15:00:05 #startmeeting kolla 15:00:06 Meeting started Wed Jan 30 15:00:05 2019 UTC and is due to finish in 60 minutes. The chair is egonzalez. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:07 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:09 The meeting name has been set to 'kolla' 15:00:15 #topic agenda 15:00:21 * Roll-call 15:00:21 * Announcements 15:00:21 * CI state https://etherpad.openstack.org/p/kolla-ci-status 15:00:22 * Debian/arm64 CI job status - https://review.openstack.org/557659 15:00:22 * placement-api split https://review.openstack.org/#/q/topic:split-placement+(status:open+OR+status:merged) 15:00:23 * Ubuntu stein UCA (python3) https://review.openstack.org/#/c/627954/ and https://review.openstack.org/#/c/625298/ 15:00:25 * Zuul promote pipeline for container publishing http://lists.openstack.org/pipermail/openstack-discuss/2019-January/002026.html 15:00:27 * open discussion 15:00:32 #topic rollcall 15:00:43 sup guys o/ 15:00:51 o/ 15:03:52 o/ 15:03:59 o/ 15:04:48 #topic Announcements 15:05:02 Hi, i dont have any update 15:05:15 any announcement? 15:06:54 #topic CI state https://etherpad.openstack.org/p/kolla-ci-status 15:07:10 kolla builds are broken due some monasca and percona repos 15:07:20 patchs are under review 15:07:46 any other gate issue im not aware of? 15:09:09 the patch at https://review.openstack.org/#/c/633906/ to fix the gate still need some love, it fixed the xtrabackup and monasca-grafana images but broke a bunch of other images 15:11:38 hrm, percona repo is disable by default to not change other deps, only for the backup image. installing from the url enables it 15:12:47 nice catch egonzalez, I'm going to revert to a previous patch set that was a bit more conservative 15:13:24 thats why toolbox and mariadb fail, we install maria from centos repos 15:14:27 anyway, any other error in CI? 15:14:36 not that I know of 15:14:37 thanks mandre for the fix 15:15:47 hrw is not around for the debian gates job topic, lets jump to the next one 15:16:08 #topic placement-api split https://review.openstack.org/#/q/topic:split-placement+(status:open+OR+status:merged) 15:16:28 status on this havent change since last week 15:16:44 it works for all source deploys/builds and centos binary 15:16:57 ubuntu binary still missing the stein UCA 15:18:47 any question/suggestion/update on the placement spli? 15:21:30 guess not 15:21:35 #topic Ubuntu stein UCA (python3) https://review.openstack.org/#/c/627954/ and https://review.openstack.org/#/c/625298/ 15:21:53 chason_ if around, whats the status of this changes? 15:24:01 o/ 15:26:15 This patch is waitlisting for some test 15:26:34 *waiting 15:26:42 cool 15:26:54 any ETA to be ready for merge? 15:28:31 Actually I do't really figure out how to test it. 15:29:41 deploying ubuntu binary? if service APIs works is that the images and ansible is fine 15:30:42 Should I build images based on the patch about this stuff of kolla and test my changes? 15:31:21 chason_ yes, that the way 15:31:38 OK, I will try it. :) 15:32:00 cool, any other thing on this topic? 15:32:25 Nothing from me now. 15:32:55 #topic Zuul promote pipeline for container publishing http://lists.openstack.org/pipermail/openstack-discuss/2019-January/002026.html 15:33:04 not sure who added this topic 15:33:52 my 2cents on this, we have our own pipeline to build and publish the images, i dont think a common openstack pipeline will work with our builds 15:34:02 at least not for source images 15:34:52 o/ 15:35:02 argh, need better calendar notifications 15:36:25 mgoddard: no problem 15:37:02 like i said, not sure who added this topic, if there is nothing else to discuss on this, we con move on to open discussion 15:37:08 it was me! 15:37:32 I think it could be good for us 15:37:41 my 2cents on this, we have our own pipeline to build and publish the images, i dont think a common openstack pipeline will work with our builds 15:37:41 at least not for source images 15:37:54 i said that ^^ 15:38:18 I like the idea of image promotion - tag with change-, push, then if testing passes, retag as master/rocky etc 15:38:43 could be good to publish a kolla-ansible image 15:39:10 currently we build both in build and gate pipelines, it's not necessary, and actually bad since we're not pushing what's been tested 15:39:45 thats true 15:39:54 ideally, if we could add a dependency from ansible jobs on build jobs, we could deploy the exact same containers 15:40:19 then if those tests pass, promote the images - just a registry API call 15:40:36 I don't think we can use the new roles, but we could use the new promote pipeline 15:41:01 is there work in progress? the promote is good for us 15:41:24 I haven't started to look at it, just read the ML post 15:42:02 anyway, that's all, just thought I'd bring it up 15:42:25 thanks, the promote would be good to have 15:42:56 any other thing on this topic? 15:45:27 #topic open discussion 15:45:38 something else missing? 15:47:18 I made this thing: https://github.com/markgoddard/bikolla 15:47:22 #link https://github.com/markgoddard/bikolla 15:47:39 uses kolla-ansible to deploy a standalone ironic cluster 15:47:56 you can also use Tenks to deploy virtual bare metal nodes for testing 15:48:25 I have some changes to apply to kolla ansible to make standalone ironic easier to configure 15:48:34 will update my patch 15:48:45 good one 15:48:45 cool, that looks really good 15:49:04 bikolla should work now though - if anyone is interested, give it a try 15:49:09 is ironic standalone deployed by kolla very different from upstream bifrost? 15:49:40 it's very similar, except it runs in containers 15:50:02 and could be multinode 15:50:05 mgoddard: loks nice 15:51:09 good timing with the recent baremetal SIG announcement 15:51:14 anyway, that's all 15:51:27 im waiting for a superuser blog about bikolla :) 15:51:29 mgoddard: do you deploy some web server/object storage to store images? 15:52:05 mnasiadka: the ironic_ipxe server is based on apache 15:52:14 mgoddard: ah, ok - got it 15:52:28 similar to bifrost, which uses nginx 15:52:51 mgoddard: true, so you just copy in the images in there - in a multinode scenario might be a bit challenging to keep them in sync 15:52:56 egonzalez: yeah, was thinking about a stackhpc blog, perhaps it will be selected for superuser if it gets enough hits :) 15:53:32 mgoddard you can propose to publish there i think, not need to wait for a request from them 15:53:35 mnasiadka: yeah, they would all need different names 15:53:47 egonzalez: cool, I'll keep it in mind 15:54:13 one nice thing about this vs bifrost is it would be easy to add in openstack services if necessary 15:54:23 e.g. if you want glance for better image support 15:54:30 or neutron for networking 15:56:24 and finally not have one container with ansible running inside as bootstrap :) 15:56:49 anything left for open discussion? 15:57:48 just FYI - skydive has finally fixed zombie problems when running in docker container 15:58:10 they will make a release in a week or so, I'll track it 15:58:46 cool, will need some upgrade tasks or just bumnping version on the images? 15:59:00 just bump the version, nothing like command args or config file changes 15:59:30 But I'll need to add some prometheus upgrade tasks and fix command args - related to bumping up version from 1.6 to 2.x 15:59:36 (another topic) 15:59:58 please add the topic for next week meeting 16:00:05 we're out of time today 16:00:13 thanks for comming 16:00:16 #endmeeting