15:00:57 <mgoddard> #startmeeting kolla 15:00:57 <openstack> Meeting started Wed Apr 7 15:00:57 2021 UTC and is due to finish in 60 minutes. The chair is mgoddard. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:58 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:00 <openstack> The meeting name has been set to 'kolla' 15:01:07 <mgoddard> #topic rollcall 15:01:14 <yoctozepto> \o/ 15:01:22 <hrw> _o°o_ 15:01:33 <mgoddard> (-o-o-) 15:01:56 <headphoneJames> (-o-O-o-) 15:04:45 <mgoddard> #topic agenda 15:04:55 <mgoddard> * Roll-call 15:04:57 <mgoddard> * Announcements 15:04:59 <mgoddard> * Review action items from the last meeting 15:05:01 <mgoddard> * CI status 15:05:03 <mgoddard> * Review requests 15:05:05 <mgoddard> * Wallaby release planning 15:05:07 <mgoddard> ** (kolla-ansible) How should we approach the switch to Quay on the user side? 15:05:09 <mgoddard> ** (kolla) Should we permanently remove support for debian/binary in Wallaby? 15:05:11 <mgoddard> ** (kolla) Or should we move to Debian 'bullseye'? We can build images but it needs work, tests and K-A support 15:05:13 <mgoddard> ** (kolla-ansible) Should we deprecate ``enable_host_ntp``? 15:05:15 <mgoddard> ** (kolla-ansible) How/when should we handle chrony container removal? 15:05:17 <mgoddard> ** Life after monasca-grafana https://review.opendev.org/c/openstack/kolla/+/784901 15:05:19 <mgoddard> #topic announcements 15:05:54 <mgoddard> #info Kolla is in feature freeze! 15:06:13 <mgoddard> Cores, please do not approve feature patches on master that do not have an FFE 15:06:32 <mgoddard> Any others? 15:06:49 <hrw> RP+2 for FFE as usual? 15:07:10 <hrw> and RP-1 for anything not-this-cycle? 15:07:41 <mgoddard> yes 15:08:10 <headphoneJames> when it the cut off date for patches? 15:08:16 <hrw> headphoneJames: week ago? 15:08:28 <headphoneJames> that have FFE 15:08:35 <hrw> joking, sorry 15:09:05 <mgoddard> I think we normally say 2 weeks after FFE, but really ASAP 15:09:46 <mgoddard> #topic Review action items from the last meeting 15:09:58 <mgoddard> mgoddard email openstack-discuss about quay.io credentials 15:10:00 <mgoddard> mgoddard switch kayobe to c8s 15:10:02 <mgoddard> no 15:10:04 <mgoddard> yes 15:10:10 <mgoddard> #action mgoddard email openstack-discuss about quay.io credentials 15:10:20 <mgoddard> #topic CI status 15:11:52 <mgoddard> monasca-grafana failing, added to allowed-to-fail images 15:12:12 <hrw> sorry, forgot to update 15:12:24 <mgoddard> Train: CentOS 7 curl unable to talk to docker.com (SSL issue) 15:13:05 <hrw> fixed should be 15:13:12 <mgoddard> ok 15:13:16 <hrw> if it was expire issue 15:13:22 <yoctozepto> it was not 15:13:29 <yoctozepto> centos7 is unable to connect to docker.com 15:13:42 <yoctozepto> it cannot negotiate tls 15:14:53 <mgoddard> bah 15:15:06 <mgoddard> train seems a bit borked 15:15:19 <mgoddard> master: rabbitmq is failing to upgrade in multinode (e.g. ceph) 15:15:32 <mgoddard> what's that one? does it always happen? 15:15:33 <hrw> hm. 'docker run centos:7; yum -y install curl, curl -L https://docker.com' works 15:16:45 <priteau> I confirm `curl -L https://download.docker.com/linux/centos/7` works fine from a c7 box 15:17:06 <yoctozepto> yeah, it is something in CI 15:17:24 <yoctozepto> or perhaps something got fixed 15:17:32 <yoctozepto> but the problem was not with cert but tls itself afair 15:17:53 <yoctozepto> mgoddard: that one is random 15:19:06 <yoctozepto> let's check train today 15:19:39 <openstackgerrit> Mark Goddard proposed openstack/kolla stable/train: Bump OpenStack versions for Train https://review.opendev.org/c/openstack/kolla/+/785222 15:19:41 <mgoddard> ^ 2 birds one stone 15:19:55 <mgoddard> I think we haven't done that in a while 15:20:07 <hrw> ;) 15:20:07 <yoctozepto> yes, we have not 15:20:14 <mgoddard> #topic Review requests 15:20:29 <hrw> +2 15:20:42 <yoctozepto> https://review.opendev.org/c/openstack/kolla-ansible/+/670104 15:20:43 <mgoddard> hrw: at least wait for CI... 15:21:19 <mgoddard> needs a second core to complete FFE 15:21:22 <hrw> mgoddard: fix is right. CI does not allow to pass broken job 15:21:46 <mgoddard> hrw: except for non-voting checks, which we always check, right? 15:22:54 <mgoddard> anyone else want to review yoctozepto's patch with me? 15:23:15 <yoctozepto> I heard mnasiadka wants to 15:23:21 <hrw> I do not understand ansible enough 15:23:49 <mgoddard> yoctozepto: please follow up with mnasiadka 15:23:54 <mgoddard> any other reviews? 15:24:09 <hrw> I am fine 15:24:28 <mgoddard> #topic (kolla-ansible) How should we approach the switch to Quay on the user side? 15:24:34 <mgoddard> yoctozepto: 15:24:50 <yoctozepto> yes, how? 15:25:33 <hrw> does it mean 'container_registry: "quay|docker" # choose one' in globals.yml? 15:25:42 <hrw> or sth like that? 15:26:11 <hrw> cause as I user I would love to see it with this level of complication 15:26:42 <mgoddard> it could be made that simpple 15:27:06 <yoctozepto> well, then it's already like that 15:27:15 <mgoddard> not quire 15:27:16 <mgoddard> quite 15:27:19 <yoctozepto> well 15:27:24 <yoctozepto> but close 15:27:26 <hrw> it can ofcourse be as 'choose this or that block of values' as registry/namespace differ 15:28:07 <yoctozepto> so 15:28:13 <yoctozepto> documenting in globals.yml 15:28:44 <hrw> user needs to edit globals.yml before deployment so it is best place imho 15:28:49 <mgoddard> I think we missed a question 15:29:04 <mgoddard> should we advertise both dockerhub and quay.io to users? 15:29:13 <yoctozepto> yes, that was the question 15:29:22 <mgoddard> I see 3 options 15:29:39 <hrw> 1. go quay 2. go docker 3. say how to choose? 15:29:39 <mgoddard> 1. support both dockerhub and quay.io 15:30:12 <mgoddard> 2. support quay.io, keep dockerhub until Victoria is dead 15:30:21 <mgoddard> 3. support dockerhub, use quay.io for CI only 15:31:29 <hrw> 2nd sounds good to me if it means that only victoria is on dockerhub. 15:31:31 <mgoddard> benefit of 3 being that if we decide quay is not the way, we can quietly drop it 15:32:37 <mgoddard> we might want to change the namespace to something like openstack.kolla.ci 15:32:43 <mgoddard> or test 15:32:57 <yoctozepto> well, as long as we don't advertise 15:32:59 <yoctozepto> it's irrelevant 15:33:11 <mgoddard> except... search engines... 15:33:26 <yoctozepto> well, I would not mind? 15:33:41 <yoctozepto> but perhaps we are sure of quay and want just offer it 15:34:21 <mgoddard> are we sure of quay? 15:35:17 <hrw> let us do whole cycle with quay. publishing to both quay and dockerhub, CI on quay. 15:35:29 <hrw> this way at end of Xena we will know 15:35:38 <mgoddard> I think that would be sensible 15:35:47 <mgoddard> lets get 2-3 months under our belt 15:37:05 <hrw> yep 15:37:07 <mgoddard> agreed? 15:37:18 <yoctozepto> agreed 15:37:25 <yoctozepto> so not doing anything more at the moment 15:37:27 <yoctozepto> agreed? 15:37:41 <hrw> looks like 15:38:36 <mgoddard> +1 15:38:42 <yoctozepto> ok, moving on 15:38:48 <mgoddard> #topic (kolla) Should we permanently remove support for debian/binary in Wallaby? 15:38:59 <mgoddard> (kolla) Or should we move to Debian 'bullseye'? We can build images but it needs work, tests and K-A support 15:39:12 <yoctozepto> I think bullseye is the way to go 15:39:17 <yoctozepto> need to swallow that change 15:39:29 <hrw> and k-a needs work 15:39:46 <mgoddard> how much work are we looking at? 15:39:48 <hrw> as mariadb/security_reset.expect does not fit 15:40:33 <mnasiadka> Well, that’s due to newer mariadb than ever I guess 15:41:01 <hrw> 10.5.9 15:41:13 <yoctozepto> not good 15:41:48 <hrw> note that this script was not changed since 2015 15:42:02 <hrw> so probably no one here touched it 15:42:25 <mgoddard> hopefully not rocket science to fix though 15:42:43 <mgoddard> might be a case of changing some questions & answers 15:42:54 <hrw> yep 15:42:57 <hrw> and keeping separate copy for Debian 15:43:06 <mgoddard> ok 15:43:16 <hrw> as we can handle it easily in kolla 15:43:17 <mgoddard> let's see how it goes 15:43:58 <mgoddard> hrw: resident debian expert, are you willing to handle it? 15:44:08 <mgoddard> perhaps with help from kevko 15:44:14 <hrw> mgoddard: I do not even know what it is supposed to do 15:44:24 <mgoddard> hrw: I mean bullseye in general 15:44:30 <hrw> in general, yes 15:44:33 <mgoddard> ok 15:44:37 <hrw> kolla part is ~done 15:44:57 <hrw> all images builds, masakari is disabled for binary but it may change 15:45:17 <hrw> now we need to deploy and check how things look 15:45:36 <mgoddard> ok, keep going and speak up if it starts getting too big 15:45:41 <mgoddard> #topic (kolla-ansible) Should we deprecate ``enable_host_ntp``? 15:46:02 <yoctozepto> should we? 15:46:06 <yoctozepto> I think we should 15:46:38 <mgoddard> it's not the default NTP daemon on any supported OS, is it? 15:46:46 <yoctozepto> it's not 15:46:54 <yoctozepto> it's not well supported any longer 15:46:55 <mgoddard> not supported on CentOS 8 at all 15:47:08 <mgoddard> deprecate 15:47:14 <hrw> can't we just say 'keep time in sync' in docs? 15:47:14 <yoctozepto> ok, action on it 15:47:54 <mgoddard> #action yoctozepto deprecate enable_host_ntp 15:48:17 <mgoddard> #topic (kolla-ansible) How/when should we handle chrony container removal? 15:48:55 <mgoddard> during upgrade? 15:49:42 <mgoddard> separate play that fires even if disabled 15:50:08 <yoctozepto> and what if someone upgrades and then decides to disable? 15:50:17 <yoctozepto> we need a strategy on these cleanups 15:50:33 <yoctozepto> deploy/upgrade is tricky as it is guarded by the boolean 15:51:50 <mgoddard> for monasca we added a separate cleanup command 15:53:06 <mgoddard> I don't think the general case is something we can decide here 15:53:10 <yoctozepto> hmm 15:53:14 <mgoddard> there are a few variables 15:53:31 <mgoddard> disable entire service? disable one container in an enabled service? 15:53:40 <mgoddard> stop supporting a service/container? 15:54:11 <mgoddard> I feel like if we remove a service, then the upgrade is the right time to handle it 15:54:28 <mgoddard> if someone else disables a service, we need a new command to clean up 15:55:00 <mgoddard> we have 5 minutes and one more item 15:55:07 <yoctozepto> so removal only in xena? 15:55:13 <mgoddard> yes 15:55:16 <yoctozepto> ok 15:55:21 <mgoddard> #topic Life after monasca-grafana https://review.opendev.org/c/openstack/kolla/+/784901 15:55:22 <yoctozepto> then no action for now 15:55:34 <yoctozepto> dougsz promised to handle this on k-a side 15:55:38 <mgoddard> right 15:55:58 <mgoddard> any more to discuss? 15:55:59 <dougsz> there is at least a sensible alternative of using the Monasca datasource with vanilla Grafana 15:56:11 <dougsz> nothing else on this 15:56:16 <yoctozepto> makes sense 15:56:25 <mgoddard> #topic Open discussion 15:56:35 <mgoddard> 4 minutes... 15:56:39 <mgoddard> anything else? 15:56:43 <openstackgerrit> Doug Szumski proposed openstack/kolla-ansible master: Support enabling Prometheus Fluentd exporter https://review.opendev.org/c/openstack/kolla-ansible/+/785229 15:59:30 <mgoddard> thanks all 15:59:32 <mgoddard> #endmeeting