[01:26] here's a quick review, https://github.com/juju/juju/pull/5410 [01:40] davecheney: reviewed [01:43] ta [01:57] Bug #1582463 opened: juju subordinate not allocating and can't be destroyed

[02:38] it takes 16 minutes to run the state tests on my machine [02:45] davecheney: I have a branch where I'm adding if testing.Short(){ c.Skip() } to all the longest tests until each package takes no more than 1 seconds to test. That should drop the time to run a smoke test dramatically. Not that it'll help when you really want to run the full suite. [02:45] davecheney: it also does that to all mgosuites and jujuconn suites in the suitesetup, skipping the entire suite, because, seriously, no. [02:46] brb [02:46] natefinch: i only care about the tests taht CI runs [02:53] davecheney: are you testing against mongo 3.2 now? [02:54] hm, doubt it [02:54] mwhudson: nope [02:56] mwhudson: what i mean is [02:57] my setup has not changed since 14.04 [02:57] so whatever was happening before i went on leave when the state tests did pass [02:57] is no longer happening [02:58] davecheney: oh right, of course [02:58] was just a stray thought anyway [02:58] mwhudson: fwiw, CI doesn't run mongo 3.2 [02:58] so, there's that [02:58] davecheney: yeah, i expect some tests still fail [02:59] could someone who's a mod on the juju-dev mailing list please approve my message that is slightly larger than the 40kb message limit [03:17] davecheney: but what about all those poor souls on their 2800 baud modems? [03:18] natefinch: some ppl believe that terrible experiences enrich souls :-P [03:20] anastasiamac_: I think some people just can't get out of the 1999 mindset [03:20] natefinch: 1999? when was that? :-P [03:22] wepl, ya'll miss out on my satorial wit [04:21] Bug #1556442 changed: juju quickstart looks for network-bridge: "lxcbr0" in local environment

[04:21] Bug #1557874 changed: juju behaviour in multi-hypervisor ec2 clouds [06:03] here is a simple one https://github.com/juju/juju/pull/5413 [06:25] Bug #1581748 opened: [2.0b7] add-credential no longer supports maas as a cloud type

[06:25] o/ [06:25] davecheney: you should have an amusing email to read now [06:26] thumper: :) [06:26] are you a mod on juju-dev ? [06:26] can you approve my less than amused email ? [06:26] yep [06:26] um... not the ubuntu juju-dev list, no [06:26] I think that is still just gustavo [06:31] thumper: no, it's Becky^WWilliam [06:31] elmo: oh, thanks [06:35] the mailing list moderation is a cluster - I'm going to arbitrarily drop all the held messages; some of them date back to 2013 [06:35] the majority are spam; and I can't believe the non-spam ones are still relevant [06:35] heh [06:45] thumper: heh, it seems I dodged it but frobware got on that plane after all? [06:45] dimitern: he will be [06:46] thumper: ah, not yet I see [06:47] sometimes not having a visa might be good for a change [06:55] Bug #1582555 opened: deploy cs:trusty/ubuntu deploys Xenial charm [07:42] rogpeppe: ping [07:42] dimitern: pong [07:44] rogpeppe: hey, I'm seeing a consistent test failure in cmd/jujud/agent/machine_test.go:TestDyingModelCleanup - http://paste.ubuntu.com/16472327/ [07:45] dimitern: and it might be because of my recent change? [07:45] rogpeppe: I did a git bisect and surprisingly it turned out your PR #5404, commit https://github.com/juju/juju/commit/2a731745eb2bb75f5ed20a887f23d9f198c16259 is the first time it happens [07:45] dimitern: interesting [07:46] rogpeppe: surprisingly, because it doesn't seem your PR touched anything near cmd/jujud/agent [07:46] dimitern: i don't think i believe that [07:46] dimitern: i don't think gce is even linked in to those tests [07:47] rogpeppe: well the only thing that comes to mind is ProviderSchema added in environs/interface.go, but not implemented by the dummy provider? [07:47] dimitern: it's just an interface declaration [07:47] dimitern: nothing uses it [07:47] rogpeppe: yeah, that's why it's a mystery.. [07:48] rogpeppe: can you try running cmd/jujud/agent$ go test -check.v -check.f MachineSuite.TestDyingModelCleanedUp on master tip and see if it passes? [07:49] might be related to different go versions.. I'm on go1.6.1 linux/amd64 [07:52] dimitern: ok, trying that [07:52] rogpeppe: hmmm.. running the test on the only commit of that PR (2a73174) passes, it fails on the merge commit itself :/ (1439ead) [07:53] elmo: i'm right here man [07:53] please wait til i sign off before calling me mesage irellevant :( [07:54] davecheney: yo! [07:54] rogpeppe: hey [07:54] davecheney: long time no see [07:54] davecheney: "see" [07:54] rogpeppe: your mate jeff hodges is looking for you [07:54] davecheney: hey o/ [07:54] davecheney: jeff hodges... name rings a bell [07:55] davecheney: ah yes [07:55] i think he wants you to make godep do something [07:55] or godeps [07:55] davecheney: i need to spend some time on godef [07:55] that's the one [07:55] davecheney: the guilt is nagging at me [07:55] davecheney: but i've had no time [07:56] davecheney: well, i guess i could've not gone for bike rides at the weekend but there are limits :) [07:56] i think you've got your priorties set just right [07:57] davecheney: along with all the other stuff i've wanted to do (fix encoding/xml, finish up my errgo proposal doc, rewrite juju etc, etc) [07:58] dimitern: i suspect it might be some kind of race issue [07:58] dimitern: FWIW i do see that test failing [07:58] rogpeppe: ok, thanks for confirming! [07:59] rogpeppe: it might be quite intermittent and that merge commit just happen to be a false positive, will dig further back [08:00] dimitern: FWIW (showdeps -a github.com/juju/juju/cmd/jujud/agent | grep gce) prints nothing [08:00] dimitern: i.e. none of the packages i changed in that PR are linked into those tests. [08:22] holy fuck [08:22] mongo failed three of my PR's that I submitted [08:22] what is the story [08:22] why hasn't that bag of shit been taken out behind the woodshed and been shot ? [08:23] dimitern: https://bugs.launchpad.net/juju-core/+bug/1581074 [08:23] Bug #1581074: LXD container interface names don't align with parent interface names

[08:24] dimitern: https://bugs.launchpad.net/juju-core/+bug/1566801 [08:24] Bug #1566801: LXD containers /etc/network/interfaces as generated by Juju gets overwritten by LXD container start

[08:25] dimitern: https://github.com/frobware/juju/tree/master-lp1581074 [08:47] fwereade: https://github.com/juju/juju/pull/5415 [08:49] davecheney, LGTM, thanks [08:52] [09:01] voidspace: standup? [09:01] dimitern: morning [09:38] davecheney, morning dave - still up? regarding http://reviews.vapour.ws/r/4850/ ParseUnitTag is different [09:44] Bug #1582620 opened: juju-upgrade-mondo typo [10:42] dimitern: Ugh - any way in mongo to find out that a collection is a capped one? [10:43] babbageclunk: I'm sure there is a way, but I don't know off hand [10:43] babbageclunk: IIRC the txn collection is capped [10:54] dimitern: certainly the log collections are. I can't see any way to get CollectionInfo in the mgo docs. [10:54] dimitern: seems like I'll just have to check the error text - yay! And then guess the right size when I recreate. [10:55] babbageclunk: it seems Collection.Create only returns CollectionInfo [10:56] dimitern: I don't want to have to try to intercept the create calls so I can stash the infos somewhere. [11:20] my new favorite thing: juju debug-log --replay -T > bacon.log [11:21] you guys/gals are killin it. <3 [11:21] and if this was always there, i just found it and it feels like my birthday all over again [11:22] lol lazyPower [11:23] o/ dimitern [11:27] lazyPower: o/ it's always good to hear a happy user with all the issues we dealt with for 2.0 :) [11:28] its been a schlog, happy to spread the cheer when I can :D [11:30] I have some questions for you specific to spaces and how we can leverage them in k8s. I'd like to replace flannel with network-spaces in our kubernetes-core bundle. But we're a couple weeks out in terms of planning. I'll poke you in #juju when we're ready - prepare yourself ;) [11:31] lazyPower: k8s ? [11:31] aah :) [11:32] lazyPower: well spaces are well supported mostly on maas, with some support on aws [11:32] ah right, its similar to storage in the fact its hamstrung by provider support [11:32] good point [11:37] welp thats something to think on [11:38] lazyPower: yeah, we plan to add support for spaces to public clouds first, but it will take some time [11:39] Bug #1582667 opened: i/o timeout when deploying/upgrading charm [12:12] dimitern completely understood. no pressure :) [12:12] i mean, networking is easy right? [12:12] said no-one ever [12:15] lazyPower: :D indeed [13:42] Bug #1582714 opened: worker/resumer: test failure during CI [13:51] Bug #1582731 opened: github.com/juju/juju/state tests time out on centos

[14:00] Bug #1582731 changed: github.com/juju/juju/state tests time out on centos

[14:12] Bug #1582731 opened: github.com/juju/juju/state tests time out on centos and windows

[14:27] natefinch: hey just wanted to say thank you for getting that bug addressed [14:27] natefinch: i'm sure ian is breathing a sigh of relief :) [14:28] katco: welcome, yeah, I was glad to be able to fix it in a timely manner [14:49] katco: trivial review of the deps change for that? http://reviews.vapour.ws/r/4852/ [14:49] natefinch: tal [14:49] natefinch: ship it [15:01] * perrito666 has sandpaper in his throat [15:02] perrito666: you should try eating food instead [15:02] natefinch: nah, I am too hardcore for that [15:02] I think Ill be making sign language for the next standup [15:22] Bug #1582714 changed: worker/resumer: test failure during CI [15:54] grep -r '"admin"' --include="*.go" . | wc -l [15:54] 231 [15:54] screw dry, I wish we could at least manage something less than soaking wet [15:54] damp? moist? [15:55] natefinch: now is your chance :) stick it somewhere sane [15:55] katco: yep [15:55] histerical raisons [15:55] there is a testing.AdminUser [15:56] oh, that's the best part. Half of these are usernames, not model names [15:56] possibly more than half [15:56] natefinch: same happened with file urls [15:56] I don't know because THEY'RE JUST STRINGS [15:56] there was a bunch constructed by hand [15:57] hows it going everyone? I feel like I'm missing something here ... is there a way around this -> https://github.com/juju/juju/issues/5411 ? [15:57] (╯°□°）╯︵ ┻━┻ [15:57] * perrito666 touches natefinch with the "well fix it" wand [15:58] thumper: package testing? [15:58] bdx: hello, sorry I dont know [15:58] bdx: we track our bugs on launchpad. here's a dupe with activity: https://bugs.launchpad.net/juju-core/+bug/1571593 [15:58] dimitern: voidspace any of you is familiar with that error? [15:58] Bug #1571593: lxd bootstrap fails with unhelpful 'invalid config: no addresses match'

[15:58] github.com/juju/juju/testing [15:58] katco: mm, we should go around closing the ones in gh every now and then [15:59] perrito666: we try. we need to turn off GH issues completely, but there always seems to be something more important to work on... :( [15:59] meh, I dont have supercow powers, I cant close the bug [15:59] I added the link to the lp one in the comments [16:00] * katco closed the bug [16:00] Oh my bad, I should be filing those in lp huh [16:00] oops [16:00] thanks [16:00] bdx: no worries, it's a common mistake. we don't make it clear at all [16:00] perrito666: that's lxd related [16:00] bdx: it is, I just saw the address part [16:01] sorry to disturb you :) [16:01] perrito666: it usually means lxdbr0 was misconfigured [16:01] perrito666: I'm doing everything I can to not use lxdbr0 [16:02] but I guess I should of scouted out the bugs further [16:02] shoot [16:02] ha [16:02] sorry [16:04] dimitern, perrito666: is it a bug with lxd then? [16:05] bdx: not really, there is a previous step to be done but apparently we are not documenting it? [16:05] bdx: run lxd init [16:05] and your problem should go away after networking setup [16:06] perrito666: yea, I've ran `lxd init` enough to no avail [16:06] mmm, very strange [16:08] try dpkg-reconfigure lxd and then restart lxd-bridge service [16:08] perrito666: I can bootstrap lxd just fine when using lxdbr0 [16:08] ah ok, then what do you do [16:08] to make it fail? [16:09] perrito666: any host only bridge works for me ... its when I try to configure lxd to use a bridge on an external interface that it fails [16:09] mm, my knowledge of lxd does not go that far sorry [16:10] ericsnow: ^ do you know more about that? [16:10] should the issue be raised with lxd then? [16:10] perrito666: not really; perhaps tych0? [16:10] ericsnow: he is not here :) [16:11] thumper: sorry, my question was - why is a production value being defined in a testing package? [16:11] bdx: ill try to summon someone from lxd, otherwise, file the bug with juju since we are the ones that break in that case [16:13] Bug #1582798 opened: After bootstrapping, juju gui isn't automatically deployed

[16:13] Bug #1582799 opened: Deploying from the cli api.jujucharms.com is not reachable

[16:16] dimitern: I've got the state tests working for mongo3.2 now, except for TestSetAdminPassword. [16:16] thumper: ^^ [16:16] perrito666: the bug already exists .... its just not very specific to the nature of what actually happens -> https://bugs.launchpad.net/juju-core/+bug/1571593 [16:16] Bug #1571593: lxd bootstrap fails with unhelpful 'invalid config: no addresses match'

[16:16] I'll add in my bits to it [16:16] babbageclunk: awesome! [16:16] babbageclunk: is it as fast as 2.4 ? [16:17] bdx: the error seems to be the same but they state a different cause [16:17] could yolu please describe your case in that bug as a follow up? [16:17] dimitern, thumper: but this change between 2 and 3 means that I think Initialize will have to work differently: https://docs.mongodb.com/manual/core/security-users/#localhost-exception [16:18] dimitern: need to rerun the whole package again now that I've whacked all the moles - hang on. [16:18] dimitern: but within a minute of 2.4 definitely. [16:19] dimitern: which is better than 100x [16:19] perrito666: surely [16:19] bdx: tx a lot [16:20] np [16:20] thank you [16:24] babbageclunk: great news then! \o/ [16:24] dimitern: But so my final problem: at the moment Initialize connects to localhost without auth and creates the collections and indexes from allCollections(), and then at some point in the future sets an admin password - is that right? [16:25] babbageclunk: AIUI 'some point in the future' is at test(suite?) teardown/reset [16:26] dimitern: No, I mean in a real bootstrapped controller. [16:26] dimitern: almost all the tests run with auth turned off. [16:26] babbageclunk: ah, so the process goes - bootstrap with generated pwd, then when the controller's jujud agent starts, it resets the admin pwd once [16:28] ok - but in state.Initialize it does all of the collection and index creation before that, right? I think the admin password just needs to be done earlier. [16:30] dimitern: oh, no - I've found it in agentbootstrap.InitializeState - it's setting the password first, isn't it. Ok, I just need to fix the test then. Sweet! [16:43] Bug #1582801 opened: Juju debug-log returns error when using during bundle deployment

=== Garyx_ is now known as Garyx [17:25] Bug #1582818 opened: SSH Key in MAAS GUI not working

[17:44] davecheney: not sure what you mean; I approved your mail before flushing the queue [17:58] perrito666, dimitern: this would explain alot -> https://github.com/juju/juju/blob/master/tools/lxdclient/client.go#L232-233 [18:00] really? we are checking for eth0? that is quite easy to break [18:02] bdx: this check might be wrong https://github.com/juju/juju/blob/master/tools/lxdclient/client.go#L241-248 [18:07] quick review anyone? http://reviews.vapour.ws/r/4854/ [18:08] yea, yea, I saw the whole part after the comments .... cry cry [18:12] fwereade: yt? [18:13] perrito666, katco, ericsnow: super easy, I promise: http://reviews.vapour.ws/r/4854/ [18:14] natefinch: rviewing [18:15] natefinch: what prompts this change? [18:15] natefinch: there, that wasn't so bad [18:16] perrito666: Mark [18:16] say no more [18:17] ericsnow: yeah, turns out there actually was a constant, it just was hard to find amongst all the other "admin" references [18:17] natefinch: reviewed [18:23] perrito666: good catch on that comment https://pbs.twimg.com/media/B3qIJLFCcAEJLWm.jpg [18:23] natefinch: I believe that I lack the etnicity to catch that joke :p [18:23] bdx: ericsnow: i think you have to use lxdbr0 right now [18:24] perrito666: the label says rice, the container contains cookies [18:24] pass it through the I am southamerican filter [18:24] there's no way to configure it otherwise [18:24] perrito666: it's about comments being out of date [18:24] natefinch: I assumed they where rice cookies (things that happen when you live with celiac people) [18:26] by default I suspect of any sort of bakery product not made/bought by me to be rice based :p [18:30] perrito666: fairly uncommon here [18:31] here, too, unless here==my house [18:35] Bug #1582841 opened: remove check for lxdbr0 [18:44] tych0, perrito666, dimitern: I named the bridge on my external interface lxdbr0 .. this tricked juju, and lxd, and got me bootstrapped :-) [18:44] lol [18:44] so simple, so sweet [18:53] just found our 15 month old in an awkward ball, face down in a chair, unresponsive.... turned out he'd just fallen asleep and kind of slumped into a pile. [18:53] and unresponsive only because he was so tired [19:01] natefinch: and how is your panic attack? [19:07] perrito666: fine once it was established he was sleeping, but geez === redir is now known as redir_lunch === redir_lunch is now known as redir === admcleod_ is now known as admcleod [20:22] when do I want to return errors.Trace(error) vs. error ? [20:22] is there a rule of thumb? [20:22] katco: natefinch ericsnow ^^ [20:22] redir: .Trace just adds stack information to the underlying object [20:22] redir: basically always trace [20:22] redir: really not a reason to not use it [20:23] redir: .Annotate et. al. will do it for you, so you don't need to wrap those in .Trace [20:23] OK just modifying a func that does both and wasn't sure why. [20:23] Bug #1582881 opened: destroying a subordinate returns an error but seems to work [20:23] redir: probably started being written before we had the errors library [20:23] redir: probably targeted fixes. .Trace is only ~2y old [20:24] should I .Trace it in the drive by? [20:24] redir: if it's not too many of them, that's fine. [20:24] redir: yeah i'd say those are easy to review, so bring the function up to code (no pun intended) [20:26] * redir nods. just one of three...easy enough. [20:27] gah, why is it get-model-config and not show-model-config? :/ everything else is show :/ [20:28] why doesn't `juju block` accept a -m arg? [20:31] is it an oversight? am i right to assume that the block command does indeed apply to one specific model? [20:31] tvansteenburgh: not sure [20:33] tvansteenburgh: it definitely is per model according to the comments [20:33] natefinch: okay, thanks [20:34] tvansteenburgh: must be an oversight... I think because it's built as a super command (i.e. juju block destroy-model .. destroy model is a command, not an argument) [20:37] tvansteenburgh: can you file a bug? [20:37] natefinch: sure thing [21:21] anastasiamac_: ping [21:41] ericsnow: o/ [21:43] ericsnow: m making coffee - 7.40am.. bbiab [21:43] anastasiamac_: np [21:43] anastasiamac_: just thinking we may want to postpone our meeting until the syslog stuff is more settled [21:54] anastasiamac_: was going to suggest the same [21:59] katco: ericsnow: sounds good \o/ [21:59] * redir doesn't know about meeting [21:59] i'll move the meetings [22:00] Bug #1582892 changed: `juju block` doesn't accept -m arg [22:00] ericsnow: katco: altho m not convinced that the user stories will actually change :D [22:00] redir: i'll fill u in during standup [22:00] anastasiamac_: yeah, probably not :) [22:00] redir: perrito666: m happy to have a standup earlier if it suits u [22:01] ericsnow: m still keen to talk to u if u want :D but will move the actual meeting booking [22:01] anastasiamac_: works for me [22:02] redir: want to do standup now-ish? [22:02] anastasiamac_: fine by me, running tests now so I have time [22:02] :) [22:03] I am in the room [22:03] Bug #1582892 opened: `juju block` doesn't accept -m arg [22:03] redir: perrito666: m going to tanzanite now then \o/ [22:06]