[00:36] menn0: your welcome [00:36] * davecheney swallows guilt about not knowing enough about what he was reviewing [00:52] menn0: https://github.com/juju/juju/pull/743 [00:52] if you're feeling like it [01:58] axw: running late for 1:1, otp [01:58] wallyworld: nps, ping when ready [02:35] * wallyworld should not leave my keyboard unlocked when I am getting lunch [02:51] axw: sorry about delay, is now ok? [02:51] wallyworld: sure, be there in a minute [03:30] davecheney: sorry about the delay. looking at that PR now. [03:30] it's been a bit of circus at my place today [03:34] no probs [03:43] davecheney: done [03:55] is anyone able to have a look at https://github.com/juju/juju/pull/732/ ? === urulama-afk is now known as urulama === BradCrittenden is now known as bac [06:14] TheMue: hey Frank, when you come on, could you take a look at https://github.com/juju/juju/pull/737 It introduces a new facade, and copies a bunch of code off Client for the EnsureAvailability() API. This will allow a newer version of EnsureAvailability() to be implemented [06:15] thanks in advance :-) [06:42] morning all === b is now known as Guest39961 [08:03] morning [08:03] wallyworld: yep, will take a look in a moment [08:14] morning all [09:00] folks - what's our general guideline for pr size? [09:00] mattyw: asap -> as small as possible ;) [09:01] mattyw: mine grow always too much, that's not good. hard to review and sometimes trouble to merge. [09:02] TheMue, I have a 500 line one and a 100 line one - thinking about merging them [09:04] mattyw: I think LOC aren't a good metric here. if it helps the reviews to see them both together a merging IMHO makes sense. [09:10] wallyworld: you've got a review [09:51] morning [10:16] perrito666: morning [10:19] perrito666: o/ [10:47] dimitern, voidspace: ho? [10:47] TheMue: omw [10:48] TheMue, me too, sorry [13:34] * perrito666 looks at an empty space where my calendar says natefinch should be [13:34] :p [13:34] perrito666: heh [13:35] I can reschedule, just tell me so I go back to headbanging with nirvana [13:35] perrito666: I'm stuck with a kid right now, so yeah, later would be good [13:36] ok, nirvana it is [13:49] Hi devs, I want you all to know that while master and 1.20 are reported as failures for the last 2 days, we don't think juju is at fault. CI may not either. utopic lxc fails and the reason might be because utopic, or that ci tests need special rules for ubuntu devel series that have lots of package changes [13:56] sinzui: tx [13:56] has anyone hit this with local? [13:56] 'error executing "lxc-create": lxc_container: No such file or [13:56] directory - bad template: ubuntu-cloud; lxc_container: bad template: ubuntu-cloud; [13:56] lxc_container: Error creating container juju-precise-lxc-template' [13:57] perrito666, no, but that looks like /var/cache/lxc has bad data? have you tried to delete it [13:58] sinzui: no I havent, tx for the advice [13:58] perrito666, when that is stale. or you get a download error, you will get messages like that [13:58] mm, seems empty [13:58] perrito666, stale images can report that when a crucial package has changed like openssl [13:59] perrito666, that is interesting. lxc will populate the cache before it creates. [14:00] perrito666, we might have a name mismatch... [14:01] perrito666, [14:01] $ sudo ls /var/cache/lxc [14:01] cloud-precise cloud-trusty trusty [14:01] mmm, why in the universe whould this happen [14:02] the ubuntu-cloud template nows how to find those series. it is provided with lxc, but might really by in the lxc-templates package [14:06] sinzui: odd, let me recompile this just in case [14:12] I'm looking at PR746 from TheMue and I don't really understand how that API versioning fits in to the juju run. Since juju run doesn't actually use Facades at all. [14:13] natefinch: standup? [14:21] wwitzel3: will tell you later, mom [14:22] TheMue: sounds good, I sent a mail to juju-dev so others can benefit from it too [14:23] TheMue: since when I asked about versioning most of the responses I got were, "no idea" [14:23] TheMue: thanks :) [14:37] wwitzel3: just writing a doc about it [14:37] wwitzel3: has been a discussion between William, John and me [14:44] does anyone know if I can trigger lxc template download? [14:44] by hand [14:44] ? [14:49] can someone have a look at: https://github.com/juju/utils/pull/27 . This is needed to fix some tests on Windows [15:08] sinzui: do you know the answer to my question? [15:09] perrito666, I do know you can do it but I don't know the command. I think it is ubuntu-cloud-somthing [15:10] perrito666, I can give you an lxc example to create a container that will trigger the series you want to try [15:11] sudo lxc-create -t ubuntu -n trusty -- -r trusty -b $USER [15:11] ^ perrito666 [15:16] perrito666, sudo lxc-create -t ubuntu -n precise -- -r precise -b $USER [15:17] will match the cache the error is about [15:38] wwitzel3: so, now I've got some time for you [15:39] wwitzel3: you're talking about version of a command, am I right? [15:39] wwitzel3: this sounds strange to me, so far we only talked about versioning for the API [15:40] wwitzel3: so in case the change of the command also needs a change of an API then it's ok [15:41] TheMue: well, that is what was suggested on the reivew, yes. [15:42] wwitzel3: ah, ic, so you're having an existing facade and you now have to change it (new method or change params or result of a method)? [15:42] sinzui: thank you very much man [15:42] TheMue: No, there is no existing facade [15:45] wwitzel3: but to realize your command you would have to add a new facade because action happens on the server and no current facade matches to it? [15:48] TheMue: hrmm, there was no facade changed or registered for this at all .. https://github.com/juju/juju/pull/705 [15:50] *click* [15:50] TheMue: instead of passing in a string, I pass in a struct .. but I don't see how versioning is going to do anything here .. if you have a new client and you run juju run --help .. you will see the new options [15:51] TheMue: and if you have the new client and provide those options, the old server will just ignore them [15:52] wwitzel3: I see it, it's apiserver/client. [15:52] TheMue: the problem is we do some validation before it ever gets to the server, for valid names [15:52] well look at that, https://bugs.launchpad.net/juju-core/+bug/1330406 there was a bug for this [15:52] Bug #1330406: juju deployed services to lxc containers error executing "lxc-create" with bad template: ubuntu-cloud [15:52] wwitzel3: now lets say the apiserver stays at the current version but you install a new client, how will they work together? [15:53] wwitzel3: the API on the server expects different params than the client sends [15:53] wwitzel3: here we have to support both, the old way as well as the new way on the server [15:54] wwitzel3: the client chooses the best version matching to its own version and uses it [15:54] TheMue: if we have a new client, and an old server .. it will work .. the server will ignore the new options [15:55] abentley: thanks for all your help [15:55] TheMue: it is not changing how the existing options work [15:55] abentley: will you need more info for the SSL stuff than what I've sent you? [15:56] ericsnow: No, that should be good. OTP. [15:56] abentley: no worries :) [15:56] TheMue: if there is an old client and new server, it will also work. The new server knows how to parse all the options from the old client [15:59] TheMue: I'm not against it, if we need to do it, I just fail to see how it is going to improve the user experience .. since a new client, with old server will still result in the user seeing options that aren't available on the server. [15:59] TheMue: unless we are going to version it so that we report an error to the user when they use those options [16:00] TheMue: but I thought the point was so that automated scripts and such don't break [16:00] TheMue: which in this case, they wouldn't [16:08] can lunch be a verb? .. [16:08] * wwitzel3 lunches === jheroux_away is now known as jheroux [16:13] there you go, I just needed to purge lxc and re install it === allomov__ is now known as allomov [16:36] wwitzel3: yes, I've seen how the client calls a number of individual server functions w/o changed args. had seen it wrong first === jog_ is now known as jog [16:52] night all, have a good weekend [16:52] mattyw: tc [17:00] ericsnow, I am running a bit late [17:00] will ping you [17:00] alexisb: k [17:09] ok ericsnow ready and joining the hangout === jheroux is now known as jheroux_away [18:17] cmars: could you take a look at https://github.com/juju/juju/pull/736? [18:25] cmars: as well as https://github.com/juju/utils/pull/33 [19:48] ericsnow, ok [19:48] cmars: thanks! [19:49] ahh my god. I hate it when I realize I've wasted hours on a stupid logic error [19:49] cmars: I just put up a couple really small ones too (749 and 750) [19:52] abentley: sorry to bug you but what's the status on copying those SSL files over? [19:53] ericsnow: Sorry for the delay. Busy, busy day. I'll do that next after this code review. [19:54] abentley: much appreciated! sorry to add to your already overflowing plate :( [19:54] anyone know anything about aufs? [19:54] natefinch: if he logic error wasn't stupid you wouldnot hate it? [19:55] perrito666: it would be less embarassing to myself if it were something complicated. [19:55] natefinch: you had no reason to feel embarrassed until you told us about your mistake [19:55] .p [19:55] :p [19:55] heh [19:56] nah.... I guess embarassed is the wrong word, frustrated with myself [20:16] export FOO=bar doesn't affect the external terminal when I put it in a .sh file and run the file.... why? And what's the right way to do that? I'm trying to write a script that'll set up a debugging environment... but the export line doesn't seem to work [20:17] ericsnow: Try https://reviews.vapour.ws/r/ [20:18] abentley: works great. Thanks! [20:18] Your connection is not private. Attackers might be trying to steal your information from reviews.vapour.ws (for example, passwords, messages, or credit cards). [20:18] that's from Chrome.... maybe because it's a self-signed cert? [20:19] natefinch: indeed, that is why. [20:19] abentley: ahh, yeah, if I click on the "advanced" link, it says that [20:19] hrm [20:24] natefinch: with that we are all set to cut over tomorrow night :) [20:24] ericsnow: cool [20:25] abentley, ericsnow: how hard would it be to get a real cert for that page? a huge warning from the browser is not exactly the most professional foot to put forward [20:27] natefinch: Well, we could ask sabdfl if he's got a copy of the Thawte root cert lying around in his sock drawer. Then it would be free :-) [20:27] abentley: lol [20:27] natefinch: It's just a matter of buying it. We have a CSR ready to submit. The domain is owned by Curtis, so I'd certainly let him make the call first. :) [20:27] abentley: nice one :) [20:28] ericsnow: I figured. Let's do that ASAP... it's like $50 or something for a cert, it seems silly not to get one... but also doesn't seem like something we should gate on. [20:28] natefinch: agreed [20:28] natefinch: I'll add it to the to-do list :) [20:30] ericsnow: just warn people to expect that error message when you send the cut-over email [20:30] natefinch: sounds good === Beret- is now known as Beret [21:49] could anyone spare me a minute for a 3-line patch? https://github.com/juju/juju/pull/749 === HankM00dy is now known as thehe === mup is now known as 16WAAEA3S === makyo_ is now known as Makyo === rodlogic is now known as Guest11715 === benji is now known as Guest28630 === bodie_ is now known as Guest70589