[00:03] GOD DAMN CI BOT [00:03] let me commit the fix that fixes that flakey test ... bastard! [00:05] menn0: yes, we need to move off code.google.com [00:05] i can take that if everyone else is busy [00:05] i think there is a bug [00:05] or maybe a few === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away [00:33] thumper: ? [00:35] ARGH, failed again! [00:36] can someone please review https://github.com/juju/juju/pull/2321 [00:36] this fixes the panic that causes the tests to fail === 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 [01:04] wallyworld: yes? [01:05] wallyworld: gah... that damn meeting [01:05] wallyworld: I've missed it every time since daylight savings [01:05] wallyworld: we need to push it out another hour === kadams54 is now known as kadams54-away [01:52] Bug #1455284 was opened: Juju run doesn't displays returncode 0 [02:08] davecheney: got a few minutes? [02:18] thumper: sure [02:18] now ? [02:19] yeah [02:19] 1:1 hangout? [02:19] give me 2 mins [02:19] downstarirs [02:19] kk [03:07] wallyworld: do you know what the regex for MAAS storage tags is? [03:32] axw: has bug 1455224 passed CI yet? [03:32] Bug #1455224: TestListBlockDevicesDeviceFiltering fails on ppc64el/gccgo

[03:33] thumper: nope [03:34] axw: actually, yes [03:34] http://juju-ci.vapour.ws:8080/view/Juju%20Revisions/job/run-unit-tests-trusty-ppc64el/ [03:34] yes, sorry [03:34] just realised I looked at the wrong place :) [03:34] * thumper marks it fix released [03:34] thanks [03:35] np, I'm trying to land something on 1.24 [04:59] axw: sorry, missed ping, after vivid upgrade, quassel doesn't make any sounds :-( [04:59] STORAGE_REGEX = re.compile( [04:59] r"(?:(?P[a-zA-Z0-9]+)\:)?" # Optional label [04:59] "(?P[0-9.]+)" # Mandatory size [04:59] "(?:$(?P[^)]+)$)?", # Optional tag list between parentheses [04:59] re.VERBOSE) [04:59] that's what it is currently [05:01] so, basically anything other than a right parenthesis? I guess it's processed further down the line again [05:01] wallyworld: thanks. I'll just leave it as it was, but I've added internal whitespace checking to validation [05:01] ok === kadams54 is now known as kadams54-away [05:41] axw: wallyworld: storage add, next layer :D http://reviews.vapour.ws/r/1702/ [05:41] looking [05:43] axw: tyvm ;D [05:44] axw: i was not too sure about input as map of map of map is scary [05:44] anastasiamac: yes, I'm about to comment on it :) [05:44] axw: i seem to recall seeing params structs [05:44] axw: would prefer to use that :D [05:44] axw: oh, must b gr8 minds and all... [05:44] :P [05:49] anastasiamac: couple of comments, please change the signature and revert apiserver code - then I'll take another look [05:50] axw: tyvm :D [05:50] anastasiamac: if you did that map^3 thing because it made testing simpler, add a helper in the test code instead [05:51] axw: i did this thing coz i was coping from a place that had it :D [05:51] anastasiamac: can you point me? just curious [05:52] axw: annotations api [05:52] * anastasiamac hides [05:52] ooh, i like hide and seek [05:52] hm yeah, that's kind of bad :) [05:53] wallyworld: can't hide from u :D [05:53] it's only in code though, wire format is fine [05:53] axw: :) [05:54] axw: i wrote it originally to have ErrorResults as rturn [05:55] axw: but then we had a discussion about having success messages [05:55] axw: actually could u do a quick hangout? [05:55] sure [05:56] see you in tanzanite [05:56] coming [07:41] Bug #1455368 was opened: Expose juju version to the hook environment [08:15] dooferlad, hey there [08:16] dooferlad, I've discovered an issue yesterday with cloudinit on precise - it seems after the centos support has landed some of my workarounds for supporting 0.6.3 c-i (the one in precise) were dropped or changed [08:17] dimitern: yikes! [08:17] dooferlad, I've also had mixed success with bootstrapping on precise and then adding a trusty machine; it seems the precise cloudinit has been upgraded to 0.7 [08:18] dooferlad, so can you spin up a few tests locally with master on maas with precise and trusty + lxc/kvm containers? [08:18] dooferlad, if you can see issues in the /var/log/cloud-init*.log around apt-get install, I think we should file a bug [08:19] dimitern: I don't think I have tested with multiple Ubuntu releases before. Fun! [08:19] dimitern: will take a look [08:19] dooferlad, :) cheers [08:20] dooferlad, the issue is around using e.g. apt-get install --target-release precise-updates/cloud-tools cloud-utils - it becomes apt-get install '--target-release foo/bar cloud-utils' [08:20] for cloud-tools packages [08:30] dimitern: so how do I ask juju to add a machine running $ubuntu_version? [08:31] dimitern: or do I just create it in maas then add-machine with ssh:? [08:49] dooferlad, assuming you "default-series" is missing from envs.yaml, it defaults to yours or trusty (not sure); when you add-machine --series precise it should spin up another node [08:50] dooferlad, you'll need precise images on maas up to date as well [08:51] dimitern: thanks. Will start pulling images [09:04] dooferlad, standup [09:05] dimitern: http://reviews.vapour.ws/r/1694/ [09:05] dimitern: I have a review from Eric - but in case you want to take a look to [09:55] perrito666: ping [10:00] Frankban pong [10:01] perrito666: hi, I am getting those errors while trying to merge a branch: http://juju-ci.vapour.ws:8080/job/github-merge-juju/3280/console [10:02] perrito666: this is due to the fact that my branch updates to the last charm.v5, which includes the new UpdateStatus hook kind, which in turn is not mentioned when validating hooks in worker/uniter/hook/hook.go:Validate [10:02] perrito666: so, to move forward, I can either 1) comment out that new kind in charm.v5, or 2) just add it in validate (perhaps line 50?) [10:03] frankban: that is master right? [10:04] perrito666: what is master? [10:04] frankban: you are trying to merge with master [10:04] perrito666: yes [10:05] frankban: pass me the v5 version please [10:06] perrito666: 39463053128b672308c459958d00f260b58ce79e (current master trunk) [10:06] perrito666: your change is in https://github.com/juju/charm/pull/122/files [10:07] perrito666: UpdateStatus is the offending hook not recognized by core [10:09] I know [10:16] frankban: there https://github.com/juju/juju/pull/2133 [10:17] that branch, which was waiting on other work to land, is now merging [10:17] it contains what you need [10:17] :) [10:18] perrito666: great [10:18] perrito666: I'll merge trunk back after it lands and try again with my branch [10:19] frankban: until recently, the head of charm.v5 had other issues that made it impossible to merge [10:19] perrito666: out of curiosity, what issues? [10:20] frankban: worker/uniter/hook/hook.go:54: undefined: hooks.StorageDetached [10:20] that changed names [10:20] it was changed in 1.24 but not in master so until it was fwported we could not use charm.v5 tip [10:21] perrito666: ok [10:22] perrito666: so, good timing [10:56] Bug #1455445 was opened: Instance ERROR status not noticed by Juju

[10:56] oh we noticed... we just didn't care ;) [10:59] Bug #1455445 changed: Instance ERROR status not noticed by Juju

[11:11] Bug #1455445 was opened: Instance ERROR status not noticed by Juju

[11:25] natefinch: are you talking to the bot again? [11:26] Bug #1454143 changed: Deployer failure upgrading to proposed 1.23.3 from 1.23.2 client

=== liam_ is now known as Guest15928 [13:07] wwitzel3: ping [13:12] ericsnow: pong [13:18] nate died... [13:19] mgz: that is what children do to you [13:26] ericsnow, can you help with bug 1454678. This issue might be a dupe of bug 1408467 [13:26] Bug #1454678: "relation-set --file -" doesn't seem to work

[13:26] Bug #1408467: juju help-tool does not reflect all options available to relation-set

[13:29] Bug #1455490 was opened: Panic: no reachable servers in unit test

[13:30] sinzui: I don't think it's a dupe, but I won't be able to look at it for a little while [13:30] thank you ericsnow [13:42] Bug #1452796 changed: template download failed

[13:42] Bug #1455222 changed: juju doesn't see MAAS' "failed deployment" state

[13:54] Bug #1452796 was opened: template download failed

[13:54] Bug #1455222 was opened: juju doesn't see MAAS' "failed deployment" state

[14:00] Bug #1452796 changed: template download failed

[14:00] Bug #1455222 changed: juju doesn't see MAAS' "failed deployment" state

[14:42] wwitzel3: ericsnow: bonus points: make the timing of the registration more explicit than init so we can control ordering [14:44] wwitzel3: ericsnow: see bug 1452422 for why that might be important. init registrations are getting overwritten by --metadata-source which should not be happening [14:44] * perrito666 sees ordering and time and panics [14:44] Bug #1452422: Cannot boostrap from custom image-metadata-url or by specifying metadata-source

[14:45] katco: k [15:07] I am SO going to build one of these https://pbs.twimg.com/media/CDC9FfDWEAAYgfS.jpg:large [15:16] perrito666: i often dream of doing something like that, but the steady stream of old computers makes it too easy to use them as servers. [15:16] katco: that seems to be something that can be built here for 1k USD [15:17] given the right prices :) [15:19] perrito666: $1k > 0 ;p [15:28] lol [15:29] katco: that is true, I dont have a stream of old computers (yet, if I did, I would make a rack like that one) [15:29] perrito666: i tend to get a new one every 5 years or so [15:29] perrito666: and the servers last that long easily [15:29] perrito666: so i use my computers ~10 years :) [15:30] katco: I have not used a desktop in 5 years [15:30] or more [15:30] I do change my laptop every one or two years [15:30] perrito666: i always have a desktop for the graphics card [15:31] katco: I felt behind too much on my hardware so I only play in consoles now [15:31] bleck consoles [15:31] bleck [15:31] katco: consoles are non updatable hardware, which is a very convenient thing [15:32] especially living in noNewHardwareGoesTroughCustomsLand [15:33] haha [15:36] I really deeply hate juju local [16:33] cherylj, ping? [16:49] mattyw: what's up? [17:53] * natefinch is totally rocking some actions for his CI tests. [17:54] /natefinch totally rocks [17:54] :) [17:55] they really make this test way easier. I'm testing log rotation, so I'm writing a charm that I can tell to dump a bunch of data into the logs, and then ask it how big the log file is :) [17:56] sweet [17:56] jw4: docs are pretty good, too. I was afraid it would be difficult to figure out how to do it, but it's totally not. [17:57] natefinch: bodie_ is responsible for all the good docs [17:57] nice, well done, bodie_ [17:57] natefinch: glad to hear its working well [17:58] jw4: I haven't actually tested it yet, but writing it has been pretty painless :) Not really any different than writing a regular hook. [17:58] natefinch: heh.. well I'll be happy to get any feedback along the way [17:59] ericsnow: ping? [17:59] cherylj: hi [18:00] ericsnow: I have some questions about systemd unit files. Got a few minutes? [18:00] cherylj: sure [18:01] hi. first time using such a (semi) automated review/merge system. my PR was marked "ship it", am I supposed to do anything else now? [18:01] jw4: I do think it's kind of weird that the input is schema'd and the output is not. I kept looking to try to find how to define in the metadata what it returns. [18:01] * bodie_ tips fedora to natefinch [18:02] bodie_: fedoras are not allowed here ;) [18:02] hehe [18:02] natefinch: yeah that was discussed quite a bit - rationale for not using output schemas is lost in the mists of my memory [18:02] jw4: haha... that's ok [18:02] Argon]: I (or someone else) will need to do the $$merge$$ comment for you [18:03] jw4: I figured it was one of those things that went back and forth 1000x :) [18:03] depends on the rationale for schemas [18:03] although I have thought it would be nice for the charm to document what would come back from various actions [18:03] and, obviously for building UI around it on the GUI, it would be good [18:04] I guess the input schema makes web page controls configurable... but for output you don't want to write to a number picker or whatever. [18:04] mgz: ah ok, that's fine, thanks [18:04] but for the purposes of validation, clearly the contents of the charm are expected to be trustworthy [18:04] Argon]: was going to get someone else to look at it as well, but there's no real need [18:06] Argon]: I was confused about storage types/virt types initially, but have now read up again and basically understand [18:12] mgz: no need to hurry. but this turns out to be a little problem (not related to merging the patch), because while I can patch my local juju, I can't patch the agent version juju is installing on the remote machine [18:12] so I can bootstrap an amazon env using t2 instances (because my local juju is patched), but subsequent actions like "deploy wordpress" (I'm following your introduction tutorial) fail because it, once again, tries to use paravirt images on a hvm type instance [18:14] Argon]: so, there's a developer hack around that [18:14] generally, we pick a released juju out of streams.canonical.com that matches the version of your local client [18:15] rick_h_: little bit of feedback on jujucharms docs... it's really hard to find the docs on hook tools and environment variables (super important and used pretty much every time you write a charm). They're under "how hooks are run"... which was not an obvious place for me to look. They're important enough that I think they deserve their own page. [18:15] but, you can tell juju to use your own streams instead, which can be whatever juju binaries you want === kadams54 is now known as kadams54-away [18:15] natefinch: please file a bug on the github.com/juju/docs on that please. We'll see if we can get the docs folks to look at/split that up. [18:16] rick_h_: cool, will do [18:16] Argon]: limits you to arches you can built locally, but that's likely not a problem for ec2 (except when I'm using my arm chromebook... why no arm instances ec2) === kadams54-away is now known as kadams54 [18:21] mgz oh that's fine, I could spin up a instance and build it there anyway. however, I'm not exactly sure how to specify this, are you talking about the config var "image-stream"? [18:23] Argon]: so, the robust way is with the `juju metadata` plugin and then the agent-metadata-url pointing at the streams you generate [18:24] Argon]: for throwaway environments, you can just bootstrap with --upload-tools param [18:25] mgz: great, thanks! [18:30] mgz: let me correct your. there are no arm instances that "run" in ec2. there are several that walk so slowly that it takes 12 hours to build a juju and jujud [18:30] ehehe === kadams54 is now known as kadams54-away [19:00] ericsnow: just fyi - the containter still didn't stop, but I didn't get that same error this time. Looking more... [19:01] cherylj: k [19:13] upgrade-juju --upload-tools will upgrade every time even if the toos have the same version? [19:15] Can anyone think of a way for a charm to get something written to the machine log? [19:15] natefinch: ? [19:16] perrito666: I'm testing log rotation with a custom charm... but it occurred to me that while the charm's output is written to the unit agent's log (making it easy to spam the log to make it big), there's no direct way for a charm to output to the machine agent's log [19:17] you mean to machine-N.log? [19:18] yeah [19:19] mmm, nope, cant think of a nice way [19:19] I suppose I could do something like stop the machine agent, manually dump a bunch of data to the log file, and then restart the agent... but I'd prefer a way to do so through normal log output. [19:30] Bug #1455627 was opened: TestAgentConnectionDelaysShutdownWithPing fails

[19:33] jw4, bodie_: can I call action-set multiple times? So like action-set result-map.foo=bar action-set result-map.baz=bat etc? [19:33] natefinch: yes [19:33] overwrites if you collide keys, btw [19:33] jw4: I thought so. It seemed implied in the docs, but not explicitly stated. [19:34] bodie_: ok... but result-map.foo=x and result-map.bar=y don't count as collisions, correct? [19:34] correct [19:34] ok [20:00] Bug #1455628 was opened: TestPingTimeout fails

[20:13] I wish juju help action do would actually do the right thing [20:13] 'juju help action do' that is.... it just returns the help for action generally [20:14] natefinch: juju help action do sounds like something yoda would say [20:14] lol [20:14] yeah, I kinda hate the way nested commands end up being reverse polish notation [20:15] natefinch: http://i.imgur.com/qVbuAPU.jpg [20:15] here for you to hang on the wall [20:16] rof,l [20:16] awesome [20:24] jw4: so, I guess we don't automatically log the output of actions the way we do with hooks? [20:26] What's the best practice for upgrading juju (from beta2 -> beta3)? After install, I run juju upgrade-juju --upload-tools so my current (local) environment is in sync. After that, though, juju status, ssh, etc all hang [20:28] jw4: nvm, may be my fault [20:37] wow, now there's an interesting bug [20:38] if you forget to make your action hooks executable.... no one actually complains, but they just don't run [20:38] jw4: ^ [20:39] aisrael, that sounds like a bug === kadams54-away is now known as kadams54 [20:50] alexisb: ack. I'll file a bug [20:55] wwitzel3: ping:q === kadams54 is now known as kadams54-away [21:06] Bug #1455646 was opened: Juju doesn't complain if action files aren't executable

[21:48] cherylj: did that work out with LXC? [22:06] ericsnow: I had to go take my daughter to the doctor for a well check, so I haven't been able to look much more at it. [22:06] cherylj: np :) [22:06] But it didn't shut down and I'm not sure why [22:06] ericsnow: I'm seeing a lot of "peer has disconnected" errors and I don't know if that's a separate issue [22:07] ericsnow: what is cloud-final? [22:09] katco: ping [22:09] wwitzel3: pong [22:09] katco: have time to jump in to moonstone real quick? [22:10] cherylj: the cloud-init service or something like that [22:10] wwitzel3: for a short bit, need to EOD [22:10] katco: shouldn't take long [22:10] katco: thanks [22:15] happy weekend all! I am off to pick up kiddo so granny can head to a dinner party [22:15] cheers, happy weekend === kadams54 is now known as kadams54-away