats.html
18:41:03 <tnnn> oh, yeah - sorry, thats what I meant :)
18:41:26 <tnnn> jstat - it's configured to show garbage collector stats
18:42:10 <tnnn> to me, it looks that jvm gc is doing its job rather well
18:42:11 <h01ger> we should switch that to the upstream java plugin from munin contrib
18:42:32 <tnnn> there was a jvm restart on 21
18:42:37 <h01ger> but after i wrote it i was to annoyed having basically rewritten existing code that i didnt bother…
18:43:02 <h01ger> "annoyed" is a bit strong…
18:43:50 <tnnn> h01ger: Can you give me a link to it?
18:44:41 <tnnn> While showing gc stats is nice, we could probably configure it to show a few different things as well
18:45:04 <h01ger> tnnn: https://github.com/munin-monitoring/contrib/tree/master/plugins/jvm
18:45:48 <h01ger> and jenkins.d.n.git/hosts/jenkins/etc/munin/plugins/jenkins_jstats
18:46:22 <h01ger> and then this jenkins_jstats plugin needs to be removed from out repo and the jvm one added… patches for that much welcome
18:46:42 <h01ger> (should be easy and thus good for getting more familar with the repo)
18:46:58 <tnnn> that munin contrib one mentions sun java 5 in few places :)
18:47:19 <helmut> so we're confident that logparser was the main leaker now?
18:47:42 <tnnn> So we should check first if it's really better that the one you wrote
18:48:09 <h01ger> helmut: i guess thats the case
18:48:23 <h01ger> tnnn: when i checked they were pretty similar
18:48:33 <tnnn> I belive I can try doing that in the upcoming week (few holidays here in Poland)
18:48:42 <tnnn> mhm, ok
18:49:52 <mapreri> helmut: empirically decided, anyway
18:49:54 <h01ger> #info next jenkins/qa meeting on May 25th at 18utc (=4th wednesday of the month)
18:50:02 <mapreri> we haven't seen those anymore since then, so.…
18:50:22 <h01ger> thank you all for contributing to Debian/jenkins/QA!
18:50:26 <h01ger> #endmeeting