[03:15] Bug #1478762 opened: juju bootstrap failed to connect to environment with error "discarding API open error" [09:01] fwereade: hangout? [09:01] dooferlad, sprinting, a bit inconvenient [09:01] fwereade: have fun! [09:02] ok [09:02] dooferlad, cheers :) [10:41] dooferlad: no answer from the maas guys so far, will ask again later. but found the reason for my failing code, a type assertion doesn't work as wanted [10:41] TheMue: thanks for chasing them. [10:42] TheMue, dooferlad: can I borrow one of you for a bit of independent sanity-verification? [10:42] fwereade: sure [10:42] TheMue, dooferlad: I have a branch called final-leadership-switch-1.24 [10:43] TheMue, dooferlad: builds fine, tests not fixed [10:43] TheMue, dooferlad: would you please run an HA environment with a bunch of ubuntu units deployed [10:43] TheMue, dooferlad: and let me know if it appears to work right [10:44] TheMue, dooferlad: just started one myself [10:44] TheMue, dooferlad: and *very* preliminary indications are good [10:44] TheMue, dooferlad: but please try to break it if you can [10:44] fwereade: if you can point me to instructions, sure. [10:44] fwereade: haven't done anything with HA yet [10:45] dooferlad, `juju ensure-availability` [10:45] fwereade: but I am always happy to try and break things :-) [10:45] dooferlad, leadership is charm-facing not user-facing [10:46] dooferlad, the hook tools you'll want to poke are `is-leader`, `leader-set` and `leader-get` [10:46] dooferlad, will be back shortly, have to go to lunch now [10:46] fwereade: OK, enjoy. [10:47] dooferlad, tyvm [10:51] f*ck [10:51] sorry [10:52] assertion fails due to usage of mocking interfaces, changing them to concrete types makes mocking harder again [10:53] have to thing about it [11:06] * dooferlad goes for lunch [11:29] dooferlad: (a) hangout is canceled, (b) found a solution for my mentioned problem, yeah [12:13] dooferlad, TheMue: any findings? +ve or -ve? :) [12:18] fwereade: I kicked off what should have been a simple MAAS-in-kvm bootstrap and ensure-availability, but the bootstrap failed. Just back from lunch so will debug and retry once I have 3+ machines again. [12:21] ah, DNS you swine. [12:27] dooferlad, cheers [12:30] fwereade: do you know why unit logging is always forced to be DEBUG ? [12:30] rogpeppe2, hmm, I *think* it might be because people got upset that we'd hidden hook output from debug-log by default [12:30] rogpeppe2, long time ago [12:31] fwereade: right - i see it changed start of 2014 [12:31] fwereade: i just reported https://bugs.launchpad.net/juju-core/+bug/1478936 [12:31] Bug #1478936: environs/config: fails if logging config is not attr=value [12:32] fwereade: which was causing my tests to fail [12:33] rogpeppe2, shouldn't your logging-config be "=DEBUG"? [12:33] rogpeppe2, not just DEBUG [12:33] fwereade: just "DEBUG" is a valid logging config [12:33] fwereade: it's equivalent to "=DEBUG" but much easier to type [12:33] rogpeppe2, ah,, good to know [12:37] Bug #1478934 opened: IPv6 openstack security group rules not managed by juju in dual stack [12:37] Bug #1478936 opened: environs/config: fails if logging config is not attr=value [12:49] Bug #1478934 changed: IPv6 openstack security group rules not managed by juju in dual stack [12:49] Bug #1478936 changed: environs/config: fails if logging config is not attr=value [12:55] Bug #1478934 opened: IPv6 openstack security group rules not managed by juju in dual stack [12:55] Bug #1478936 opened: environs/config: fails if logging config is not attr=value [12:55] Bug #1478943 opened: Rsyslog certificate fails when using IPv6/4 dual stack with prefer-ipv6: true [12:58] Bug #1478943 changed: Rsyslog certificate fails when using IPv6/4 dual stack with prefer-ipv6: true [13:07] Bug #1478943 opened: Rsyslog certificate fails when using IPv6/4 dual stack with prefer-ipv6: true [13:32] ok, so running 8 kvms in a MAAS cluster containing a HA MySQL and Ceph cluster may slow my machine down a bit. [13:33] Maybe I should upgrade to 32GB RAM [13:36] dooferlad: I had the same thought, then I finally admitted that I should not be doing that on a laptop :p [13:36] perrito666: oh, this is a desktop. [13:37] perrito666: it is fine now, it was just bootstrapping several machines at once that seemed to be a problem. Now down to 13GB in use. [13:39] I think I could put 5 physical nodes + 1 master together, but power control would be painful. The 2 nodes, 1 master set up I have is enough for most work. [13:39] trust fwereade to break things :-) [14:07] I definitely need to make git co run godeps [14:13] Bug #1478983 opened: Cannot resolve a blocked unit [14:22] mgz, pinjg [14:23] what is that, chinese ping? [14:24] perrito666, ;p [14:29] fwereade: hey [14:30] mgz, so, I would like to distribute a deeply hacky build of juju, and it has been suggested that a ppa would be a good way to go [14:30] mgz, is there a quick easy way I can cause this to occur? [14:30] nope. [14:31] for starts, if you care about more than the client [14:31] mgz, ok, problem solved, I'll send a .tgz with the two binaries and --upload-tools ;) [14:31] whoever you want to use this either needs a very temp env with --upload-tools and only one arch, or set up their own streams hosted somewhere [14:31] so a ppa gets you nothing [14:32] fwereade: I could help you there [14:33] if you want people to test a fix, just giving them some tools built for the right arch and instructions on how to avoid using published streams by mistake is probably what you want [14:33] if they're intending to leave that environment up, that ends up being painful [14:33] fwereade: all make sense? [14:34] mgz, yeah, think so, have to come back to it later anyway [14:42] TheMue: do you have a moment to help with a compile error? https://ide.c9.io/dooferlad/juju [14:42] dooferlad: sure [15:09] TheMue: ok, so I am going to have to re-think this. It sucks. [15:10] TheMue: I need to get more state out of the functions I want to test, but I can pass things in. I think. [15:12] dooferlad: yeah, had it the same, needed state and wanted to avoid state in tests and work with a subinterface as mock [15:13] dooferlad: had to change the entity watcher interface for it, thankfully I'm so far the only user. [15:52] Bug #1479024 opened: TestRootDiskTags localServerSuite fails on ppc64el

[16:04] Bug #1479024 changed: TestRootDiskTags localServerSuite fails on ppc64el

[16:14] Bug #1479024 opened: TestRootDiskTags localServerSuite fails on ppc64el

[18:53] Is there some traditional way of delivering a patched version of juju to a bug reporter for verification that it fixes their problem? [18:54] cherylj: not that I know [19:00] google drive or dropbox... used to be Ubuntu One but... yeah no [23:32] perrito666: che? [23:33] axw: directed to anastasiamac being revolutionary :) [23:33] aha :) [23:34] perrito666: axw: D someone has to take the stance, rite :D [23:35] * perrito666 sends the beret, cigar and fake beard pack to anastasiamac [23:38] perrito666: I think that tent is more important - I'll neee to live somewhere too :)