[01:02] anastasiamac: could you look at https://github.com/juju/juju/pull/3961 (it's not getting into reviewboard for some reason) [01:09] anastasiamac, thanks! [01:10] mwhudson: of course \o/ [01:13] anastasiamac: thanks [01:13] mwhudson: lgtm :D [01:16] anastasiamac: thanks. how do i merge it? :-) [01:17] mwhudson: i believe that there is a bot on this repo... have u tried $$merge$$? [01:18] no, let's find out [01:18] k [02:05] hey good morning [02:05] perrito666: hi o/ [02:06] how is sunshine coast? [02:06] nice and sunny [02:06] aeesome \o/ [02:06] i hope hot too - m removing all long-sleeve stuff from suitecase :D [02:24] cherylj: i think it's a one line, fix, http://reviews.vapour.ws/r/3386/ [02:27] wallyworld: lgtm. Let's hope that's it :) [02:27] cherylj: i think it is jusdging by the error [02:30] cherylj: the cutover for alpha2, that's in a few hours? [02:30] yeah, no. I don't think we're going to get an alpha1 out this week. [02:31] cherylj: works well for me then - i have a couple of series in metadata fixes i want to do later today [02:31] heh, sounds good :) [02:32] about about to leave for sprint, will do them later tonight [02:32] have a safe trip, wallyworld :) [02:32] cherylj: will do, it's only a 2 hours drive :-) [02:32] not a 16 hour flight :-) [02:32] ha, nice [02:43] cherylj: and i will b driving - so wallyworld will be safe \o/ [02:43] :D [09:58] frobware, hey, sorry I missed our 1:1 [09:59] dimitern, np [10:52] dimitern: Error handling subnets "Default space" is not a valid tag [10:52] dimitern: that's great!! [10:52] voidspace, you should upgrade your maas :) [10:52] dimitern: lots of problems with rc3 [10:53] staying with rc2 was deliberate [10:53] dimitern: think it's worth upgrading? [10:53] dimitern: frobware has had various issues with the upgrade (node addresses and leases) [10:53] voidspace, since rc2 or thereabouts spaces are now called "space-0" and fabrics like "fabric-0" [10:53] dimitern: anyway, I'll just add name translation [10:53] dimitern: facade registration worked [10:53] voidspace, is "Default space" the name or the description? [10:54] dimitern: name [10:55] voidspace, I'd first try renaming that to "space-0" [10:55] voidspace, it used to be forbidden to change the default space's name [10:56] voidspace, if you can change it with $ maas space update 0 name=space-0 [10:56] dimitern: yeah, I did that - now I get "foozle" is not a valid tag, so there's obviously more I need to do in my code [10:56] ("foozle" was the random name I picked) [10:57] ah, maybe I'm using space name on subnets when I should be using tags [10:57] voidspace, yeah - that was what we talked about yesterday I think - space tags are "space-" and should be used instead of space name in api params (except for creating a new one I guess - except to check for dupes) [10:57] dimitern, can we sync? [10:58] frobware, sure - standup HO? [10:58] yep [12:06] Bug #1526296 opened: deploy charm with --series --force doesn't reject invalid OS [12:09] Bug #1526296 changed: deploy charm with --series --force doesn't reject invalid OS [12:18] Bug #1526296 opened: deploy charm with --series --force doesn't reject invalid OS [13:59] cherylj: Good morning. Bug #1526063 has only been observed in the maas-spaces branch. I meant to mark it incomplete for master, since we don't have evidence that it's there. [13:59] Bug #1526063: TestNoContextWithLock

[14:00] cherylj: But if we do mark it high in master, we have been assigning it to a milestone also. [14:36] dimitern: what's the correct way to create a tag - first check it's valid and then call New*Tag ? [14:37] voidspace, it depends on what New*Tag() does, but most of them IIRC call IsValid*Tag() and panic if it's not [14:37] dimitern: right, that's why to check if valid first [14:37] voidspace, so it's better to check before calling it [14:37] dimitern: if *not* that, then what? [14:37] so that is the correct way [14:37] thanks [14:37] voidspace, where are you getting the name from? [14:38] dimitern: subnet.CIDR and space.ProviderId [14:39] (I've got past the tag probl [14:39] oops [14:39] abentley: I marked it as triaged for both, as both seem to have the same code [14:40] voidspace, so ProviderId can be anything, so but you're not validating it anyway, right? [14:40] dimitern: I'm doing a simple transform first in my current code [14:40] voidspace, as for the CIDR, if it's invalid and you want to convert it to a subnet tag, then it's an error [14:40] sure [14:40] it's the actual mechanics of creating the tag I was asking about [14:42] currently I have a panic on the apiserver when trying to list spaces - nil pointer dereference [14:42] voidspace, hmm does it say where? [14:42] calling space.ProviderId() [14:43] in reflect/value.go from ProviderId() [14:43] which is new [14:43] interesting [14:44] if a field is "omitempty" in a doc then will accessing it get you a nil pointer dereference? [14:44] I just assumed that for a string, omitempty would default to an empty string [14:45] Bug #1517258 changed: juju 1.24.7 precise: container failed to start and was destroyed

[14:51] dimitern: ah, spaceShim doesn't have a ProviderId [14:51] I think that's it [14:51] Bug #1517258 opened: juju 1.24.7 precise: container failed to start and was destroyed

[14:52] that makes more sense [14:54] voidspace, right :) missed that one [15:00] Bug #1517258 changed: juju 1.24.7 precise: container failed to start and was destroyed

[15:02] dooferlad, you've got a review [15:03] dimitern: thanks! [15:03] Bug #1517258 opened: juju 1.24.7 precise: container failed to start and was destroyed

[15:06] Bug #1517258 changed: juju 1.24.7 precise: container failed to start and was destroyed

[15:14] dimitern: you didn't miss it, ProviderId is new [15:15] dimitern: I didn't realise it would need to be added to spaceShim as well [15:15] re-bootstrapping now [15:15] dimitern, you about for 10-15 mins after the OS call? [15:16] voidspace, heh I meant you added that recently to AddSpace et al, right? :) [15:17] frobware, sounds good [15:18] dimitern, as a heads-up just trying to add to the release notes for the AWS support post 15.10 [15:21] frobware, ah, nice [15:22] frobware, I though I already added something about it in Juju 1.25.0 Release Notes doc [15:23] dimitern, it's the stuff we've done since then, mostly bug fixes IIRC [15:23] frobware, +1 [15:43] dimitern: space imported on bootstrap! [15:43] voidspace, awesome!! \o/ [15:44] voidspace, so as soon as my other branch lands, they should be usable in constrains right after bootstrap [15:44] dimitern: hmm... but not the subnets [15:44] dammit [15:44] ok, next step of debugging [15:44] voidspace, ah :) [15:58] dimitern: replied to review - just a query about your comment about the comment... [16:03] hey dimitern, are you still working on bug 1525280? [16:03] Bug #1525280: 1.25.1 with maas 1.8: devices dns allocation uses non-unique hostname

[16:03] cherylj: did you already bump the min version blueprint to a2? === ddellav_ is now known as ddellav [16:04] katco: yes, let me re-generate the tracking page [16:05] cherylj: ty for doing that [16:08] dooferlad, cheers - replied [16:09] cherylj, hey, I'm still sorting out a few issues and testing but it will be ready today [16:09] awesome, thanks dimitern [16:29] dimitern, if you would like a tester for your change I can help [16:30] frobware, I'll push the changes during the OS call [16:30] frobware, it will be good to get a second verification [16:33] voidspace, openstack call? [16:35] dimitern: omw [16:55] katco: you said no retrospective this friday, right/ [16:55] natefinch: yeah i'm going to cancel [16:55] katco: ok, cool, just making sure. [16:55] natefinch: you still need to show up for work though :) [16:56] katco: awww man [17:02] dimitern: hmm... just had a thought, I bet adding a subnet with a spaceTag isn't sufficient to associate it with that space is it [17:02] I bet I need to do more... [17:02] hmmm... no, that's not the root problem - subnet list is also empty even though it found the subnets from the provider and called AddSubnet [17:03] I wasn't checking the result error field in the first cut though - I've added that now [17:03] so at least I should see an error if one occurs [17:07] dimitern: error creating subnet SubnetTag and SubnetProviderId cannot be both set [17:07] dimitern: so that's straightforward [17:08] voidspace, ah :) [17:08] voidspace, yeah - use the provider id [17:10] dimitern: ok [17:10] dimitern: which means we don't use the CIDR at all creating the subnet [17:25] voidspace, we do on AWS [17:25] as an alternative to using provider id [17:26] voidspace, the idea is, if you have provider-id, you can uniquely identify the subnet, whereas with just a CIDR it might be ambiguous (depending on how it's specified) [17:30] dimitern: next issue is in the maas provider itself [17:30] dimitern: exited "discoverspaces": error creating subnet cannot get provider subnets: instance "" not found [17:30] dimitern: that's new code, so obviously needs some work [17:30] (the new subnets code) [17:30] on the apiserver side [17:30] it's an interesting journey :-) [17:44] voidspace, dooferlad, frobware, please have a look http://reviews.vapour.ws/r/3388/ - fixed the critical bug 1525280 [17:44] Bug #1525280: 1.25.1 with maas 1.8: devices dns allocation uses non-unique hostname

[17:45] voidspace, I've fixed that in my other branch btw [17:49] dimitern: this is address allocation only right? [17:50] voidspace, have a look please - one more thing to do and will set it to land: https://github.com/juju/juju/pull/3921 [17:59] mgz, nope [17:59] mgz, it's the default behavior now [18:00] mgz, but I tried not to break the addressable containers as well [18:01] dimitern: sorry, should have been clearer, this is only an issue when we have explict address allocation on, as in 1.25 or 1.24 with flah? [18:02] mgz, ah, well.. that's a good question [18:03] mgz, I think it's fine with the flag still [18:03] mgz, but just doing a quick test now on 1.25+1.9rc4+address-allocation [18:43] dimitern: EOD, but I can look later tonight [18:44] voidspace, cheers, if you can I'll appreciate it [18:45] dimitern: yep [18:45] dimitern: ah, I thought empty instance id already worked! [18:45] dimitern: I thought I'd done that, my fault then :-) [18:50] dimitern, should I expect a different name. I see: instance-id: juju-machine-0-lxc-0 [18:53] frobware, in status it will be the same, but if you check the node details on the web UI, you should see a section about containers and VMs and it will be there with the unique hostname [18:54] natefinch: wwitzel3: have you had a chance to review ericsnow 's patch? http://reviews.vapour.ws/r/3375/ [18:54] katco: almost done with mine [18:55] mgz, it's good that I checked, as it turned out I did break the addressable containers unintentionally; fortunately it's quite easy to fix [19:05] dimitern, mgz: we need (developer-centric) automation in this area. Too much manual testing... :( [19:05] natefinch: also, is your patch ready to be reviewed? [19:05] natefinch: or still a wip? [19:05] dimitern, in the new year we should have a spike on automation [19:05] katco: it is ready to be reviewed [19:06] natefinch: sweet tal [19:07] dimitern, its says "fixed" for me. fixed-kitten-juju-machine-0-lxc-0.maas :) [19:09] frobware, awesome, isn't it? :) [19:09] dimitern, want me to sanity test with multiple nics or shouldn't it matter? [19:10] frobware, it might matter, the more tests the better - but don't feel obliged [19:20] dimitern, looking again at your revised patch [19:24] frobware, cheers - pure luck I got that on the first attempt (a second test triggered it after a few attempts) [19:27] natefinch: review up. should have some things to address [19:27] katco: cool, thanks. [19:28] dimitern, how come allocate-address worked without requiring a pointer before? [19:30] frobware, :) long story - it basically expected to get a concrete address to try allocating [19:30] frobware, so it didn't need to return anything but an error [19:30] yuck [19:35] dimitern, I could do this in the review but figured we might go quicker here: splitting host on '.' what happens for some-host.qacdo.maas? [19:36] frobware, it will work as funny-joke-juju-machine-1-lxc-1.qacdo.maas" [19:36] frobware, hmm.. but I think I check if the parts are exactly 2 - good point! [19:36] dimitern, right [19:37] dimitern, line 1405 I think is the issue, not necessarily the split [19:38] frobware, yep, good catch [19:51] dimitern, Is vs IS - is the plan to fix them up as we find them? [19:51] dimitern, correction: Id vs ID [19:57] frobware, that's a can of worms right there :) the consensus seems to point towards ID recently though [19:57] dimitern, fwiw - for me it would be better to do this in a different PR - less to review [19:58] frobware, agreed - these are just a few drive-by fixes [20:03] ericsnow: looking at your patch now [20:03] katco: ^ [20:10] dooferlad, fwiw the merge bot doesn't take kindly $$c o m m e n t s$$ with spaces apparently [20:18] dimitern, why did so many of the IP addrs get bumped? 2.1 => 2.2, for example. [20:20] katco, ericsnow, wwitzel3: Forgot to mention this mornnig - I'll be in an out tomorrow. My wife had a dentist appointment rescheduled which is screwing up a good part of the afternoon. [20:21] frobware, because tests were simpler before and didn't care about the default gateway, and assumed all networks of a node have .1 IPs [20:21] katco, ericsnow, wwitzel3: should be good for the morning, though. [20:21] dimitern, ha. makes sense. thx. [20:21] dimitern, subtle though. [20:22] frobware, yeah :) [20:23] gotta take off a little early, but will be back later tonight === natefinch is now known as natefinch-afk [20:24] and the maas-spaces live test with even more complex constraints works! [20:32] dimitern, great! [22:12] katco, natefinch-afk, wwitzel3: could you also take a look at https://github.com/juju/charm/pull/184 [22:15] ericsnow: sure [22:22] katco, natefinch-afk: I've addressed all review comments except for the one about the CharmStore interface [22:22] katco: regarding that one, did you have any alternatives to recommend? [22:29] ericsnow: i'd have to look at the interface again [22:29] ericsnow: one sec [22:32] ericsnow: maybe like ResourceLister? [22:33] ericsnow: if we stick with nouns, i agree CharmStore is fine [22:33] ericsnow: and also it's not really a huge deal :) [22:34] katco: I was going to say that we are going to add a bunch of other methods, but actually that's not true [22:34] katco: so I'm fine with doing the normal Go interface name thing [22:34] ericsnow: well and if we do, i would suggest we pass in multiple interfaces that are grouped by functionality [23:29] cherylj: that fix did the trick for 1.25 \o/ so I assume you're looking to release real soon now? [23:33] wallyworld: for 1.25.2, yes. We're just waiting on dimitern's fix for bug 1525280 [23:33] Bug #1525280: 1.25.1 with maas 1.8: devices dns allocation uses non-unique hostname

[23:34] awesome, will be good to see that fixed [23:40] wallyworld, thanks for the quick fix on the windows issue yesterday [23:40] sure, np [23:40] wallyworld, how is the beach? :) [23:40] terrible :-( [23:40] Yeah - surf is a bit small at the moment [23:40] are you trying to make me feel better [23:40] i have a horrbile photo i'm going to share with you soon [23:41] punk [23:41] I could barely catch a wave at all this morning [23:41] blahdeblah: where are you based? [23:41] wallyworld: Sunshine Coast [23:41] blahdeblah: ah, we're sprinting at moolooabah [23:41] nice [23:42] we're having a chrosmas party on the weekend [23:42] want to come? [23:42] When is it? [23:42] sat night? [23:42] haven't locked it in yet [23:42] Will have to check with the social director, but sounds good. [23:42] ok, let me know asap if you can as places along the esplanade fill up real quick atm [23:43] will do [23:43] and yes, waves are a bit small :-) [23:44] mooloolaba will always be small, but even the open beaches are small this week [23:44] indeed [23:44] Not that it will stop me having a go. ;-) [23:44] me either