17:00:32 <tardyp> #startmeeting weekly
17:00:32 <bb-supy`> Meeting started Tue Jan 24 17:00:32 2017 UTC and is due to finish in 60 minutes.  The chair is tardyp. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:32 <bb-supy`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:32 <bb-supy`> The meeting name has been set to 'weekly'
17:00:41 <tardyp> anyone here?
17:00:45 <rutsky> hi!
17:00:56 <tardyp> #topic introduction
17:01:10 <skelly> ah yes, Tuesday
17:01:10 <tardyp> https://titanpad.com/buildbot-agenda
17:01:41 <tardyp> anybody else?
17:03:07 <bb-github> [13buildbot] 15noc0lour commented on pull request #2635 1402c5878: Yep, thanks for your comments! 02https://git.io/vMNfl
17:03:30 <noc0lour> I guess I'm here this week
17:04:40 <tardyp> cool
17:04:51 <tardyp> #topic week in reiew
17:04:54 <tardyp> #undo
17:04:54 <bb-supy`> Removing item from minutes: <ircmeeting.items.Topic object at 0x806891950>
17:05:00 <tardyp> #topic week in review
17:05:22 <tardyp> this week we got lots of activity on the py3 side
17:05:30 <tardyp> and we closed 2340! yeah!
17:06:40 <tardyp> I'm working on enabling status on py3 to know how many tests are left to pass
17:06:56 <tardyp> my last run was ~300 on 2400
17:07:09 <rutsky> tardyp: where this status will be seen?
17:07:27 <rutsky> are those errors mostly syntax or non-syntax?
17:07:42 <tardyp> it is sent to event.buildbot.net, and I'm working on a dashbord in kibana to display tht
17:07:46 <tardyp> very simple thought
17:08:33 <rutsky> huh, sounds interesting!
17:08:43 <tardyp> https://github.com/buildbot/buildbot/pull/2607
17:08:52 <tardyp> the last commit is about that
17:09:18 <tardyp> #topic releases
17:09:46 <tardyp> I've got some news from sa2ajj. he is fine, but just quite busy
17:10:32 <tardyp> #info no release for eight until sa2ajj can find more time or somebody take over
17:11:12 <tardyp> #info 0.9.5 will be released next week to keep up with beg of month release plan
17:11:29 <tardyp> #topic trac2github
17:11:45 <tardyp> do somebody wants to take this task?
17:12:31 <rutsky> I will not have enough time for that
17:12:53 <noc0lour> I'd be totally in for that, but currently my plate is full
17:13:16 <tardyp> I'd probably look at it after the release if nobody step in
17:13:40 <tardyp> #topic single file distribution
17:13:46 <skelly> what about 0.9.4 before 0.9.5? ;)
17:13:46 <tardyp> rutsky: anything to report?
17:14:21 <tardyp> skelly probably a good idea to first release 0.9.4
17:14:34 <rutsky> if 0.9.5 will be in the next week, I don't think I will be able to do it yet
17:15:12 <rutsky> I need to recover e2e tests in order to check single-file distributive
17:15:29 <rutsky> maybe this time e2e will be good enough for merging in the main tree :)
17:15:36 <tardyp> well, we are not in a rush. but 0.9.4 will have py3 support for worker
17:15:54 <tardyp> ah that's why you are working on that.
17:15:56 <rutsky> nice!
17:16:29 <rutsky> yes, I think if single-file binaries will not be autotested they became broken quite soon
17:16:59 <rutsky> next topic?
17:17:15 <tardyp> #topic nine.bb.net
17:17:42 <tardyp> with the whole load of PR merged for py3 we got ansible scripts running every hour, and restarting the master
17:17:43 <rutsky> is it about build failures?
17:18:24 <tardyp> and the restart did not work well, so this meant lots of nine.bb.net downtime
17:18:40 <tardyp> so I disabled automatic deployment for this server
17:18:42 <skelly> would be nice if we could have it follow tags instead
17:18:56 <tardyp> we need to have a more stable method to update it
17:19:05 <tardyp> skelly: I think it is good that it follows master
17:19:17 <tardyp> it allows to see ui regressions for example (got one last week)
17:19:19 <skelly> could do clean restarts
17:19:24 <skelly> that's what I do here
17:19:50 <tardyp> yes, I guess there is not much to do, but this is something to implement
17:20:06 <tardyp> I was also thinking about a/b multimaster
17:20:09 <rutsky> what are typical reasons of failures?
17:20:21 <rutsky> a/b multimaster?
17:20:46 <skelly> running multimaster live would be nice as a live proof it works
17:20:50 <tardyp> I mean green blue deployment
17:21:16 <tardyp> during the upgrade you keep one master alive while you restart the other
17:21:41 <tardyp> when the new one is updated, then the first one can reboot
17:22:09 <tardyp> but this needs pgsql + crossbar, and all that stuff. Its a bit of work, but its clearly doable
17:22:48 <tardyp> I think trace2gh is more important
17:23:03 <tardyp> I'll just run the ansible script manually when there is no load
17:23:28 <tardyp> #topic buildbot builder failures
17:23:38 <tardyp> rutsky: can you explain?
17:24:05 <rutsky> sure
17:24:16 <rutsky> take a look at commits and their test status: https://github.com/buildbot/buildbot/commits/master
17:24:47 <rutsky> I never saw builds like https://nine.buildbot.net/#/builders/3/builds/949 successful
17:25:11 <rutsky> and I always noticed same error: fatal: Could not parse object '36fbfefb99526a79e18864b2297dbebf4bbae566'
17:25:31 <skelly> the depth might not be high enough
17:25:32 <rutsky> probably because of "git clone --branch master --depth 1 https://github.com/buildbot/buildbot ." above
17:25:52 <rutsky> yes, I think it's always not enough :)
17:26:07 <tardyp> probably we can increase it to 100
17:26:11 <noc0lour> Yeah, if the commit you are looking for is not the newest one you"r done
17:26:31 <tardyp> it might be hardcoded into bbtravis
17:26:48 <rutsky> I don't see --depth in  	.bbtravis.yml
17:26:57 <skelly> what about shallow?
17:27:01 <rutsky> well, there are
17:27:03 <rutsky> git:
17:27:05 <rutsky> depth: 300
17:27:20 <skelly> unfortunately that doesn't work for bbtravis
17:27:23 <tardyp> its not taken in account
17:27:34 <skelly> I think I know a way to fix that actually
17:28:42 <nicolas17> oh
17:29:20 <nicolas17> it's cloning with --depth 1 and trying to use commit X, when the latest in the repo (= the only one fetched by the clone) is already X+1 or more?
17:30:08 <bb-trac> [trac] #3667/undecided (v:master) created by rutsky (buildbot builder on nine.buildbot.net always fails due to shallow clone) http://trac.buildbot.net/ticket/3667
17:30:28 <rutsky> tardyp: skelly: I opened ticket for that, lets move forward
17:30:38 <tardyp> its due to https://github.com/buildbot/buildbot_travis/commit/4161ac8416ff4a50bd2596f5bb739b26b5c7b992
17:30:58 <tardyp> I think the topic list is finished
17:31:08 <tardyp> anything more to talk about?
17:31:10 <skelly> just hit 30 minutes too
17:31:22 <rutsky> I closing then     http://trac.buildbot.net/ticket/3661
17:31:26 <tardyp> #endmeeting