[00:01] * thumper sighs [00:01] why oh why is the bot fubared === Ursinha is now known as Ursinha-afk [00:31] wallyworld_: thumper sinzui axw https://docs.google.com/a/canonical.com/document/d/1m9R2n6LPLNLGjdopcNkQYVG8D5V4FTyvc1vvn-9ZifM/edit# [00:32] davecheney: thanks, will look [00:32] davecheney: ta [00:32] thumper: bot is farked because or a tarmac issue [00:32] i think [00:32] :( [00:33] it has merged 4 branches to it's local tree but not pushed to lp [00:33] i'm trying to see if those branches actually passed the tests [00:33] haha [00:33] if so, i can push manually [00:36] thumper: ffs. [00:36] 2014-03-06 18:39:30 WARNING ConnectionReset calling 'Branch.last_revision_info', retrying [00:36] 2014-03-06 18:39:32 WARNING ConnectionReset calling 'Branch.last_revision_info', retrying [00:36] 2014-03-06 18:39:35 INFO Committing to: /home/tarmac/trees/src/launchpad.net/juju-core/ [00:37] so if it can't push to lp, it gives up and just merges locally [00:37] which results in complaining ever after [00:37] can the machine see launchpad? [00:37] hang on [00:37] it must be able to [00:37] it can and has - i think those are transient [00:37] otherwise it wouldn't merge... [00:37] yep [00:38] so transient errors = it's all broken [00:38] gah [00:38] i'll look for the branches that have passed testing in the log file and push those manually [00:48] wallyworld_: while you're in there [00:48] that's what she said [00:48] wallyworld_: want to 'go get github.com/juju/loggo' ? [00:48] sure [00:48] ta [00:49] wallyworld_: so since it failed once, it no longer tried to push to lp? [00:49] yep :-( [00:49] i had to push manually [00:49] there were 5 i think [00:49] trying to take a crack at fixing a few small bugs to get my head in the game with Go -- one problem, i'm not sure which part of the codebase this refers to on Github [00:50] https://bugs.launchpad.net/juju-core/+bug/1217868 [00:50] <_mup_> Bug #1217868: move from state.D to bson.D

[00:50] it says juju-core, but i'm not seeing a juju-core repo [00:50] bodie_: juju-core is still on launchpad [00:50] is there a doc that explains how things are put together? [00:50] ah [00:51] go get launchpad.net/juju-core/... [00:51] alrighty [00:51] wallyworld_: I have a physio appt during our standup [00:51] wallyworld_: when do you need to finish today? [00:52] around 3pm [00:53] thumper: when does your appt finish? we can just delay the standup [00:53] I should be back just after 3pm my time [00:53] noon your time [00:53] that suits me, I've got a school thing at 2pm [00:53] ok, so let's delay by 30 mins then [00:53] wallyworld_: sent you an email. [00:54] waigani: changing writeSampleConfig by itself is not sufficient - i was just point out where the config is used [00:54] you need to add a method to optionally add to the sample config [00:55] do this during set up [00:55] wallyworld_: right, thought so [00:55] and have writeSampleConfig use that [00:55] wallyworld_: perhaps by 45 minutes [00:55] as I need to get home [00:55] sure [00:55] axw: man of the moment! Sorry for spamming your inbox [00:56] * axw opens inbox [00:56] thumper: btw, i can see the bot landing your latest branch [00:56] running tests i mean [00:57] axw: waigani: standup will be delayed while thumper gets repaired. approx 45 minutes [00:57] hehe. no worries [00:57] wallyworld_: ok, cool [00:57] wallyworld_: we'll see if it pushes when done [00:57] yeah [00:58] waigani: re "do you think it is worth while making the validation optional but on by default in UpdateEnvironConfig" -- yes, it is already optional, by opening state with a nil policy [00:59] axw: haha, point taken [01:00] wallyworld_: okay 3:15. My girl asked me to watch her at assembly which is on now. So that works well. [01:00] cool :-) [01:04] waigani: first, when reopening don't use Initialize, use "Open". second, if you follow through to JujuConnSuite you can see how it opens a state connection. It uses juju.NewConn, passing in the environ [01:05] waigani: take a look in juju/conn.go to see what NewConn does, and how it opens a state conn [01:05] waigani: you can then do something similar in your test, but pass nil in as the policy [01:09] wallyworld_: is the bot finished? [01:09] * thumper doesn't see a push [01:09] * wallyworld_ looks [01:10] thumper: looks like push to lp has failed again [01:10] :( [01:10] thumper: i'll mention to martin [01:10] does the bot have a ssh key that lp knows about? [01:11] don't know, seems not i guess [01:11] actually [01:11] wallyworld_: how are you pushing to lp from there? [01:11] it must have [01:11] cause i'm pushing from the bot [01:11] so it just may be a small setup issue === Ursinha-afk is now known as Ursinha [02:15] * thumper joined standup hangout [02:16] axw: http://pastebin.ubuntu.com/7047615 [02:19] axw: standup time :-) === Tribaal_ is now known as Tribaal === bodie_ is now known as Guest31737 [09:54] https://code.launchpad.net/~rogpeppe/juju-core/511-instancepoller-aggregate/+merge/209505 [09:54] wwitzel3: https://code.launchpad.net/~rogpeppe/juju-core/511-instancepoller-aggregate/+merge/209505 [09:54] one time special offer [09:54] review our branch, again! [09:54] https://codereview.appspot.com/71490045/ [09:56] wwitzel3: https://code.google.com/p/irssi-libnotify/ [10:12] wwitzel3: it's a trap! [10:13] morning all [10:13] Morning natefinch [10:13] should we get the video online? [10:13] wwitzel3: yes please [10:16] natefinch: https://plus.google.com/hangouts/_/7acpj3k2j7gc114cifi39l6sf0 [10:36] wwitzel3: lp:~rogpeppe/juju-core/512-maas-bootstrap-bridge-utils [10:37] wwitzel3: lp:~rogpeppe/juju-core/512-maas-bootstrap-bridge-utils === bloodearnest_ is now known as bloodearnest [11:01] A long and difficult CL for someone [11:01] https://codereview.appspot.com/72450043 [11:52] mgz, does this look familiar @ 1.17.4 [11:52] unit-percona-cluster-6: 2014-03-07 11:51:52 ERROR juju runner.go:220 worker: exited "upgrader": cannot read tools metadata in tools directory: open /var/lib/juju/tools/1.17.4-trusty-amd64/downloaded-tools.txt: no such file or directory [11:52] I get that continually from all of my service units [12:09] jamespage: bug 1285901 [12:09] <_mup_> Bug #1285901: error starting unit upgrader on local provider

[12:10] mgz,good oh [12:10] I'll stop making noise [12:12] voidspace: https://github.com/golang/lint [12:16] adeuring, basically LGTM, ping me if any of my suggestions are unclear [12:16] adeuring, oh! did we want handling for JUJU_TEST_MODE as well? [12:17] fwereade: what do you mean with "handling for JUJU_TEST_MODE"? [12:18] adeuring, actually forget I said anything [12:18] ;) [12:18] adeuring, it'll cause more problems than it solves === smoser` is now known as smoser [14:25] dimitern: what version of jujuclient were you using yesterday when you deployed trying to reproduce our issue? I'm wondering if you had the newest version which I guess has a workaround for the issue [14:27] cjohnston, i'm using juju-core from trunk [14:28] dimitern: client... jujuclient [14:28] cjohnston, i don't quite know what's jujuclient [14:28] python-jujuclient ? [14:28] cjohnston, didn't have to install that [14:28] we also started pinning a specific revno from trunk a couple days ago as a work around for this [14:28] cjohnston: wat? [14:29] why not just use one of the released 1.17 ones [14:29] or are you talking about kapil's thing? [14:30] not sure what kapil's thing is, but he has been helping us with this issue [14:30] cjohnston: specifically, what do you mean by "jujuclient"? [14:30] python-jujuclient [14:30] AIUI it has a workaround for the api watcher issue we are hitting.. [14:30] ev, hazmat ^ [14:31] okay, so you are talking about kapil's thing [14:31] mgz, i had a workaround for the api server going awol and disconnecting all clients [14:31] in python jujuclient so that it recovered for extant watches and reconnected and restablished the watch [14:32] mgz, i think the question is if the underlying issue of the api server going awol is addressed [14:32] mgz: so I am trying to figure out how dimitern tested it, and it appears as though it would need to be retested without hazmat's workaround in order to try to reproduce.. [14:32] they've only given us steps to repo that include spinning up 15 manchines [14:32] however we have a number of people on our team who have been hitting it [14:32] so, dimitern has been struggling to actually get traction on it [14:32] mgz: that's how we produce the error [14:33] mgz, specifically there's a log entry in syslog showing mongodb with no clients [14:33] we don't have time right now to go look for other ways to produce it.. we have a deadline that we have to meet [14:33] cause of a client error from the api server, which then restarts the api server worker [14:34] mgz, cjohnston i also have no idea how to reproduce this.. it does seem more prevalent on canonistack.. [14:34] cjohnston: sure, I know you guys are pretty limited in what you can put in here [14:34] sinzui: bug #1289376 conversation should be here when you're around [14:34] <_mup_> Bug #1289376: Cannot build r2379 on amd64+precise

[14:34] sinzui: so certainly you should have been using newer golang [14:35] but the failures in that bug look exactly what I would expect if you tried to compile juju-core with go 1.0 [14:35] (the default in precise) [14:35] sinzui: I'd be willing to help debug it with you [14:35] jam, I will look further...oh the makefile didn't exec and add the deps... [14:36] jam, I switched the rules to run tests since I saw failures everywhere...but I didn't switch back the to the makefile tests. [14:36] sorry [14:36] sinzui: no problem, I'm happy it was easy to fix :) [14:36] we'd love to get 1.17.5 today (as a potential 1.18 candidate) though things are looking pretty slim at this ponit [14:36] point [14:36] bug #1289316 is serious [14:36] <_mup_> Bug #1289316: lxc not installed from ubuntu-cloud.archive on precise

[14:42] jam, I will release 1.17.5 when CI blesses a recent rev. It is 7 revs behind as we did rapid landings. [14:42] jam, I will release over the weekend if I need to [14:44] sinzui: sure, just that if we don't have fixes like bug #1289316 landed, we can't make it 1.18 as it is regressed from 1.16, and arguably we shoudln't make it 1.17.5 either [14:44] <_mup_> Bug #1289316: lxc not installed from ubuntu-cloud.archive on precise

[14:44] jam, noted [14:45] sinzui: I'll hopefully have a short list for you as I uncover things, and target them to 1.17.5 as Critical, will that help you ? [14:47] sinzui: or would you prefer a different way of indicating blockers [14:55] ? [15:04] jam, sorry, distracted by meeting. I would love a list of blockers. we can target the bugs to 1.17.5 to be clear they need to be fixed in it [15:05] jam, https://launchpad.net/juju-core/+milestone/1.17.5 has a few bugs that stakeholders care about. We can push the off to the next release though since I can release every week === Guest31737 is now known as bodie_ [15:49] jam: https://bugs.launchpad.net/juju-core/+bug/1288141 [15:52] sinzui: did you start duping the bugs against https://bugs.launchpad.net/juju-core/+bug/1227450 ? [15:52] we have probably ~3 bugs I think that are related enough to be a dupe [16:07] <_mup_> Bug #1288141: Errors when a 1.17.4 unit is rebooted

[16:07] <_mup_> Bug #1227450: juju does not retry provisioning against transient provider errors

[16:07] jam, hazmat did with my blessing. [16:07] jam, looks like CI is catching up I see r2390 has started testing [16:19] fwereade_, natefinch, jam: https://codereview.appspot.com/72310044 [16:26] mgz LGTM'd. Thanks. === niemeyer_ is now known as niemeyer [17:45] mgz: https://codereview.appspot.com/72270044 [17:45] also dimitern ^