[00:25] wallyworld: The subnets I add when the provider doesn't support spaces should have a space of "", right? [00:26] yup [00:26] "" is considered the default space [00:27] wallyworld: at the moment the AddSubnets api method doesn't accept "", I'll change it to now. [00:27] ah, ok [00:30] wallyworld: The AddSubnets method is shared between the Subnets and DiscoverSpaces facades - do you think I should allow space="" in both places? [00:31] babbageclunk: yeah because "" is universally the default space [00:31] * wallyworld says without looking at code [00:31] wallyworld: yay, that's easier! [00:49] thumper: i fixed the race https://github.com/juju/juju/pull/7056 [00:49] wallyworld: awesome [00:49] race was one line, drive bys a few more :-) [00:50] wallyworld: lgtm [00:50] ty [00:55] I wish we had someone around that understood the 1.25 networking model and the changes for 2.0 [00:55] * thumper sighs [00:56] blahdeblah: hey... [00:56] blahdeblah: didn't you say that you had some 1.25 environments and 2.0 models with similar capabilities? [00:56] blahdeblah: are they maas or openstack? [00:57] thumper: yes; openstack - all in canonistacks [00:57] well... that is a start at least [00:57] thumper: they're not identical, but similar [00:57] blahdeblah: what's the chance you can get me a mongo dump of both? [00:57] sure. [00:57] thumper: Did you get my email about that yesterday? [00:58] probably... [00:58] * thumper looks [00:59] blahdeblah: well that sucks [00:59] sorry :-) [01:02] blahdeblah: how conversant are you with the mongo shell? [01:02] thumper: I can share my screen with you via hangout and monkey-type whatever you tell me like an absolute pro! :-P [01:03] :) [01:04] blahdeblah: what we could do is add the index on the fly to test [01:04] and if that works, which it should, get it in to 2.1.1 [01:06] thumper: while you're in there maybe check how many rows there are in the collection? [01:06] thumper: sure - just need to step away for a minute, but we can work on it after that [01:06] babbageclunk: yeah [01:06] blahdeblah: sure, ping when back [01:06] will do [01:09] thumper: There's a comment in MachineAgent.startModelWorkers saying "we should try to find the right values for these numbers". [01:09] haha [01:30] * thumper goes to make coffee [01:35] thumper: I'm back - can look at this index when you're ready [01:52] blahdeblah: ok, just eating a muffin [01:53] thumper: you want to try hangouts for this again? [01:58] blahdeblah: yeah [01:59] blahdeblah: just on another call, with you shortly [02:19] thumper: I'm gonna have to run for lunch appointment in 10 mins, and I'm guessing you'll be EOD-ish by the time I get back. Can you pastebin or email what you want me to do? [02:22] blahdeblah: yeah... can do [02:28] * babbageclunk goes for a run [03:33] thumper: still around [03:33] ? [03:40] thumper: just want to confirm that I'm doing this indexing stuff on 2.1.0 [03:40] (not 1.25.10) [03:53] blahdeblah: yeah 2.1 [03:54] thumper: Just created a ticket to track and added your email; the index thing worked perfectly and I'm moving on to getting the dumps [03:54] blahdeblah: which index, first or second? [03:54] first [03:54] https://pastebin.canonical.com/181228/ [03:56] blahdeblah: sweet [03:56] I'm a little surprised that the first one worked, but kinda happy [03:56] babbageclunk: ^^ index without _id worked [03:59] blahdeblah: that is very helpful, would still like a mongo dump of 1.25 if you can :) [03:59] On its way [03:59] blahdeblah: awesome, thanks [03:59] Now I'm on the hunt for some maas 1.25 environments to get a dump from [04:00] and some maas 2.1 models to do equivalent queries against [04:00] bradm: ^ [04:08] thumper: oh, nice! [04:10] thumper: well, I guess I should have had more faith in mongo [04:10] babbageclunk: https://github.com/juju/juju/pull/7058 [04:10] babbageclunk: not sure why you should have more faith, I didn't [04:10] well, I had ~25% faith [04:10] does that count? [04:13] thumper: lgtm'd [04:19] babbageclunk: https://github.com/juju/juju/pull/7059 another [04:20] babbageclunk: indices are ensured whenever state gets opened [04:24] thumper: This is the one I thought the other one was going to be - you got me with a classic bait-and-switch! [04:24] thumper: lgtmd [04:27] axw: added revert to the same PR. still +1? happy to merge? [04:28] anastasiamac: yes, thanks [04:28] axw: awesome! shiping.. [04:56] jam: will be 5-10 mins late, I'll ping you. I'd like to chat about storage spec revision today [04:56] k [05:07] jam: I'm in [08:24] axw: do we have a bug about "not able to open /dev/loop" spamming the log files with the lxd provider? [08:25] (it might be fixed in 2.2, I'm just seeing it testing out 2.1) === frankban|afk is now known as frankban [08:25] jam: not that I know of [08:25] jam: I don't think it's fixed in 2.2 [08:30] bug #1669297 [08:30] Bug #1669297: diskmanager log spam about /dev/loop

[08:30] (just filed) [08:47] wallyworld: did you see my response? [12:00] axw: did we break "juju upgrade-juju" with development tools in 2.1.1/ [12:00] ? [12:00] I swear it always worked for me to just keep running that command [12:00] now I'm getting "newer version found" [12:00] 2.1.1.1 [12:00] but it should be creating 2.1.1.2 [12:00] ah.. just a bad error [12:01] running 'upgrade-juju' in not-the-controller-model gives a bad error [12:01] jam: --build-agent ? [12:01] perrito666: no, I just needed to "-m controller" [12:01] yes [12:01] build-agent told me that "you can only do that in the controller model" which pointed me in the right direction [12:02] its a bit dumb though that if you --build-agent it fails to upgrade because not on the controller instead of upgrading the controller... I used --build-agent I know what I want [12:04] axw: yeah, saw it but had to go to soccer. i realise the tests are faulty because the before hook reuses the app object instead of getting a new one, so will fix [12:06] perrito666: math.rand is returning deterministic results... was that intentional? [12:06] I've bootstrapped 2x, and both got "10.0.18.0" for lxdbr0 [12:07] jam: I would definitely not expect that, the docs dont mention it [12:09] perrito666: its standard for C libs [12:09] 'rand' is deterministic so you have to call srand first [12:09] perrito666: crypto/rand is properly always random but probably doesn't have Perm [12:09] usually its srand(time()) [12:10] perrito666: care to poke at that? [12:10] maybe we could open a bug on 2.1.1 so that we know its being fixed [12:10] jam: at all, Ill fix it, so much for a high level languaje [12:10] https://golang.org/pkg/math/rand/ [12:11] says: use a default shared Source that produces a deterministic sequence of values each time a program is run. [12:11] perrito666: ^^ [12:11] 'use the Seed function' [12:12] jam: yup, I read the doc for Perm only, my bad [12:13] using Seed seems fine for our use case [12:13] yup, Seed with timenano should fix it [12:13] * perrito666 writes a lib that sends an email to a bunch of D&D nerds to throw dice and reply with an int slice [12:20] jam: did you merge this with 2.2 or 2.1? [12:21] perrito666: heh [12:21] perrito666: its in 2.1 [12:22] jam: you did some nassty things there [12:22] jam: look https://github.com/juju/juju/compare/2.1...perrito666:use_random_seed_for_lxd_ips?expand=1 [12:24] something doesn't look right with that [12:24] did I target the wrong version? [12:25] jam: well for once the test diff is not there [12:25] :p [12:25] so it is even weirder [12:25] https://github.com/juju/juju/pull/7054 [12:26] perrito666: did you "git update remote" first? [12:26] sorry git remote update [12:26] jam: mmm I git fetched [12:26] but the reset might not have taken [12:26] * perrito666 goes again [12:27] * perrito666 pulls -r [12:30] jam: btw, put a linter in your editor ffs :p [12:53] jam: there you go, I had messed my own repo https://github.com/juju/juju/pull/7061 [12:54] wallyworld: LGTM, thanks [12:54] lgtm [13:02] perrito666: I see you're working on the container networking, does this have anything to do with bundles being forced to specify spaces? [13:03] recently I've started to see: juju.provisioner failed to prepare container "20/lxd/1" network config: unable to find host bridge for space(s) "" for container "20/lxd/1" [13:06] http://github.com/juju/juju/pull/7062 [13:08] SimonKLB: I am not sure [13:08] jam: plese do some cleanup on that pr, your commit tree is a mess, do some squashing [13:09] I'm morally opposed to push --force, but I'll squash it a bit before $$merge$$ [13:12] axw: ty for review [13:34] jam: perhaps you could shed some light on it? im trying to run my bundle on the lxd provider, but spaces dont seem to be supported there yet? are the warnings nothing to worry about? [13:35] SimonKLB: as a quick question, are these running trusty? [13:35] I believe I have a fix in flight that should improve that case with 2.1.1 so you won't get a warning [13:36] jam: nope they are all xenial [13:37] jam: ship it [13:37] perrito666: so 'git log upstream/2.1...HEAD' shows the extra commits but "git rebase -i upstream/2.1" won't show me anything since the last time I merged 2.1 [13:38] perrito666: I tried --no-fork point which seemed hinted by the readme [13:38] but haven't found anything that will actually let me squash the 2 merges into a single one [13:39] jam: git rebase -i [13:40] sinzui, ping [13:40] hi niedbalski [13:42] sinzui, hey curtis :) , I am having issues while trying to set the apt-mirror config option in the model, I don't see this sed replacement (https://github.com/juju/juju/blob/eb98521dc187927bebe390e0172136033f80487e/cloudconfig/cloudinit/cloudinit_ubuntu.go#L157) appended on the cloud-init userdata. (https://pastebin.canonical.com/181281/) , are you aware on a similar issue? [13:44] niedbalski: This is new to me. I recall there are are a few issues related to security updates only come from ubuntu.com [13:46] perrito666: I was able to cleanup one or 2 of the commits, but the others were actual conflict resolution/bringing in changes from 2.1 that I actually need. [13:47] sinzui, yes, the security updates are part of the issues I am tracking, do you know any lp bug related to it? [13:47] jam: ok, I forgive you :p [13:48] niedbalski: Bug 1606487 bug 1599886 [13:48] Bug #1606487: apt-mirror does not override security.ubuntu.com for controller [13:48] Bug #1599886: apt-mirror does not override security.ubuntu.com for containers on trusty

Yakkety):Fix Released> [13:48] jam: I am usually only dense when I see things like "went to lunch" "changing computers" or "typo" [13:51] jam: ideally each commit you send should compile [13:54] sinzui, thanks [14:05] do I have to do anything special when deploying a bundle with lxd containers when running on an lxd provider? [14:05] im getting: juju.provisioner cannot start instance for machine "0/lxd/0": Failed to change ownership of: /var/lib/lxd/containers/juju-51d4de-0-lxd-0/rootfs [14:06] seem to be a problem with nested lxd? === mskalka_ is now known as mskalka [14:20] fixed it by adding 'security.privileged: "true"' in the juju-default lxd profile === frankban is now known as frankban|afk [20:52] well.... shit. [20:54] thumper: that triggers an automated response by me to go and check the mail [20:54] perrito666: no... no email about this yet [20:54] I'm just wrapping my brain around 1.25 -> 2.x issues [20:55] and I think what I need to do is to effectively have both codebases vendored (effectively) into my upgrade app [20:55] rather than dealing with external repos [20:55] due to needing both versions [20:55] I need to be able to use 1.25 state [20:55] and 2.0 api [20:55] and both client disk formats [20:55] heh, well fuck is a strong understatement [20:56] so... [20:56] hmm [20:56] fun... [20:56] oh good [20:56] menn0: got a minute [20:56] I need to discuss something / get a sanity check [20:57] can I deploy a kvm container in an lxd provider? [20:57] perrito666: no [20:57] at least I don't think so [20:57] we used to explicitly stop any container in lxc containers [20:57] I assume that lxd also stops kvm [21:03] thumper: hi, yep, 1:1? [21:04] menn0: ack [21:57] wallyworld: ping? [21:58] yo [21:59] wallyworld: I got a bit confused yesterday when I redid the tests for the discover spaces worker and realised it never sends the CIDRs of the subnets to the API? [21:59] correct, there's a backend cache which does it [21:59] when the first api call is made [21:59] wallyworld: seems weird [21:59] it does [21:59] did you want a quick HO? [22:01] wallyworld: sure! I mean, I don't think I need one - the tests are working now, just threw me for a bit. [22:02] babbageclunk: let's chat quickly just make make sure we are on the same page [22:02] wallyworld: sounds good [22:03] babbageclunk: in 1:1 [23:32] thumper: can you please merge https://github.com/juju/description/pull/2? [23:32] seems I do not have rights [23:44] axw: yep [23:44] axw: done [23:44] thumper: cheers