14:00:33 #startmeeting kayobe 14:00:34 Meeting started Mon Jun 24 14:00:33 2019 UTC and is due to finish in 60 minutes. The chair is mgoddard. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:37 #topic rollcall 14:00:39 The meeting name has been set to 'kayobe' 14:00:43 \o 14:01:15 o/ 14:01:44 around dougsz & jovial[m] ? 14:01:53 jovial[m] might be busy 14:01:54 \o 14:01:58 sure 14:02:08 #topic agenda 14:02:18 * Roll-call 14:02:20 * Announcements 14:02:22 ** Vote on kayobe becoming a kolla deliverable: http://lists.openstack.org/pipermail/openstack-discuss/2019-June/007250.html 14:02:24 * Review actions from last meeting 14:02:26 * Kayobe whiteboard https://etherpad.openstack.org/p/kayobe-whiteboard 14:02:28 * Discussion 14:02:30 ** Stein release https://etherpad.openstack.org/p/kayobe-stein-release 14:02:32 * Open Discussion 14:02:34 #topic announcements 14:02:37 #info Vote on kayobe becoming a kolla deliverable 14:02:42 #link http://lists.openstack.org/pipermail/openstack-discuss/2019-June/007250.html 14:02:59 open to kolla cores, but opinions from others in the community welcome 14:03:04 any other announcements? 14:03:31 None from me 14:03:43 Nope 14:04:11 #topic Review actions from last meeting 14:04:32 mgoddard to enquire about GitHub repository transfers 14:04:34 mgoddard to schedule bug review and squash session 14:04:41 bad mgoddard, didn't do either 14:05:11 given that we are potentially close to becoming official through kolla, I don't think we need to do 1 14:05:24 #action mgoddard to schedule bug review and squash session 14:05:26 Agreed 14:05:29 +1 14:05:40 #topic Kayobe whiteboard https://etherpad.openstack.org/p/kayobe-whiteboard 14:06:34 Can't get to zuul builds results for some reason 14:07:03 They can be very slow to load 14:07:09 got there eventually 14:08:10 main issue seems to be semi-regular failures in kayobe-seed-upgrade-centos 14:08:13 http://zuul.openstack.org/builds?project=x%2Fkayobe&job_name=kayobe-seed-upgrade-centos 14:08:48 I think it's a mariadb upgrade issue, probably need to be better at gracefully stopping the mariadb service 14:09:09 http://zuul.openstack.org/builds?job_name=kayobe-seed-upgrade-centos 14:09:13 that's a better link 14:09:36 only seems to fail on master 14:09:58 I'll try to investigate 14:10:11 There were some failure on stable branches on 2019-06-17 14:10:12 #action mgoddard to investigate errors in kayobe-seed-upgrade-centos job 14:10:19 was that due to cliff? 14:10:24 Oh, that was something else 14:10:26 yeah, cliff 14:11:15 thanks for updating the cumulus patch priteau 14:11:40 It still needs more work but I haven't found the time to work on it lately 14:11:42 hold up, actually you just commented :) 14:12:13 we have quite a few patches requiring review 14:13:45 http://tiny.cc/kayobe-review-dashboard 14:13:46 dougsz just added a good comment on the custom blazar configuration 14:14:06 Just wondering about that in general 14:14:15 The kolla_extra_{service} thing 14:15:10 Should we only use it if Kayobe itself provides some custom configuration to Kolla Ansible? 14:15:16 https://tiny.cc/kayobe-review-dash 14:15:37 Nice ^! 14:15:52 dougsz: ideally, yes. But at the moment it's necessary for files like kolla/config/blazar.conf 14:16:55 #topic discussion 14:17:05 #topic Stein release https://etherpad.openstack.org/p/kayobe-stein-release 14:17:05 It looks more complex that it should be for services that kayobe doesn't customise 14:17:10 #undo 14:17:11 Removing item from minutes: #topic Stein release https://etherpad.openstack.org/p/kayobe-stein-release 14:17:28 it definitely is more complicated than it needs to be 14:17:51 Just to make sure I understand correctly how it works: the operator writes custom configuration in etc/kayobe/kolla/.conf, which gets moved to a variable, which gets templated into etc/kolla/config/ 14:18:17 I started this but didn't finish :( 14:18:19 https://github.com/markgoddard/kayobe/commit/f069ea53fb4b07d5ed379ddb85b783f67da68135 14:18:58 priteau: correct, except operator writes to etc/kayobe/kolla/config/.conf 14:19:15 Right, that's what I meant :) 14:19:44 And it's a bit complicated because the same code is used for files that kayobe also customises, such as ironic.conf? 14:19:51 yes 14:20:35 Is your WIP commit just doing a plain copy or template for files that Kayobe doesn't customise? 14:20:44 it's still a template 14:20:55 I keep wondering if we just want a plain copy mode 14:21:03 might be quicker 14:21:06 Ah, no templating 14:21:09 yeah 14:21:10 That could be nice 14:21:26 Particularly for Prometheus config 14:21:32 just another if to worry about though 14:21:41 which uses go templates, so there is some awkward escaping 14:21:51 Yeah, not a big issue 14:22:26 simple fix is just wrap the whole file in {% raw %} 14:22:30 Could we lose some capabilities if we stop templating on the kayobe side? 14:22:38 yes 14:22:52 sometimes it's useful to access kayobe variables 14:22:55 Yeah, but you can't use the {% raw %} thing twice 14:23:13 I've once tried to template my ceph.conf file to dynamically gather IPs of the ceph mon servers, but didn't manage to make it work... 14:23:34 we could make it easier to pass through variables to kolla-ansible, then access those 14:24:43 (example of work around for escaping twice: https://github.com/RSE-Cambridge/cumulus-kayobe-config/commit/307cd1129b57394fb0eb83d8afdfd2a8dcd9e9b3) 14:25:41 ouch 14:25:53 {% endraw %}{{ '{% endraw %}' }} 14:26:01 so that's escaping for kayobe & kolla jinja? 14:26:17 Yeah 14:26:19 inception templating levels 14:26:21 nice 14:27:32 Maybe I could add some documentation on double escaping, escaping etc and that would be enough for now 14:27:35 I suppose passing through variables is quite easy if you use globals.yml 14:27:50 that's true 14:28:16 I propose that we leave it as it is for now and take it as discussion for a design session 14:28:22 +1 14:28:35 #topic Stein release https://etherpad.openstack.org/p/kayobe-stein-release 14:28:41 This one again 14:29:02 kolla release is still not quite there :( 14:29:27 I think we are nearly on top of the last ceph issue 14:29:35 now there is a galera upgrade bug 14:30:20 but I will try to crack it soon as we do need to get this thing out 14:31:18 #topic open discussion 14:31:25 Anyone have anything for today? 14:32:19 There are some small patches pending review 14:32:30 Not important ones, but easy 14:32:38 https://review.opendev.org/#/c/664870/ 14:32:56 https://review.opendev.org/#/c/662483/ 14:33:07 https://review.opendev.org/#/c/660973/ 14:33:44 +1 14:33:52 or +2 even ;) 14:34:42 One other thing, we should probably be a bit more systematic about backports 14:35:20 In kolla I've started a ~monthly backport, and keep the last checked commit in the whiteboard 14:36:03 What would be our backport policy? Important bug fixes obviously, but what about documentation fixes or small features? 14:36:10 Sounds like a good plan to track them, it's easy to forget 14:36:28 Do you think anyone reads the stable docs? 14:37:03 I try to use the correct version of docs for other openstack projects. I don't so much for kayobe 14:37:45 +1 for docs fixes. Small features maybe, if there's a clear benefit and minimal risks 14:38:16 if we become official we'll probably have to stick to the openstack stable backport policy 14:38:18 That's what I do as well. Though looking at the latest docs for Kayobe could be unhelpful for things like which kolla configuration is copied 14:38:46 it's always a risk - docs may improve over time, or go stale 14:38:58 #link https://docs.openstack.org/project-team-guide/stable-branches.html#appropriate-fixes 14:39:35 #link https://docs.openstack.org/project-team-guide/stable-branches.html#review-guidelines 14:40:07 thanks 14:40:20 The best would be if we could support a few older OpenStack releases from the latest Kayobe version 14:40:56 This way operators could upgrade to kayobe stein to get the new features, while keeping their OpenStack on Rocky or Queens. But it's a lot of extra testing... 14:41:29 yeah, could be difficult 14:41:41 still fairly coupled to the kolla version 14:42:26 mainly through the inventory 14:42:41 if we kept old inventories around it could work 14:44:10 anything else for this week? 14:44:23 Not from me 14:44:34 ok, have 15 minutes back :) 14:44:40 thanks priteau and dougsz 14:44:43 #endmeeting