[00:26] does juju have a wrapper around exec.Command? [00:30] katco: iirc there is one in utils [00:30] sadly there is more than one around [00:30] perrito666: ty sir [00:31] it does little to wrap it :) [00:31] yw [00:31] * perrito666 wonders why is he even reading this channel at thos hour [00:32] perrito666: because you love it here - admit it [00:32] :) [00:32] jw4: well have no life :p [00:32] :) [00:32] code is life! [00:32] (of course there are more important things, but coding is a lot of fun :)) [00:32] * jw4 whispers "don't tell katco any different" [00:33] lol [00:33] :) [00:33] well right now my daughter is asleep by 6:30 [00:33] 18:30 [00:33] wow - how old is she? [00:33] i'm sure when that changes i'll be a little more scarce [00:33] she is 8 months [00:33] awww [00:34] does she sleep through the night though? [00:34] most of the time [00:34] nice! [00:34] she's a good little baby :) [00:34] yeah; our first slept through the night at 8 weeks, but our son (second born) took much much longer [00:35] how many do you have john? [00:35] two [00:35] daughter 13 and son 9 [00:35] wow, how nice :) [00:35] the teenage years haha [00:35] don't envy the daipers and sleepless nights even though some monster occasionally replaces my daugter [00:36] well, i'm having fun so far :) [00:36] good [00:36] I get nostalgic until I think about some of the practicalities of infants [00:37] hehe [00:37] perrito666: I forget - you don't have kids right? [00:38] he took himself seriously [00:38] haha [00:41] katco, jw4: we make baaad sleepers [00:41] katco, jw4: my son is 19 months and he's only /just/ started sleeping through [00:42] oh man [00:42] menn0: i feel lucky for sure [00:42] menn0: yeah? That's not unusual I think... but it's a long time to have disruptive sleep [00:43] menn0: for the parents I mean [00:46] jw4: nope, it shows right? [00:47] perrito666: you don't look weary? ;) [00:47] well if when I have kids they are anything like me the will begin to be good sleepers around 21 [00:48] years that is [00:48] perrito666: lol [00:52] menn0: are you close to merging your env worker branch ? [00:53] wallyworld: you are a hard client to please :p [00:53] :-( [00:54] did what i say make sense though? [00:54] wallyworld: not really but I havent read the last review just the email [00:54] the review is just a note or two [00:55] we have existing code to set machine/unit status [00:55] so we need to use that instead of duplicating new code [00:55] lets go priv so logs dont suck later [00:55] we add the agent status setter and the apiserver layer thunks accordingly [00:55] ok [02:11] axw: wallyworld: boom! 2015-01-29 02:10:34 INFO juju.storage.provisioner provisioner.go:128 block devices created: [{1 1771eb70-b0ca-49e3-84d4-7b73b1401d18 /dev/loop1 [02:11] 1 false}] [02:11] katco: sweet :) [02:11] whoot [02:11] going to sanity check the image rq, but we might be good [02:11] seems so [02:12] katco: are you generating a UUID for the volume ID now? [02:12] axw: currently, yes. with todos to use the algorithm you suggested [02:12] axw: i just didn't have a machine id to hang off of [02:12] UUID would probably be fine TBH [02:13] assuming you're creating the image file with that too [02:14] yes [02:14] axw: i think we'll still have to put some thought into how to name these things so persisting across reboots isn't a pain [02:14] axw: but that is trivial i think [02:14] /dev/loop1: [0801]:51515891 (/var/lib/juju/storage/block/loop/var/lib/juju/storage/block/loop/1771eb70-b0ca-49e3-84d4-7b73b1401d18) [02:14] [02:15] cool [02:15] charm is hello worlding along just fine, i'm going to push these changes and have dinner [02:15] i'll check back in after to see if there are any issues [02:18] axw: https://github.com/wallyworld/juju/pull/4 [02:18] looking [02:18] axw: fair warning, the lxc tests are not passing (stupid trivial stuff to fix) [02:18] wallyworld: ^^ [02:19] ok i'm eating some dinner. bbiab [02:21] ty [02:30] mattyw: let me know when the data is loaded and will check it out. It needs rating info because that's the api we hit [02:30] mattyw: so it needs the sub, the plan, and a starting rating data so that the my accout page in the UI had some records to it [02:31] rick_h_, stop [02:31] mattyw: party [02:33] thumper, ping? [02:33] mattyw: hey [02:34] mattyw: wwitzel3 reviewed that branch :) [02:34] thumper, hey there, ok cool - I didn't get around to much reviews yesterday so I'm going to try to get to some today [02:34] thumper, quick question for you [02:34] shoot [02:35] thumper, recently I've been wanting to find the envuuid from juju via the command line - I don't think we have one at the moment, and it feels like it might be something that would be useful to have in juju status [02:35] thumper, wanted to know what thoughts/ plans you had for that kind of thing [02:35] mattyw: there is one [02:35] * thumper bootstraps to check [02:36] $ juju api-info environ-uuid [02:36] 607aa9ab-dd2e-4943-86c3-3fc65b800450 [02:36] mattyw: ^^ [02:36] although, in status? [02:36] maybe [02:37] thumper, awesome! [02:37] thumper, that certainly makes me happy for now [02:37] thumper, I didn't know about the api-info command - feels like Christmas :) [02:37] thumper, thanks for the help :) [02:37] np [02:38] thumper, as along as it's available somewhere I'm happy for now [02:39] kk [02:46] axw: another oarsum hack https://github.com/wallyworld/juju/pull/16 [02:46] wallyworld: looking [02:49] wallyworld: reviewed [02:49] ty [03:01] wallyworld: i've addressed your concerns. ok if i merge? [03:01] katco: would be awesome [03:02] wallyworld: i think you might have to do it since i don't have write access [03:02] katco: oh ffs i keep forgetting [03:02] katco: merged \o/ tyvm [03:03] i'll try it out now [03:03] after meeting [03:03] wallyworld: i would feel a lot better if you or axw could poke at things and make sure this is working (multiple block devices etc.) [03:03] will do [03:03] wallyworld: i'm not familiar enough yet to know what edges there are [03:03] katco: sure, I'll have a play with it [03:03] axw: tyvm [03:04] ok, i'm going to take care of a few things... please ping me if you find something! [03:53] wallyworld: FYI, the loop devices won't get cleaned up if you destroy the machine [03:54] wallyworld: we should be marking these as "persistent" later, and preventing env destruction till they're gone [03:54] axw: i just noticed that [03:54] i ate my disk with a large loop device [04:01] axw,wallyworld: either of you remember where is the cloud-init.log file is on cloud machines? [04:01] thumper: I thought it was just /var/log/cloud-init-output.log ? [04:02] axw: ta [04:02] what he said [04:13] axw: sorta works, but something is not quite wired up - attached hooks complaining storage foo/0 is not provisioned, just digging into that now [04:13] this is for loop on local [04:13] wallyworld: yeah, I'm having some issues too but I've modified the provisioner [04:14] so close [04:33] axw: wallyworld: there r separate discussion going on.. [04:34] axw: wallyworld: is it pool manager responsibility to vaidate provider type supplied [04:34] axw: wallyworld:? [04:34] axw: wallyworld: on pool create that is... [04:35] depends if we think pool manager should know about envieons [04:35] I guess not [04:35] it should validate the storage provider type, but probably not whether it is valid for the current environment [04:35] that is my initial thought, but i'm happy to be wrong [04:36] it does make me a little nervous that something else will come along and use the pool manager without validating, that's all [04:36] may need another component - EnvironPoolManager [04:37] or maybe that's overkill [04:37] that might make it clearer, but you could still go through the PoolManager directly (unless that gets hidden) [04:37] yes [04:37] not too fussed either way [04:38] was just trying to reduce external dependencies of pool manager [04:38] would i think prefer an EnvironPoolManager to live in /environ/storage [04:38] and the api calls that [04:39] I vote we defer this decision, and do validation in the apiserver code for now [04:39] yep, sounds good [04:39] was about to say that :-) [04:40] am annoyed we have so much business logic in apiserver layer - need to fix that sometime [04:44] heh [04:45] wallyworld: I'm sure it is because we don't have a business layer [04:45] we have apiserver and state [04:45] and people didn't want it in state [04:45] so guess where it goes [04:45] thumper: i know too well the reason - been annoyed about it for ages as you know [04:45] yeah.... I know [04:45] * thumper has the munchies [04:49] thumper: if i just install django, is there a default page like for apache or tomcat etc? [04:50] kinda [04:50] you mean locally or the charm? [04:50] wallyworld: sorry, I think your SetProvisionedBlockDevices method is broken. it's only calling state.DemoSetProvisionedBlockDeviceInfo if err != nil above [04:51] thumper: i just want to expose the service and point a browser to it to see it worked [04:51] wallyworld: it might work, but not entirely sure [04:51] kk [04:51] axw: ffs i'm stupid [04:52] wallyworld: I'm changing the storage provisioner to periodically check for unprovisioned storage instances [04:53] wallyworld: not using a watcher because the entities are in the wrong places (not associated with machine, but units) [04:53] can you fix that stuff up while you're there? [04:53] yes [04:53] ta [04:53] just testing now [04:53] dontcha love hacks for demos [04:58] wallyworld: you too huh? :) [04:58] noooooo, we aren't using *any* dirty hacks [04:59] we're far better than that [05:00] wallyworld: of course not. we aren't either. ;-/ [05:00] glad to hear it === kadams54 is now known as kadams54-away [05:38] wallyworld: I think we should be running those LXC changes by someone, could be opening up security holes [05:39] wallyworld: in particular, looks like we're dropping all AA protection [05:39] i think we can tighten up the rules [05:39] after the demo :-) [05:39] of coursed === anthonyf is now known as Guest44978 [08:45] axw: nice branch. we should also be able to register the new providers for ec2 as well as local i would expect? [08:45] wallyworld: yep [08:45] wanna add that real quick? [08:46] wallyworld: sure, just testing a tmpfs provider.. I think it needs more lxc config hacks to work [08:46] likewise with formatting/mounting block devices in local, I think [08:46] ah bollocks [08:47] wallyworld: actually I think I've just mucked up the mount syntax, but could still be broken... will come back to it [08:47] righto [08:50] wallyworld: testing rootfs on ec2, will take a few mins [08:50] sure, np [08:50] my connection has been shit today, so ec2 testing is painful [09:02] wallyworld: updated [09:03] ty looking [09:06] thanks axw, i'll merge [09:06] cool [09:13] wallyworld: I was worried about nothing, tmpfs works on local [09:13] farking oarsum [09:19] wallyworld: https://github.com/wallyworld/juju/pull/19 [09:19] looking [09:20] sigh, ec2 is slooooow right now [09:20] axw:wallyworld: running tests after rebasing, m getting [09:20] cmd/juju/storage/listformatters.go:11:2: cannot find package "github.com/dustin/go-humanize" [09:20] axw: ? [09:20] anastasiamac: godeps -u dependencies.tsv [09:20] axw: i did have a way of registering common providers but removed it due to import loops [09:21] i plan on adding it back later [09:21] wallyworld: cool [09:21] * anastasiamac rolles eyes [09:21] axw: of course :D [09:22] wallyworld: unless you think there's anything else we need for next week, I'm going to take a step back from code and go back to the model rework [09:27] axw: should storage-dir in tmpfs default to /storage or similar? [09:27] libdir? [09:27] the juju lib dir setting [09:27] can't remember the name [09:27] comes from agentconfig [09:28] wallyworld: data-dir? [09:28] wallyworld: as in /var/lib/juju/storage? [09:28] yeah [09:28] isn't that where I put them? [09:28] /var/lib/juju/storage/fs [09:29] ah, haven't seen that in the diff yet [09:30] ah there it is [09:30] wallyworld: I really think we need to change the way that works, since (as I noted in a TODO) the data-dir isn't necessarily the same on all machines [09:30] at the bottom [09:30] it should be decided late [09:30] so it can't be a property of the pool, but of the provider as instantiated in the agent [09:30] yep [09:31] will do for now :-) [09:31] yep [09:32] axw: i'll fix that failing test anastasia mentioned. on my machine the cmd/juju tests *always* hang and time out :-( [09:33] okey dokey [09:33] thanks [09:36] wallyworld: actually before I go off to do model things, one more thing... we should be able to use the storage subordinate now [09:36] axw: only web app i can find that works with postgres is discourse and it fails to install - apt issues, fixed, but now sig verification failure. just seems too out of date [09:37] :( [09:37] oh storage subordinate [09:37] hmmm [09:37] if there were a bundle with that with a web app charm, might be good [09:38] i'd still like to find a web apps just just uses postgres like wordpress uses mysql [09:38] i might see about making mysql chamr use storage [09:39] but first food [09:40] wallyworld: https://jujucharms.com/rails-example-single/6 ? [09:40] I'll test it out.. [09:40] nice, how'd you find that? [09:40] my charm store search foo must be bad [09:40] wallyworld: jujucharms.com, bundles [09:41] looked for one with an elephant :) [09:41] :-) [09:41] right, i need to eat first then charms [10:14] TheMue: https://github.com/voidspace/juju/compare/dimitern:wip-ec2-addressable-containers...wip-preparecontainerinterfaceinfo [10:16] TheMue: there's a comment in there "// TODO: (mfoord) we should add the machine ID to the IP" [10:16] TheMue: that's out of date and can be removed (the call to addr.AllocateTo does this) [10:24] axw: data=rootfs,10M <- forces the use of the 10M or else it errors. we shouldn't need that for filesystemmounts [10:25] wallyworld_: that's not always true. Ceph will need it... [10:25] rightio, so it needs to be optional [10:26] wallyworld_: yep. it should default to 1G anyway, which will then be ignored [10:26] yep [10:33] axw: loop file systems, as accessed from inside container, are owned by root. we will need to make the ownership ubuntu.ubuntu [10:33] wallyworld_: do we? the charms run as root [10:33] unity also notified me of a new device and showed the disk icon in the launcher. and i can see what files i create from inside the container [10:33] nice :) [10:33] ah yeah, fair enough [10:34] wallyworld_: that fails app fails when I try to install it... [10:34] joy [10:34] wallyworld_: apparently you can use django, but I have nfi about it [10:34] me either [10:34] i was hoping django had a welcome page [10:34] like apache, tomcat etc [10:35] but i didn't see one [10:39] axw: i think juju storage list should be sorted by owner. thoughts? i had a few units and would prefer to see storages for the units listed together [10:39] anyone able to look at this one line fix: http://reviews.vapour.ws/r/820/ [10:39] wallyworld_: yeah, probably [10:39] wallyworld_: in the tabular format anyway [10:39] menn0: oh, you ask all the hard questions [10:40] axw: yep [10:40] wallyworld_: sorry :) [10:40] wallyworld_, axw: thanks guys [10:42] anastasiamac: can you please sort the tabular storage list by owner? when you are free [10:52] morning [11:21] perrito666: morning [11:26] axw: this fixes some tests (yours and mine :-) and also makes default size 1G and default count 1 always https://github.com/wallyworld/juju/pull/20 [11:27] wallyworld__, axw: pushing a branch now that fixes the panic that's currently being frequently seen during merge attempts [11:28] which test? [11:29] wallyworld__: the leadership feature tests are failing frequently [11:30] the problem isn't actually with those tests but with my recent change to how the machine agent workers are started [11:30] subtle race [11:30] ok, i hadn't noticed as ee've been landing to a feature branch [11:30] great that you caught it [11:30] wallyworld__: http://reviews.vapour.ws/r/821/ [11:31] wallyworld__: funnily enough, this related to a review comment you made :) [11:31] oh, i hate to ask, which one? [11:32] wallyworld__: where we were discussing whether the envWorkerManager should clean up the returned runner even when an error was returned [11:32] ah right [11:32] wallyworld__: I was using an unusual pattern, which you picked up on [11:32] ancient history, > 1 week ago :-) [11:32] wallyworld__: this fix in the same area [11:34] menn0: done, ty for fix [11:35] wallyworld__: thanks. good idea re the comment [11:35] menn0: i can't recall, maybe even a comment next to where singular runner is used if it's releavnt to do so [11:36] wallyworld__: I actually have a better idea which is clearer and reduces the assumptions about the singular runner's behaviour [11:36] ok [11:43] wallyworld__: pls take a look again. this is much clearer IMO. http://reviews.vapour.ws/r/821/diff/ [11:43] ok [11:44] menn0: can we remove "&& singularRunner != nil" now? [11:45] looks much nicer [11:45] wallyworld__: probably but I've been bitten by not have a guard like that before [11:45] wallyworld__: let me try it [11:45] menn0: i can't see that it's needed as we exit above with an err before the defer [11:46] wallyworld__: yeah I know, but I've thought that before and had problems [11:46] wallyworld__: that was in a slightly different case though [11:46] menn0: runner and singular runner usage looks the same and runner has no guard [11:47] oh wait [11:47] sorry [11:47] can't read [11:47] wallyworld__: it does for the defer that cleans [11:47] LGTM [11:47] wallyworld__: i'm going to remove them and re-run the tests and my little tester that was triggering the panics [11:47] wallyworld__: the guards irk me too [11:47] ok [11:49] wallyworld__: yep, doesn't work. panics due to nil pointer dereference [11:49] doh [11:49] wallyworld__: I don't get why... something special about defers perhaps and how they get executed? [11:49] nfi right now [11:50] wallyworld__: I don't care right now. bed awaits and I still have some other (non-work) jobs to do [11:50] yep, land tha sucker [12:15] wallyworld__: that's in. thanks for the late night review. [12:15] np, catch you later [12:15] really late for you [12:15] yeah :) [12:18] wallyworld__: anything to go over before I see you in Capetown ? [12:19] fwereade: ^^ [12:19] jam: nothing urgent - just need go over storage, health etc prior to seeing mark [12:20] jam: there's a few things re: storage we can disuss over there === kadams54-away is now known as kadams54 [13:23] wallyworld__: axw: hey, just waking up. how are things with storage? [13:23] katco: good :-) [13:23] wallyworld__: woohoo! [13:23] loop provider works [13:23] was tweaked a bit [13:23] wallyworld__: cool [13:24] axw added tmpfs and rootfs [13:24] wallyworld__: anything i should know going into today? [13:24] very cool [13:25] katco: we do need to 1. evaluate the permissions, especially the aa ones on the container (but that can wait); 2. loop at mounting an external block device [13:25] look not loop [13:26] and bind mounting into local provider [13:26] wallyworld__: understood on both accounts. 1 is part of a lot of cleanup we'll have to do [13:26] yep [13:26] wallyworld__: axw: i was planning on looking at 2 today [13:26] doesn't matter for demo of course [13:26] great :-) [13:26] cool [13:27] wallyworld__: axw: ok i'm off to get ready. you guys rock! [13:27] anastasiamac: and you as well if you're around [13:27] katco: later, hope you have a good day [13:28] axw: not sure if you saw before, i fixed some tests and added default size handling https://github.com/wallyworld/juju/pull/20 [13:28] wallyworld__: I didn't. looking now [13:28] i made count default to 1 regardless of size [13:28] if count wasn't specified [13:37] wallyworld__: reviewed [13:37] ty [13:44] axw: changes pushed [13:45] wallyworld__: LGTM, thanks [13:45] merged [13:48] katco: also, if you're bored, there's also investigating how to provision block devices on openstack (as we do with EBS volumes on AWS) [14:04] jam, hey, as OCR can you have a look at this please? http://reviews.vapour.ws/r/822/ [14:16] wallyworld__, axw, ^^ [14:16] well not as OCRs but if you have some time :) === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [15:08] wallyworld__, axw you guys must be robots that don't need sleep [15:17] dimitern: Could you review http://reviews.vapour.ws/r/823/ please? [15:17] jamestunnicliffe, sure, looking [15:18] ....snow === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [16:05] perrito666: are we having standup? (no natefinch, no wwitzel3) [16:05] ericsnow: we can wave it [16:05] ericsnow: yet Ill priv you a question [16:05] perrito666: k [16:10] voidspace: why do you think so? [16:10] TheMue: no reason... [16:16] dimitern: http://reviews.vapour.ws/r/824/ should work. Deleted the old merge and review request [16:25] jamestunnicliffe, reviewed === kadams54 is now known as kadams54-away [17:14] mgz, ping [17:15] dimitern: hey [17:16] mgz, hey! so a quick question about the merge bot [17:16] hit me [17:16] mgz, jamestunnicliffe - our new team mate wants to land a PR - https://github.com/juju/juju/pull/1502 - and he is in juju Hackers group on GH, has access to RB and all [17:16] mgz, but it seems the bot is not picking up his $$merge$$ comment [17:17] jamestunnicliffe: have you set your membership of the juju group to public? [17:17] mgz, ah! that might be it :) [17:17] mgz, he went for some tea - will ask him in a moment :) [17:18]