[01:00] thumper, wallyworld: message clarity around creds mngmt, part 1 -https://github.com/juju/juju/pull/8363 :D PTAL \o/ [01:14] anastasiamac: lgtm, yeah credential mgmt is confusing for people [01:19] wallyworld: ta \o/ === thumper is now known as thumper-afk [03:50] wallyworld: as discussed, PR that filters out cloud cred tag from modelinfo unles u r model admin PTAL? https://github.com/juju/juju/pull/8364 [04:16] anastasiamac: sorry, was interviewing, lgtm [06:07] jam: u and guild recently worked on upgrading mongo to 3.4(I guess .7 coz this is what m seeing)... m running featuretests on develop and m getting https://pastebin.ubuntu.com/p/FJHJx6RjVD/ ... after that all subsequent tests in the suite understandably panic... any idea wot's up? [06:09] anastasiamac: it looks like our work on testing 3.4 was incomplete. I think it was a case of "we fixed the ones we knew about" but nobody kept running the suite against them (eric was running the suite, I was fixing). Anyway, I'll debug for a bit [06:11] jam: sounds great! thnx! m a bit surprised that this featuretests package is not run as part of landing/CI since it's our basic sanity check that layers are glued correctly - from cmd to api to state [06:21] anastasiamac: are you sure it isn't run? where did you get 3.4.7 from? Is it just that you're running Bionic? [06:21] I'm pretty sure the bot isn't testing 3.4 [06:22] jam: m running develop tests locally on my machine after doing a pull and dependencies re-run couple of hrs ago [06:23] anastasiamac: sure. my point is that the bot could very well be running featuretests, just they don't fail with mongo 3.2 which is the one available on !bionic [06:23] jam: u have no problems running featuretests on ur machine? [06:23] ah [06:24] testing now [06:24] mayb it should then? since u r landing this code?... would b nice to have a stable development env, at least on my machine... ;) [06:24] (seeing if I can reproduce with a 3.4.X of mongo) [06:24] thnx [06:25] anastasiamac: dblogtest passes even with a mongod that should be 3.4.12, will keep trying to see if I can reproduce [06:25] jam: it's debugLogDbSuite that fails [06:26] anastasiamac: sure, but the entire dblog_test.go file just passed, still investigating [06:28] jam: right, m running all tests in that directory, not isolated to a file... so maybe try 'go test -gocheck.v' within the directory? [06:28] anastasiamac: I'm doing exactly that [06:28] anastasiamac: does it fail every time for you? [06:28] or just once in a while [06:28] every time [06:31] jam: juts out of curiosity.. which go r u runing? [06:31] go test --check.v passed completely for featuretest for me. [06:31] go 1.9.4 from snap [06:31] on Xenial [06:32] and I'm grabbing the builds from Mongodb themselves, vs the bionic package, so its possible its a bionic thing. Though the error doesn't look specific to how it would be built [06:32] ah, its cause we're preferring /usr/lib/juju/bin [06:33] anastasiamac: you *can* just do "apt install juju-mongodb3.2" if you want to work around this [06:35] anastasiamac: so setting JUJU_MONGOD= to the correct 3.4 path did see the same failure you're seeing. [06:36] jam: I would prefer not to *work* around this :) i'd rather this was fixed :) [06:37] jam: m going to comment out this test (and others if any) locally to keep going... but we should address it, I *think* [06:38] jam: thnx for confirming m not just seeing things \o/ [06:38] anastasiamac: so, we absolutely need to support mongo 3.4. I'm just saying that if this is the failure, it isn't just this test. [06:38] anastasiamac: if you want a working environment, I'd recommend installing 3.2 which is what the bot is using [06:38] until we finish supporting 3.4 [06:38] jam: we r in violent agreement then [06:38] thnx [06:40] anastasiamac: given your recent changes to skip dying models, I'm guessing bug #1742699 should be fixed? it was causing CI fails from time to time [06:40] Bug #1742699: juju show-status fails during teardown [06:42] anastasiamac: so it seems that test does something nasty and resets how we're running the mongo db server, which then stays in that state for the rest of the tests... [06:43] hm. theoretically not, as in TearDown it resets it, but I think the failure is probably that TearDown is broken because of the compat failure, and then it never runs the cleanup step [06:43] jam: yes for the bug and ur theory on teardown seems to fit the symptoms... [07:16] anastasiamac: https://github.com/juju/testing/pull/135/files === thumper-afk is now known as thumper [07:16] its a change to the testing package, but seems to fix things here. can you test? [07:18] anastasiamac: if it works for you, I'll propose an update to 2.3 based on it [07:29] jam: the code looks good but i'll have to take it for the test run in part 2 of my day [07:31] anastasiamac: you can just do "go test --check.v --check.f debugLogDb" to just test the one that was failing [08:22] ugh... [08:22] brain has stopped working [08:22] yay slides... [08:22] that said, I'm increasingly happy with the content [08:22] laters peeps === salmankhan1 is now known as salmankhan [13:55] anyone care for a very small code review: https://github.com/juju/juju/pull/8368 [13:55] anastasiamac: ^^ that brings in the patch to Juju itself [13:56] manadart: externalreality ^^ [14:16] jam: I can look. [16:49] balloons: I have a PR that has been in "$$merge$$" for 2 hrs with no comment from the bot [16:49] https://github.com/juju/juju/pull/8368 [17:03] jam: second time is a charm apparently [18:05] wpk: its also possible the bot has decided it doesn't like me for some reason [18:05] wpk: thanks [19:09] Bug #1749272 opened: Adding existing subnets to space fails [20:08] morning team [20:15] evening boss [21:27] wpk: oh hai [21:27] wpk: still hanging around? [21:41] thumper: for a moment but yes [21:41] thumper: final mailcheck [21:41] all good, we can chat another tiem [21:52] thumper: with manifold output funcs, can we assume that they won't be called before start returns? [21:57] thumper: if that's something light and not requiring lots of thinking then I'm open [22:12] wpk: nah, we're all good at the moment, been on calls, so very async