[00:00] so for anyone that cares: http://blogs.msdn.com/b/looking_forward_microsoft__support_for_secure_shell_ssh1/archive/2015/06/02/managing-looking-forward-microsoft-support-for-secure-shell-ssh.aspx [00:00] still waiting for the announcement saying that windows is adopting the linux kernel [00:09] waigani: thanks for fixing maas bug, auto pilot guys will be happy [00:10] wallyworld: are you sure that not fixing bug 1457225 for 1.24 is a good idea? [00:10] Bug #1457225: Upgrading from 1.20.9 to 1.23.3 works, but error: runner.go:219 exited "machiner": machine-0 failed to set status started: cannot set status of machine "0": not found or not alive

[00:10] wallyworld: seems pretty serious [00:11] menn0: from what i'm told, not really reproducable so the plan is to include in a point release if needed [00:12] wallyworld: let me have a quick poke at it [00:12] ok, ty [00:12] wallyworld: i can guess what it is [00:12] wallyworld: welcome. I've put my notes on the wiki, so next time will be a bit quicker: https://github.com/juju/juju/wiki/Update-Launchpad-Dependency [00:12] waigani: ty [00:12] menn0: i haven't looked into the bug in detail, just going on conversations, so if you can guess, that would be good [00:13] wallyworld: ok [00:50] natefinch-afk: when you are around, can you look at this bug which arose with the move conf files from logs work? https://bugs.launchpad.net/juju-core/+bug/1461246 [00:50] Bug #1461246: UpgradeSuite.TestUpgradeStepsHostMachine consistently fails on utopic

[01:01] Bug #1461339 was opened: rsyslog on bootstrap node causing instability === natefinch-afk is now known as natefinch [01:18] wallyworld: I'll TAL [01:18] ty [01:19] wallyworld: I have been unable to reproduce bug 1457225 and i've requested more information [01:19] Bug #1457225: Upgrading from 1.20.9 to 1.23.3 works, but error: runner.go:219 exited "machiner": machine-0 failed to set status started: cannot set status [01:19] of machine "0": not found or not alive

[01:19] wallyworld: dropping it from 1.24 is probably ok [01:19] thanks menn0, that matches my understanding also [01:19] wallyworld: it's not what I thought it was at least. it looks like the upgrade started but never completed for some reason. [01:19] yeah [01:19] good to check though [01:20] axw_: allowing only count for storage constraints is only applicable to storage-add hook tool, rite? [01:20] anastasiamac: correct. "juju storage add" should be able to do whatever [01:20] axw_: tyvm :D [01:20] anastasiamac: "storage-add" should just take a name and optional count, and take the rest from constraints [01:21] natefinch: I believe that the failing upgrade test runs all the upgrade steps [01:21] natefinch: it probably doesn't need to [01:21] natefinch: it's an old one [01:21] menn0: ahh interesting [01:21] natefinch: mocking out the actual running of the upgrade steps and just confirming that they would be run is probably enopugh [01:22] natefinch: the steps and upgrade running infrastructure is all tested comprehensively elsewhere [01:22] * menn0 has to go for a bit [01:39] sinzui: tell me i'm crazy - i run "juju get-env" and the output includes secrets. why shouldn't we be masking those out by default? [01:49] wallyworld: :( [01:49] Bug #1461342 was opened: unit storage add should only accept count as constraint

[02:07] wallyworld: the two tests that are marked as failing pass just fine on my machine (and I'm on utopic). That specific test, when I run it, just says it's skipping all the copying of files, because they don't exist. But on the CI machine, it says there's one there that it doesn't have permission to remove. Seems like an environmental problem - there's a file in /var/log/juju/ that shouldn't exist. [02:07] wallyworld: given that it's only failing on utopic makes me suspicious that it's just that CI machine that has the errant file lying around, since the code isn't any different for different series. [02:08] natefinch: that makes sense i think [02:08] maybe mark the bug with a comment [02:08] natefinch: also, i recall a change with "-" in the branch? [02:09] with rsyslog config? [02:09] i just saw this, could be related? http://pastebin.ubuntu.com/11533025/ [02:09] seems there's an extra "-" in front of machine-1 [02:09] weird... might have been an errant typo [02:10] natefinch: also. maybe the upgrade step should just skip files it can't move? witha warning? [02:10] more robust that way [02:10] wallyworld, natefinch: i was going to say what wallyworld just said [02:11] you don't want an upgrade completely failing b/c of this [02:11] wallyworld: yeah, that makes sense. No need to kill the upgrade step.. the whole reason we're moving the files is so that if you blow away the folder, you don't mess up our config... but we don't really care if the old files are there anymore or not. [02:11] yep [02:13] I'll make a PR [02:15] natefinch: https://bugs.launchpad.net/juju-core/+bug/1461354 - i'm working another bug atm, but for your day tomorrow, if i don't get to this - could you look at it from the perspective it could be caused by the "-" change? [02:15] Bug #1461354: invalid tag and panic seen in state server logs

[02:16] wallyworld: absolutely... though I don't know what the "-" change is, but I can look for it. [02:16] oh, actually, I know what you're talking about [02:16] natefinch: in the log move work - you change rsyslog config [02:16] sorry, didn't know how to explain :-) [02:17] yeah yeah... I made the Namespace not include the - in the field, but added it to the logic of writing out instead [02:17] i may be jumping to conclusions [02:17] but seems like a reasonable place to start looking [02:17] it does seem suspicious [02:19] Bug #1461354 was opened: invalid tag and panic seen in state server logs

[02:27] OMG export_test is the worst [02:41] is the bot broken? [02:45] ahh maybe just slow [02:48] maybe it's just the reviewboard integration that's broken [02:49] wallyworld, menn0: care to do a very quick review on github? https://github.com/juju/juju/pull/2476 [02:49] ok [02:49] Bug #1461358 was opened: Resource tags should be periodically updated to reflect reality

[02:54] menn0: you and i obviously think simiarly [02:55] wallyworld: well in one case. [02:55] then I found all the trivial stuff and you found what could be actual problems :) [02:55] well, you saved me from the gc.IsNil things :-) [02:56] So, I almost put a comment in the actual code about not using PatchValue. I'm not embedding whatever base suite has that... adding it would run a ton more code for the tests for no reason, and would actually add more lines of code to the file.... all to avoid doing my own defer. [02:58] As for warning.... I don't know? I guess I assumed that if we don't care if the old file is there.... we don't care. No need to warn. Warn is something I feel is like "this may indicate there's a problem you should look at" .... but a file we can't delete that we don't read or care about is not really a problem for Juju. [02:58] I started to write it as Warning and then changed to Info. I'm only like 40% committed to it though, so if you guys think warning is more appropriate, that's fine. [02:59] natefinch: that BaseSuite embeds CleanupSuite which has PatchValue [02:59] menn0: weird, I swear I tried to type it in and it complained at me. Maybe I typoed it,. [03:03] menn0: so it does... my autocomplete was not finding it for some reason. [03:05] menn0, wallyworld: Fixes ready, do you want me to make the log message Warning? [03:06] yes please [03:06] ok [03:10] updated: https://github.com/juju/juju/pull/2476 [03:13] * thumper is getting pretty grumpy at all the intermittent failures with the bot [03:19] waigani: ocr? trivial one please for ci blocker http://reviews.vapour.ws/r/1843/ [03:21] wallyworld: shipit [03:21] tyvm [03:25] YAY (NOT) new intermittent test failure: filter_test.go:146: s.assertAgentTerminates(c, f) [03:25] uniter/filter [03:27] thumper: you running vivid? [03:28] our tests on vivid are worse [03:37] Bug #1461150 changed: lxc provisioning fails on joyent

[03:37] Bug #1461246 changed: UpgradeSuite.TestUpgradeStepsHostMachine consistently fails on utopic

[03:44] wallyworld, menn0: care to check my update? https://github.com/juju/juju/pull/2476/files [03:44] ok [03:45] natefinch: +1 [03:45] awesome [03:51] axw_: you resources pr - the provisioner version bump - it should be ok with an older env, and upgrades should work too, from what i can tell. i think? [03:53] wallyworld: the provisioner has a bit of code that loops while it gets a NotImplemented error [03:53] so I think it should be fine [03:53] ok [03:54] wallyworld: no [03:54] wallyworld: and that failure is from the bot [03:54] ah ok [03:57] axw_: lgtm, love the deleted code, much cleaner [03:57] wallyworld: cool, thanks [05:25] Bug #1461385 was opened: apiserver/instancepoller: data race in test [05:55] Bug #1461393 was opened: worker/deployer: multiple data races [07:38] wallyworld: https://github.com/axw/juju/blob/resource-tags-units-storagename-etc/environs/tags/tags.go [07:38] wallyworld: quick look over please, it was missing in the diff before [07:39] ok [07:39] overzealous .gitignore [07:39] axw: actually i did wonder about not seeing that file in the PR [07:40] axw: LGTM [07:40] ta [08:43] anyone here got a graph theory background? [08:49] fwereade: the answer depends on why u r asking :D [08:49] anastasiamac, I'm wondering if there's any literature I should look at regarding *potential* cycles [08:50] anastasiamac, i.e. I have a graph built up from a bunnch of successor lists which do not always agree, and may thus imply cycles [08:50] anastasiamac, I know that each node appears at most once in each successor list [08:50] fwereade: so you're looking to run Tarjan's algorithm on your data set? [08:51] Spads, not quite [08:51] Spads, I'm lookinng to determin whether a data set will produce the same tarjan sort for all possible extensions of the original data [08:51] ahhhh [08:51] Spads, given that the inputs can change in somewhat restricted ways [08:52] Spads, I thought I had something and, well, the failure rate on a workload designed to trigger the problem is down from 1 in 10k to 1 in 2m [08:52] Spads, which is sort of good [08:53] Spads, but sorta implies that I've missed something subtle, hence wondering if anyone remembers any related literature or anything :) [08:54] fwereade: this is outside my two-decade-old memory of my discrete methods/graph theory classes. Let me poke someone who may know [08:54] Spads, tyvm [08:55] Spads, no rush, I have a meeting in a moment anyway [08:55] fwereade: yeah, she may need to go dig a bit anyway [08:55] Spads, cheers [08:57] fwereade: are your rules aware of the tarjan weights of the nodes they're connecting to? [09:02] voidspace, it's just you and me today :) [09:03] Spads, I'm not aware of any weighting here [09:09] voidspace, ping [09:10] dimitern: sorry [09:10] dimitern: was helping wife with car!! [09:10] dimitern: omw [09:11] voidspace, np [10:47] anyone know what the Settings.NoProxy value means in github.com/juju/utils/proxy ? [11:38] evilnickveitch: here's as far as i've got so far: http://paste.ubuntu.com/11541062/ [11:38] evilnickveitch: i've got descriptions for everything now, at least [11:45] rogpeppe, that looks good - I will update the table in docs accordingly [11:45] evilnickveitch: i could probably generate the table automatically from that table if you'd like - might save a bit of time [11:45] s/from that table/from that data/ [11:47] rogpeppe, if you want to give that a go, i'm not stopping you :) the markdown syntax for tables is pretty straightforward [11:47] evilnickveitch: should only take a minute or two [11:47] evilnickveitch: i'm just formalising the defaults [11:48] cool [12:13] jam: got time for a brief chat about spec? [12:14] evilnickveitch: http://paste.ubuntu.com/11541625/ [12:14] evilnickveitch: gist.github.com doesn't like it though, so it may well be invalid md syntax [12:15] evilnickveitch: i left deprecated entries in there - you'll probably want to separate them out [12:16] evilnickveitch: this is better: https://gist.github.com/rogpeppe/c7ce1f02258a39898d37 [12:16] wallyworld: I'm meeting with alexis now, can I chat with you after 30min or so? [12:16] sure [12:18] voidspace: i'm led to believe you might know something about github.com/juju/utils/proxy ... is that true? [12:18] rogpeppe, yup, that will certainly do the job. thanks! [12:18] evilnickveitch: np [12:19] evilnickveitch: let me fix those multi-line descriptions first [12:20] ok [12:25] evilnickveitch: this is better: https://gist.github.com/rogpeppe/8015a6c7122c85878b43 [12:27] rogpeppe, it is. I will get this added shortly [12:27] evilnickveitch: it really needs default values for all the other rows too [12:27] yes. I am not sure they all have them [12:27] rogpeppe, AFAIR, some keys simply don't exist if they aren't specified [12:28] evilnickveitch: yeah, but at least all bool vals should document what the default bool is [12:28] perhaps they should definitively say 'none' in that case [12:29] agreed [12:29] evilnickveitch: FWIW this is the code i used to generate the table (and check consistency with the real code): http://play.golang.org/p/6ci9O6GOz3 [12:29] rogpeppe, ok, thanks, i am sure that will be useful in the future too [12:30] evilnickveitch: the stuff after the ///// comment was pasted directly from the juju-core config code [12:41] Bug #1461529 was opened: juju upgrade-charm has no effect for subordinate charms [12:45] ericsnow: ping [12:48] Hello Folks, [12:49] I am trying to deploy HA openstack keystone in 3 lxc nodes. [12:49] I am getting this error: ERROR: grp_ks_vips: id is already in use [12:56] rogpeppe: I don't believe so, no [12:56] voidspace: ok, thanks [12:58] :-) [12:58] sorry [13:09] Bug #1461534 was opened: presence.$cmd {getLastError} taking 100ms

[13:17] mgz: I was about to try a reboot from the horzon console [13:17] evilnickveitch: this also adds information on which attributes are immutable: https://gist.github.com/rogpeppe/cff7976fbf56f135af4b [13:18] * sinzui can move two jobs to another slave [13:19] rogpeppe, ah, useful! We may need to coordinate with web design again before making the table any bigger - there is precious little width to display the current info! [13:19] evilnickveitch: that's your domain :) [13:20] I wish it were! But it is my domain to ask the people who can fix it, at least [13:35] wwitzel3: pong [13:35] Syed_A: not a lot of open stack experts on here... try #juju [13:36] ericsnow: hey, want to pair up? [13:36] wwitzel3: yep [13:41] ericsnow: going to moonstone now [13:45] jam: ping, you still around? [13:47] wwitzel3: I'm 45 minutes past needing to help my son with homework, so I have to pretend I'm not here, sorry [13:47] jam: understand completely :) [13:48] we could do cooperative help with the homework and get you back faster? :p [13:54] natefinch: can you triage bug 1461339? I see several related bugs tagged “rsyslog”. This is also an older juju; the issue might be fixed in newer juju [13:54] Bug #1461339: rsyslog on bootstrap node causing instability

[13:57] sinzui: that sounds a lot like what happened a couple weeks ago with that big customer problem. I definitely think it's something we should be looking into addressing. This is really katco's dept though :) I'm just a code monkey now ;) [13:58] natefinch: If you feel it is still a real issue, then we now the issue medium, and if we want to fix it in 1.25, it is high [13:58] thank you natefinch [13:58] sinzui: tal [13:58] natefinch: you going to look at bug 1461354? [13:59] Bug #1461354: debug-log EOF: invalid tag and panic seen in state server logs

[13:59] katco: you saw the credentials one in our fix - it's a one liner [13:59] wallyworld: eh? context pls! :) [14:00] the cloudsigma crednetials leak bug we talked about this m orning [14:00] it's a private bug on beta6 - a few hours ago they added a note to the bug [14:00] wallyworld: ah no i haven't looked at that [14:00] we need to change one line of code to fix [14:01] wallyworld: ok, it sounds like it's proposed? [14:01] proposed? [14:01] oh on rb? [14:01] yeah [14:02] not that i can see [14:02] but i've been busy on other bugs and things sadly and it was past my official eod when i saw the comment and then i had soccer [14:02] wallyworld: ah ok. we'll try and get it fixed [14:02] ty :-) [14:02] past midnight here [14:03] wallyworld: yipes, gtb! [14:03] will do, sorry for not fixing, thought it best to nag you at this stage of the evening :-) [14:03] wallyworld: definitely :) [14:04] wallyworld: share the load [14:04] and the only other one is the critical debug log issue i discovered today [14:04] debug log entries from machines != 0 are busted [14:04] the logged tag seems to have an extra "-" prepended [14:05] might? be syslog related, not sure [14:05] wallyworld: I think I know the exact problem [14:05] awesome :-) so a fix is on the way? [14:05] wallyworld: yep [14:05] tyvm :-) [14:06] so we can get beta6 out today, yay [14:09] cherylj: ping, need a review when you have a moment, http://reviews.vapour.ws/r/1798/ [14:18] wwitzel3: looking now, but I am not a graduated reviewer yet. [14:24] Bug #1461561 was opened: juju run fails with "Permission denied (publickey)" on manual provider

[14:32] cherylj: thats ok, it will help get the ball rolling, thank you [14:51] jw4: you around? [14:51] natefinch: indeed [14:53] jw4: what's up with juju action not taking the -e flag until after the subcommand? It's different from every other command we have, where you can do "juju deploy -e [charm]" for example, but for action it's juju action do -e [action [14:53] natefinch: that's weird [14:54] natefinch: I suspect it's because all of our commands are subcommands, and we don't have any top level command code [14:54] natefinch: "bug" [14:59] omg, how is it possible that the chrome hangouts plugin still does not accept multi account [15:42] abentley: addressed your concerns. Should be a pretty quick review. https://code.launchpad.net/~natefinch/juju-ci-tools/logrot/+merge/259750/+index [15:42] natefinch: Thanks. On the phone. [15:42] abentley: np [15:42] anyone know which commit fixed the second issue mentioned in this comment? https://bugs.launchpad.net/juju-core/+bug/1339770/comments/2 [15:42] Bug #1339770: Machines are killed if mongo fails

[15:56] man what I wouldn't give to have issues on github so when I fix a bug I don't have to go through and copy and paste all the bug data into the comments. [16:01] ericsnow: reviewboard doesn't seem to be picking up commits to 1.24... can you investigate? [16:01] sure [16:01] wallyworld: fix for your issue: https://github.com/juju/juju/pull/2483 [16:01] natefinch: is your 1.24 up-to-date? [16:02] ericsnow: ish? [16:03] ericsnow: I think I updated yesterday [16:07] cherylj: can you review this? https://github.com/juju/juju/pull/2483 fixes a bug I introduced here: https://github.com/juju/juju/pull/2458/files#diff-29cf702f1f3ee55354dc999b19a2e391R208 [16:07] (very quick review) [16:08] natefinch: yeah, I'm finishing up a review for wwitzel3, give me a couple minutes. [16:14] cherylj: np, thanks [16:15] natefinch: re: reviewboard, I'm not seeing a problem [16:15] natefinch: was it a particular PR? [16:16] ericsnow: oh, no, maybe something is just really slow... the review is posted now [16:17] natefinch: k [16:17] natefinch: yeah, looks like RB is a little slow right now [16:17] (where really slow might be anything greater than my patience level of ~20 seconds ;) [16:23] natefinch: r=me. Shall I merge it for you? [16:25] abentley: yes please [16:25] abentley: thanks for all the time you spent reviewing the code. [16:26] natefinch: You're welcome. Thanks for your follow-through. [16:45] Bug #1461605 was opened: juju action commands require -e in the "wrong" place

[16:57] gsamfira: bug 1455627 [16:57] Bug #1455627: TestAgentConnectionDelaysShutdownWithPing fails

[16:58] mgz_: ahh. Cool. This was the only test that sometimes failed against Go 1.4.2 [16:59] mgz_: tested on windows and Ubuntu. The rest all passed [17:00] mgz_: also did a live test. Bootstrapped new env and deployed a noop charm [17:01] on windows and ubuntu [17:01] all worked fine [17:01] cool [17:02] mgz_: it should probably be tested for a couple of weeks in the CI before trusting that all is well, but initial tests looked okay :) === kadams54_ is now known as kadams54-away === kadams54-away is now known as kadams54_ [17:38] if anyone fancies a review, here's a change to environs/config that starts us down the road that will let us create sensible UI forms for environment config attributes: https://github.com/juju/juju/pull/2487 [18:38] gah... dammit ubuntu... first my screen freezes, now I reboot and my mouse cursor is invisible [18:41] at least reviewboard didn't drop all my comments on the floor... that's ncie [18:43] natefinch: well if it makes you feel better, each time I use my kb after taking a bath 2 finger scroll goes crazy [18:43] That sounds like a euphemism for something [18:46] gsamfira: I'd like to see tests with multi-byte characters... I wonder if those "interleaved nulls" are really just UTF16 padding for ASCII characters. [18:48] natefinch: that does not sound right at all [19:03] btw, evidently this was just released: https://github.com/aws/aws-sdk-go [19:04] "aws-sdk-go is the official AWS SDK for the Go programming language" [19:04] currently at "developer preview release" [19:05] https://aws.amazon.com/blogs/aws/developer-preview-of-aws-sdk-for-go-is-now-available/ [19:06] natefinch: that sounds a lot like alpha :p [19:06] "we have moved our GitHub repository from the awslabs organization to aws/aws-sdk-go. This signifies that this SDK is no longer in an experimental state." [19:06] *shrug* [19:46] dammit git [19:48] natefinch: dont complain or we will make youuse bzr [19:50] somehow it thought my branch was missing a whole buttload of commits... so it wouldn't let me repush [19:55] did you rebase? [19:56] perrito666: yes, and this is what I got: https://github.com/juju/juju/pull/2483 [19:56] only the first and last commit are supposed to be there [19:57] ah because rebasing usually results on you not being able to push your branch if you had before [19:57] that's ok, I just made a new branch and cherry-picked the right commits: https://github.com/juju/juju/pull/2489 [19:57] taht makes no sense [19:57] all the commits there (which are mine) are merged into 1.24 [19:58] cherylj: if you have a second, this is the same as the last one you reviewed, but with an added test. http://reviews.vapour.ws/r/1855/diff/# (the key to the test is the "" in NewRsyslogConfigWorker.... we have another test that does the same thing, but with a non-empty namespace, but no test with an empty one. [19:58] natefinch: there, it is fine now [19:59] seems like github was having a bad day [19:59] natefinch: looking now. [20:06] perrito666: not sure what "there" means... still looks wrong to me [20:07]