[01:22] waigani: would you mind looking at: http://reviews.vapour.ws/r/2456/ [01:22] menn0: sure [01:22] waigani: it's a fix for intermittent allenvwatcher test failures that i've noticed [01:22] waigani: there's another problem to fix as well but this one issue is happening more often [01:25] menn0: shipit [01:36] waigani: thanks [01:39] i love the fact that all our upgrade and provisioner tests use "ppc64le" as the "i cannot possibly happen" operating system [01:41] davecheney: isn't it easier to support impossible operating systems? :D [02:06] we're basically using quantal-ppc64le as our "nope, cannot happen" sentinal [02:37] davecheney: good think the ubuntu architecture is called ppc64el then [02:37] *THING [02:37] doh can't brain today [02:49] axw_: godeps: cannot get information on "/home/dfc/src/launchpad.net/gwacl": bzr revision-info has unexpected result "244.1.1 andrew.wilkins@canonical.com-20150810080433-4l0x47qsc16ure7c\n" [02:50] mwhudson: oh god [02:50] don't mention the war [02:50] or the worst joke in the world that continues to keep on giving [03:09] state tests now take 450+ seconds to pass [03:09] % go test ./state/... [03:09] ok github.com/juju/juju/state 438.504s [03:09] golf clap [03:09] FAIL github.com/juju/juju/state/leadership 600.006s [03:09] fucking fantastic [03:12] Bug #1487947 opened: provider/ec2: localServerSuite.TestNetworkInterfaces [06:46] Bug #1487947 changed: provider/ec2: localServerSuite.TestNetworkInterfaces [07:16] rogpeppe, hey, have you seen https://code.launchpad.net/~dimitern/godeps/fix-bzr-revision-info-parsing/+merge/268223 ? [07:16] dimitern: nope :) [07:16] dimitern: only just back from 2 weeks' holiday [07:17] rogpeppe, have a look when you have a moment :) it's just 2 characters change [07:17] dimitern: looking [07:17] dooferlad, TheMue, hey guys, it seems net-cli is blessed for merging in master [07:20] dimitern: interesting. out of interest, do you actually want to use non mainline revision numbers as dependencies? [07:21] rogpeppe, I suppose not, but at least when there is such a case now godeps refuses to work anymore [07:21] dimitern: yeah, i agree it shouldn't fail in that case [07:21] rogpeppe, maybe print a warning when there are dots in the bzr revision and go on? [07:22] dimitern: i think i'll just go with your fix actually. if you really want a non mainline revision, you can have one [07:22] dimitern: it should be obvious in the .tsv file anyway [07:22] rogpeppe, sounds good [07:26] dimitern: reviewed [07:31] rogpeppe, thanks! [07:32] rogpeppe, I think you should manually merge the MP, as there's no bot to pick it up [07:38] dimitern: merged (ooo, my bzr-fu is getting rusty!) [07:40] rogpeppe, :) mine too [07:57] dimitern: great! Lets do it and go drink some beer! [07:57] * dooferlad knows that TheMue would rather drink whisky [07:57] dooferlad: good idea, got time next week monday. wonna meet in London? [07:57] \o/ [07:58] dooferlad: ah, you know me right [08:00] TheMue: I think you meen tuesday. Well, maybe Monday evening... [08:01] dooferlad: yeah, Monday evening [09:01] dooferlad, voidspace, standup? [09:01] dimitern: trying, it's not letting me in :-/ [09:01] restarting the browser [09:01] voidspace, oh boy :/ monday hangout issues [09:02] dimitern: not having used this machine for six weeks issues too... [09:02] 350mb of updates [09:02] including a new firefox - that maybe the issue [09:02] that looks better [10:03] dimitern: ready [10:03] voidspace, omw [11:33] fwereade, hey, are you around today? [12:49] dimitern: btw, the provider I meant has been the vsphere one. i has a partial implementation of the networking interface [12:53] TheMue, ok, so for vsphere then SupportsSpaces will be false, but the other method will remain [12:53] dimitern: yep, did it that way [12:54] TheMue, +1 [13:11] hi, im having some problem with juju-gui [13:11] machine-0 logs says: error stopping *state.Multiwatcher resource: unit not found [13:12] everytime i load juju-gui on browser [13:12] and no icons appears [13:12] this is a juju bug or gui? [13:18] i open this issue: https://bugs.launchpad.net/juju-gui/+bug/1485249 [13:18] Bug #1485249: Juju gui is not loading. [13:56] anyone know about the openstack provider? [13:57] dimitern: ^ [13:59] Bug #1486106 changed: TestNetworkInterfaces fails on vivid/wily

[13:59] in particular at this moment I'm interesting in understanding the different roles of the tenant-name (taken from env var OS_REGION_NAME) and region (no env var) attributes [14:01] rogpeppe: you have that mixed up, tenant-name is taken from OS_TENANT_NAME [14:02] it's a user-specific grouping of resources [14:02] mgz: so they are - not sure how i read that wrong [14:02] whereas regions are a cloud-level seperation [14:03] mgz: for some reason i'd copied/pasted the wrong env vars and didn't check back to the source [14:03] mgz: i did think it was potentially very confusing :) [14:04] mgz: one weird thing: why does openstack fall back to using the ec2 passwords? [14:07] rogpeppe: there's a semi-compatible ec2 api, so many of the rcs set the same vars for ec2 [14:07] mgz: it seems like a potential security leak to me [14:08] mgz: and i don't see any other values that do the same thing [14:08] fwereade: ping [14:08] wwitzel3, pong [14:09] fwereade: katco isn't feeling well, are you able to move up our discussion to now? so she can take off and get some rest [14:09] wwitzel3, sure, let's [14:09] mgz: and surely if it was using AWS_SECRET_ACCESS_KEY it ought to use AWS_ACCESS_KEY_ID too? [14:09] fwereade: we are in moonstone, https://plus.google.com/hangouts/_/canonical.com/moonstone?authuser=1 [14:12] rogpeppe: the names used are just based on what was looked at in practice by the clients [14:15] and there's not really a security issue, the access ley/secret key are not passed along unless you say you're using that authmode [14:15] mgz: do you think it might be reasonable to drop support for using AWS_SECRET_ACCESS_KEY and EC2_SECRET_KEYS for the openstack password? [14:15] so you need to be pretty deliberate about wanting it [14:16] mgz: isn't userpass the default auth mode? [14:17] mgz: so ISTM that it'll send that password in the usual case [14:17] and it's basically always going over https to somewhere that won't be loging your auth attempts [14:18] you'd need to have and endpoint url manually set to something hostile, or dns poisoned [14:18] mgz: from the p.o.v. of the user that's not necessarily true - they don't know what's at the other end [14:18] and also have accidentally sourced your ec2 creds [14:18] I don't think the ec2 provider is robust against that [14:19] we don't check that the ec2 endpoint is really amazon [14:19] mgz: i think we do, don't we [14:19] mgz: 'cos we use https [14:19] mgz: without using InsecureSkipVerify [14:19] ? [14:19] that just verifies it has been signed by something in your chain [14:20] it doesn't verify what it is. [14:20] mgz: well, same security as any website :) [14:20] mgz: and, yes, i'm well aware that https security is a farce [14:21] the ec2 signing mechanism is better though, in not passing along the secret [14:21] mgz: but we all blithely go on assuming that verisign et al are somewhat reliable [14:22] mgz: i mean, fair enough i guess if people actually use AWS_SECRET_ACCESS_KEY for their openstack password, we don't wanna break that [14:22] mgz: it just triggers my Something Wrong Here sense. [14:23] I think it's likely not really used any more, all the current rc things will at least include both, and generally only the OS_ versions [14:23] just whether it's worth futzing with [14:24] mgz: i'm moving the env vars into the openstack provider to do the environschema stuff, and it made me wonder... [14:25] mgz: i mean, maybe goose.v1/identity should export the env var names [14:29] Bug #1465404 changed: worker/provisioner: fail lxcBrokerSuite.TestStartInstanceLoopMountsDisallowed

[14:29] Bug #1473209 changed: github.com/juju/juju/service/windows undefined: newConn

[14:35] Bug #1465404 opened: worker/provisioner: fail lxcBrokerSuite.TestStartInstanceLoopMountsDisallowed

[14:35] Bug #1473209 opened: github.com/juju/juju/service/windows undefined: newConn

=== redelmann is now known as popi_ === popi_ is now known as redelmann === redelmann is now known as s0plete === s0plete is now known as redelmann [14:39] fwereade, hey [14:40] fwereade, does this look in line with what we discussed? http://reviews.vapour.ws/r/2463/ [14:41] mgz: this is what i've got: http://paste.ubuntu.com/12184056/ [14:41] Bug #1465404 changed: worker/provisioner: fail lxcBrokerSuite.TestStartInstanceLoopMountsDisallowed

[14:41] Bug #1473209 changed: github.com/juju/juju/service/windows undefined: newConn

[14:41] mgz: does that look reasonable to you? [14:42] mgz: actually, i should avoid copy/pasting those env vars [14:45] rogpeppe: you have EC2_SECRET_KEYS ... with an S [14:46] mgz: that's what goose uses [14:46] also, not really to do here, but we should add tenant-id already [14:47] heh, in which case you probably can just delete it, as it's been wrong for a while then [14:47] mgz: :) [14:47] mgz: i suspect the same of AWS_SECRET_ACCESS_KEY too [14:47] Bug # changed: 1458717, 1461393, 1463641, 1467372, 1474788, 1478660, 1484617 [14:47] mgz: but i'm actually thinking i don't want to risk breaking anything at all [14:48] rogpeppe: no, that one is right, the old naming there was just dumb [14:48] mgz: i know it's right - i just think that using an AWS name for openstack creds is silly [14:49] mgz: particularly when it doesn't use AWS_ACCESS_KEY_ID for one of the user names [14:49] it was the only way to use non-password auth on the old hp cloud [14:49] as there aren't OS names for the different values [14:51] mgz: ah, makes sense [14:59] mgz: ok, here's a review for you: https://github.com/go-goose/goose/pull/14 [14:59] Bug #1459064 changed: worker/leadership: data races in test and code

[14:59] voidspace, dooferlad, TheMue, please have a look at these last 2 PRs we need to unblock merging net-cli into master: http://reviews.vapour.ws/r/2463/ and http://reviews.vapour.ws/r/2464/ [14:59] Bug #1460893 changed: many unhandled assigned values

[15:01] * dimitern steps out, bbl [15:02] hello all - new starter on the Juju Sapphire team today. [15:02] rogpeppe: looks fine. as I hinted earlier, there are some bigger bits that could do with fixing in the os auth stuff, but nothing that blocks this change. [15:02] frobware: hey there! [15:02] mgz: thanks [15:03] frobware: another Andrew? [15:03] mgz, tis the way... [15:06] fwereade: we didn't mean to chase you away! :) [15:11] Bug #1459064 opened: worker/leadership: data races in test and code

[15:11] Bug #1460893 opened: many unhandled assigned values

[15:14] mgz: could you land my goose branch for me please? i'm not a member of goose collaborators currently [15:17] Bug #1459064 changed: worker/leadership: data races in test and code

[15:17] Bug #1460893 changed: many unhandled assigned values

[15:18] anyone know about juju-gui? [15:18] im trying to debug an issue [15:18] redelmann: probably better to ask in #juju [15:19] thank natefinch [15:19] dooferlad: I see you reviewed the first one, are you looking at the second too? [15:19] rogpeppe: erm... can't remember exactly but the bot may actually just still check if you're a juju memeber [15:19] voidspace: yes [15:19] rogpeppe: added the comment anywa [15:19] mgz: thanks [15:21] dooferlad: ok, I'll leave it to you :-) [15:21] thanks [15:21] voidspace: I haven't graduated yet. I suggest you look at the first one first - it seems to have been committed with the second as well. [15:21] voidspace: then you know which bits to ignore :-) [15:22] ah [15:22] I did think the second had some extraneous stuff in it [15:23] dooferlad: dimitern: first looks good and it tells me which bits of the second I can ignore too [15:24] in which case the second looks trivial [15:24] dooferlad: although I missed the concern you raised, need to read the code agian! [15:24] *again [15:29] Bug #1488139 opened: juju should add nodes IPs to no-proxy list [15:35] Bug #1488139 changed: juju should add nodes IPs to no-proxy list [15:48] Bug #1488139 opened: juju should add nodes IPs to no-proxy list [16:01] is the answer to why godeps is used rather than godep a simple one? [16:01] welcome frobware ! [16:05] bodie_: godeps is older than godep [16:07] I see. thanks mgz o/ [16:26] mgz and anyone else: fancy a smallish review of a change to the openstack provider code? http://reviews.vapour.ws/r/2465/ [16:27] katco & moonstone: fwereade texted me about an hour ago his power's out and wanted to convey this apologies [16:27] voidspace, dooferlad, thanks for the reviews guys! [16:30] rogpeppe: looks fine in general, the descriptions are a bit all over the place though, do you want nitpicks? [16:30] mgz: specific suggestions would be great [16:30] mgz: i copied the descriptions from the boilerplate [16:31] mgz: much appreciated [16:31] ah, so they're now it two places? that's a bit unfortunate. [16:31] mgz: when i get some extra time, the plan is to generate the boilerplate from the schema [16:31] how much will you hit me over the head if I ask you about generating the boilerplate from the schema? [16:32] eheh, you win. [16:32] :) [16:33] mgz: i'm also not liking my decision to go with partial sentences for the descriptions. using go-like full sentences seems like it's a better plan in the end. [16:34] mgz: but these things are fairly easy to change [16:38] rogpeppe: review'd [16:38] mgz: ta v much [16:39] mgz: not sure about mentioning that control-bucket will be auto-generated - that's kinda true of every parameter that's optional. [16:41] I'm a little antsy on that one as setting it can be actually harmful... but it can also just be required [16:46] when would it be required? [16:47] mgz: perhaps a comment in the description saying "(do not set this unless you know what you are doing)" might be appropriate [16:47] rogpeppe: yeah, but leave it for now, don't need to futz with here. [16:48] mgz: k [16:48] we need it for ci testing on canonistack, I *think* it's related to being able to cleanup reliably === urulama is now known as urulama__ [16:57] Bug #1488166 opened: Service with just one unit left which doesn't think it's the leader

[17:21] bbl groceries shopping [17:44] ericsnow: getting this error when running featuretests: charm not found in "/home/nate/src/github.com/juju/juju/testcharms/charm-repo": local:quantal/workload-actions [17:46] natefinch: is it actually there (/home/nate/src/github.com/juju/juju/testcharms/charm-repo/quantal/workload-actions)? [17:47] natefinch: perhaps still named "proc-actions"? [17:47] natefinch: (which should be fixed) [17:49] ericsnow: doh, yep [17:49] I blame wwitzel3 [17:49] natefinch: :) [17:51] ericsnow: I'll submit a PR to rename them [17:51] natefinch: sounds good [18:19] :( [19:28] I think I got whatever katco has. I should remember to wear a mask to hangouts from now on. [20:54] Bug #1488245 opened: Recurring lxc issue: failed to retrieve the template to clone

=== natefinch is now known as natefinch-afk [21:47] ping http://reviews.vapour.ws/r/2458/ [22:22] davechen1y: ship it, I am happy that that is being worked :) [22:22] there is only a typo in the description [23:10] perrito666: thanks