[00:04] Unit tests enhancement by monkey patching utils.NewUUID : PTAL http://reviews.vapour.ws/r/537/ [01:02] jogging with god, that's deep [01:02] wallyworld_: get me to look at what? [01:03] axw: thumper fixed it, the previous fix for tools symlinks broke upgrades from 1.20 [01:03] oh :/ I'll take a look to see what I did wrong... [01:03] https://github.com/juju/juju/pull/1226 [01:03] thanks [01:04] axw: see comment on https://bugs.launchpad.net/bugs/1396792 [01:04] Bug #1396792: jujuc symlink creation broken on upgrade to 1.21.beta4

[01:05] ah, I see. === menn0_ is now known as menn0 === _thumper_ is now known as thumper === kadams54 is now known as kadams54-away [02:17] wallyworld_: FWIW upgrading from 1.21 -> 1.22 has the tools with their full paths, so the current code works... [02:19] ok, ta [02:19] it's just 1.20 then [02:20] wallyworld_: just 1.21 [02:20] which has landed [02:20] it was the 1.20 -> 1.21 upgrade that was buggered [02:20] yeah, that's what i meant [02:21] so we should be good now [02:21] great, ty [02:21] although I think I'd like some tests added to ensure that the symlinks are absolute rather than relative [02:44] wallyworld_ anastasiamac_: FYI ericsnow is doing some work to do with exposing AZs to units. this will be relevant for zone constraints [02:44] ok, ta [02:44] as his work will require that we store the zone in state [02:45] axw: is there a spec for his work do you know? [02:46] wallyworld_: I think it's https://docs.google.com/a/canonical.com/document/d/1yVlzgKqfhKccUbm3WcZBq7WnzglBygknscr_I_u4bUg/edit# [02:46] not sure if there's anything else [02:46] i'll read, thanks === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [03:08] review for fix of bug 1396796 please: http://reviews.vapour.ws/r/538/ [03:08] Bug #1396796: local provider all-machines.log has only machine-0

[03:08] thumper: ^^^ === kadams54 is now known as kadams54-away [03:10] * thumper looks === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away [04:04] wallyworld_: I this today while investigating other issues: bug 1396862 [04:04] Bug #1396862: Panic when deploying to the bootstrap node [04:05] oh joy [04:05] yeah, and I can't make it happen again [04:06] Fix Committed :-) [04:07] wallyworld_: so that's how you guys get so many bugs fixed.... [04:07] :-D === TheRealMue is now known as TheMue [04:10] axw: this wget certificate stuff is giving me the shits. i've tried several ways to get it to work. i even tried extracting the certificate from the running web server (in case I was using the wrong pem on disk) as per these instructions http://lnotestoself.blogspot.com.au/2011/02/wget-and-ssl-certificates.html [04:11] so far i've had no luck without the --no-check-certificate [04:12] wallyworld_: I'll have a play [04:12] right after I have some lunch [04:13] axw: ok. i had to generate the pem file as we weren't writing out the one for the state server from what i could see [04:13] i could have cut and pasted from env-get also [04:14] but writing it out on state server startup work, generating the same data as I got from using firefox to get the cert [04:16] * wallyworld_ bbiab [04:30] * thumper EODs [04:31] anyone seen "inconsistent definition for func" errors? [04:32] this happen during a merge attempt in Jenkins: http://paste.ubuntu.com/9261637/ [04:32] i can't see how it relates to the branch [04:37] wallyworld_: this works for me: juju get-env ca-cert > /tmp/ca-cert.pem && wget --ca-certificate=/tmp/ca-cert.pem https://localhost:17070/tools/1.22-alpha1.1-utopic-amd64 [04:39] menn0: weird. looks like a tooling issue [05:11] axw: that seems to work for me also, yet the cert extracted using firefox is different. no idea why [05:19] axw: and confusingly, it's not the same cert that's in StateServingInfo [05:19] wallyworld_: StateServingInfo has the server's certificate, not the CA certificate [05:20] the CA certificate signs the server cert; given the CA cert you can prove that the server is trusted by the CA [05:20] that makes sense, i know little about ssl sadly [05:21] axw: so to make it easy for wget i could do what rsyslog worker does and write out the cert when the state server starts up, or will that be a security issue? [05:22] i guess that ingo is already in env-get so no [05:22] wallyworld_: you shouldn't need to do that [05:22] CA cert is in the env config, you should just write it out as a temporary file when you write the wget wrapper [05:23] i didn't want to have to invoke juju get-env each time [05:23] but i guess i coud [05:23] wallyworld_: hold on, let me refresh my memory about your branch [05:24] right now i generate the wget wrapper each time [05:24] but i could do it once somewhere [05:24] I think this is happening infrequently enough that that is fine [05:26] ok [05:26] wallyworld_: if you have the ImageURLGetter return a CA certificate as well, it can use that. the API client already has the CA certificate in memory [05:27] yeah, i'll look into how to make it as nice as possible [05:27] first gotta read the news - Phillip Hughes died :-( [05:28] ah :/ === kadams54 is now known as kadams54-away [08:44] morning [08:51] jamespage: you ok for juju status meeting in 10 minutes? hopefully you got the invite [08:52] dimitern: can you take a look over bug 1395908? might be related to bug 1345433 which was fixed [08:52] Bug #1395908: LXC containers in pending state but no error message

[08:52] Bug #1345433: cloud-init network error when using MAAS/juju

[08:53] lxc containers that came up got no ip address [08:53] wallyworld_, sure, will have a look [08:53] hence they stayed in pending [08:53] ty [08:53] wallyworld_, is it reproducible ? [08:54] dimitern: not sure, happened obviously for that maas deployment, not sure if they tried again [08:54] i would have said it was fixed, so might be something new [08:54] wallyworld_, hmm.. I really like those bug reports :) [08:54] yeah [08:57] fwereade_: not sure if jamespage is around for juju status meeting [08:58] wallyworld_, it's 9am in the uk I think? worth waiting around a little [08:58] rightio [08:59] fwereade_: i won't log in to hangout until he pings back [09:02] fwereade_: (how/where) does juju prevent upgrading a charm when metadata changes incompatibly between revs? [09:02] axw, inside state, we check for... [09:02] axw, active relations that have changed or disappeared [09:02] axw, change in subordinacy [09:03] axw, not sure about anything else [09:03] fwereade_: cool, that's where I want to look. thanks [09:03] axw, Service.SetCharm(URL?) I think [09:04] yep === liam_ is now known as Guest33558 [09:12] wallyworld_, just coming - was travelling [09:12] sure, np [09:14] gnuoy: are you around for a juju status meeting? [09:15] wallyworld_, sorry, yes. Bad nights sleep last night and I'm still catching up [09:15] gnuoy: np at all, we're running alte, just joined the hangout ourselves [09:32] morning all [09:51] * fwereade_ going for a walk and a think [10:01] dimitern: I guess most of the Americans are off work today... [10:03] voidspace, thanksgiving? right [10:04] jam, are you around? [10:20] dimitern, jam thanksgiving I think [10:24] fwereade_, I suspected that, but then again - do they celebrate it in dubai ? :) [10:24] dimitern, probably not, but he does with his family (and I think he's marked as on holiday) [10:25] fwereade_, right, I've missed that [10:25] dimitern: ah, panic because I did "if err != nil" where I meant "if err == nil"... [10:25] dimitern: :-) [10:25] dimitern: so I was returning a nil subnet [10:26] voidspace, :) ah, there it is [10:26] rerunning the test... [10:26] dimitern: and now the test I marked with "XXX this should fail" [10:26] dimitern: now it fails [10:27] voidspace, \o/ [10:27] dimitern: so you were correct - the unique index *was* working, just failing silently [10:27] voidspace, good; sorry I didn't mention this earlier [10:28] voidspace, it took me a day to discover it initially :) [10:28] dimitern: well remembered [10:28] dimitern: sounds like a fun day [10:28] voidspace, to put it mildly, yeah :) [10:31] dimitern: the bad news is that the non-unique provider id triggering the failure is "" [10:32] dimitern: so I have to fix that... [10:32] dimitern: trying sparse [10:34] dimitern: currently indexes doesn't support sparse - I'll have to add it [10:34] voidspace, good to know unique indexes do not work with null values out of the box [10:34] voidspace, to mgo? [10:34] dimitern: no, just our code [10:34] dimitern: I have to add a new column to all the index definitions [10:35] voidspace, sparse: false ? [10:36] dimitern: yep, effectively [10:36] dimitern: I have to add a new sparse entry in the index struct - make it false for all the existing ones and true for the new one (subnetsC) [10:36] done it, just running the test now [10:37] field not column [10:37] voidspace, sgtm [10:38] dimitern: works [10:38] dimitern: I'll need to run the full test suite to make sure I didn't break anything else [10:38] I shouldn't have done though - I've just made what was the default before (Sparse: false) explicit [10:39] I'll run the whole test suite while I make coffee [10:40] voidspace, cheers === mgz is now known as mgz_ [11:40] jam: in an ec2 environment, i'm trying to connect to the state server using https. but the certificate common name set to "*" causes it to fail [11:40] ERROR: certificate common name ''*'' doesn''t match requested host name ''10.73.184.8'' [11:40] why do we use "*", and is there a way around it do you know? [11:43] fwereade_: not that you don't have enough to deal with, but is it possible your last branch (uniter-extract-operations) borked upgrades? [11:44] fwereade_: http://reports.vapour.ws/releases/2120 [11:47] mgz, well, I'm certainly not going to say it's impossible, and I admit I didn't try an upgrade with the final version of the code [11:47] mgz, will investigate [11:48] fwereade_: thanks, given all the jobs failed I'm pretty certain upgrade is unhappy, but haven't tracked down the whys yet [11:49] mgz, I see this in the logs: machine-0: 2014-11-26 23:34:18 ERROR juju.worker.upgrader upgrader.go:157 failed to fetch tools from "http://juju-dist.s3.amazonaws.com/testing/tools/releases/juju-1.22-alpha1-precise-amd64.tgz": cannot unpack tools: tarball sha256 mismatch, expected 1d45ea821ab50e719df68d7a887449ff5f63f7d3347b665f2e522a99703a2577, got 00c0b630d88383bf0866a5a9846b1b945cac1936b88242a163d2d84a6eed83aa [11:50] right, that's what I'm looking at [11:51] mgz, I don't believe I touched any related code [11:51] that feels like it should be a non-code issue... but it's only getting generated on the fly in the testing, so am not sure what's changed [11:51] it must be a fiddling in the simplestreams output in a change somewhere [11:51] mgz, if I'd broken it I'm 99% sure it would be the uniter failing to come up and complaining about an invalid state file [11:52] mgz, (which should not happen, and did not happen at least once, fwiw, but if you see it you should certainly blame me) [11:52] I'll file a bug with no specific blame and we can go from there, I'll also requeue the run to be sure [11:52] mgz, cheers [11:52] fwereade_: thanks [12:07] https://bugs.launchpad.net/juju-core/+bug/1396981 [12:07] Bug #1396981: Upgrade fails with tools sha mismatch

[15:30] mgz, rogpeppe1, if any of you is around, can you have a look at this MP for goamz please? https://code.launchpad.net/~dimitern/goamz/update-aws-api-version-to-latest/+merge/243057 [15:30] dimitern: sure [15:31] mgz, ta! [15:32] dimitern: i'm +1 in principle but i'd like assurance that that API version works across all the likely endpoints that people are using goamz for [15:32] rogpeppe1, how do you suggest to guarantee that? :) [15:32] dimitern: as a start, i know that aws has different behaviour in different regions, so it would be good to test that that version is available in all aws regions [15:33] dimitern: i've no idea, but i believe that amazon isn't the only provider that people use goamz with [15:33] rogpeppe1, the api version is the same for all regions [15:33] dimitern: ok [15:33] dimitern: i'd like to get it ok'd with gustavo before landing it. [15:34] rogpeppe1, true, but also there are a lot of forks of goamz for various reasons - not keeping up with the api version changes for now [15:34] s/for now/for one/ [15:34] rogpeppe1, sure, I intend to [15:34] dimitern: busy now, but i'll take a proper look at some point [15:35] rogpeppe1, np, when you can, cheers [15:46] dimitern: have you tested this on a non-default-vpc account? I guess we're fine in that case the way amazon do their api versioning, but would be nce to check :) [15:47] actually changing juju to use this goamz ver has some risk, but the change in this codebase would be fine... the complication being that there are also github versions [15:53] dimitern: ready for re-review: http://reviews.vapour.ws/r/534/ [15:54] voidspace, looking [15:55] mgz, I have no access to such an account [15:55] mgz, but you're right - it should work just as well [15:55] mgz, I did test it on a vpc-only region (us-east-1) though [15:57] oh ffs! my full-stop key stopped working [15:57] dimitern: restrict yourself to one sentence at a time and you'll be find [15:58] *fine [15:58] dimitern: and avoid attribute access / method calls [15:58] voidspace, :D and don't bother trying to type and struct fields [15:59] yeah :) [15:59] ... [15:59] sorry, just showing off [16:00] dimitern: reviewed [16:01] mgz, thank you [16:08] voidspace, reviewed [16:10] dimitern: thanks [16:10] dimitern: CheckValid was jam's name [16:10] dimitern: I'm happy with Validate - don't mind [16:11] voidspace, yeah, I'll leave the decision whether to rename it to you, I don't really mind [16:12] dimitern: thanks! [16:12] voidspace, I do mind about adding isAliveDoc assert in EnsureDead though :) [16:12] dimitern: fixing the other two, will make a decision on the third [16:12] dimitern: done, just running tests... [16:12] voidspace, cheers [16:12] and then I'll merge the bugger [16:13] voidspace, as you were testing with goamz recently, would you like to have a look at my branch at some point? https://code.launchpad.net/~dimitern/goamz/update-aws-api-version-to-latest/+merge/243057 [16:14] dimitern: sure [16:14] voidspace, ta! [16:14] dimitern: 1025 lines!! [16:15] voidspace, yeah, but most of them are simple renames [16:15] ah right [16:15] :-) [18:20] EOD [18:20] g'night [18:44] anyone here able to tell me where on the launchpad website I can find lp:charms/juju-gui ? [18:44] so confused by it haha [18:46] oop found it :) [18:46] ask for help and you find it :D always happens haha === kadams54 is now known as kadams54-away === kadams54 is now known as kadams54-away [20:22] menn0: I caught the tail end of what sounded like an interesting bug you hit? [20:22] waigani: the one that happened during a merge attempt? [20:23] http://paste.ubuntu.com/9261637/ [20:23] it only happened once and the next attempt succeeded [20:23] and as mwhudson also found, Google has nothing on it [20:24] hmph, interesting === benonsoftware is now known as \b === \b is now known as benonsoftware [20:56] menn0: adds env-uuids to statusDoc: http://reviews.vapour.ws/r/543/ [20:59] waigani: sorry... I need to get this CI blocker sorted. I might be a little while. [20:59] menn0: np === mup_ is now known as mup === kadams54 is now known as kadams54-away [23:17] This is the fix for the CI blocker: http://reviews.vapour.ws/r/544/ [23:17] review please [23:31] thumper or wallyworld_: can you have a look at review 544 pls? [23:31] sure [23:31] wallyworld_: thanks [23:40] menn0: sorry, just got off phone, looking now [23:50] wallyworld_: np [23:53] wallyworld_: good catch re the error message. that needs to change with this change in behaviour [23:53] yeah, great, i thought so