08:09:43 #startmeeting multi_arch 08:09:44 Meeting started Tue Mar 31 08:09:43 2020 UTC and is due to finish in 60 minutes. The chair is ricolin. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:09:45 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 08:09:47 The meeting name has been set to 'multi_arch' 08:09:50 hey ricolin ! 08:10:21 hope everyone is fine 08:10:27 tonyb: yeah, kids facetime and minecraft and whatnot ... they're very adaptable which is good 08:10:32 it's a crazy week 08:11:24 tonyb, assume you didn't need to adjust much since you already WFH:) 08:12:10 ricolin: Not too bad really but I reliased how much I have come to appreicate a quiet house ... now that it isn't ;P 08:12:23 lol 08:12:32 ricolin: how're things for you? 08:12:49 ACtually we should chat after the meeting :) 08:13:17 haha 08:13:20 fine! 08:13:37 :) 08:13:45 I put in two topic today 08:13:50 CI (as usual) 08:13:54 and Virtual PTG 08:14:02 anything else we should discuss today? 08:14:21 no from me 08:14:35 s/topic/topics/ 08:14:55 not from me 08:15:01 okay! 08:15:03 #topic VPTG planning 08:15:16 so what's plan! 08:15:33 *the plan 08:16:06 fyi infra has been working on jitsi i think 08:16:16 How much time do people want to dedicate to it? 08:16:46 If we could commit to say 4 hours we could do a jitsi call and hack on some aspect of the CI 08:17:02 for current status, I assume 1/4-1/2 day? 08:17:06 peobably couldn't merge much in that window but who knows? 08:17:21 4 hours sounds reasonable 08:17:24 true i think it's just about "i'm doing this and nothing else for a set period" 08:17:41 what about 2 hour sessions? 08:17:45 with a gap day? 08:17:59 It'd mean we had a few days to merge stuff ? 08:19:03 you mean like 2 hours in first day, and 2 for the third day? 08:19:13 ricolin: Yeah soemthing like that 08:19:21 ricolin: as a "hackfest" thing 08:19:27 works for me 08:19:37 we could also do status/education stuff 08:19:41 will it make sense if we also host BOF this time? 08:19:51 Sure 08:20:02 tonyb, good idea, but what to educate about? 08:20:54 ricolin: I don't know exactly .. the SIG mission, "sponsors" consumers? 08:21:05 it woudl be good if devstack was working, and people could maybe try their plugins 08:21:31 ianw: Yeah that's be good 08:22:18 https://zuul.opendev.org/t/openstack/build/efdb93a189194b6b9df912d07c59e726/logs is a run i did this afternoon, it looks like a lot of it passed 08:22:29 ianw, we still got about two months so it might work that time:) 08:22:32 anyway, picking bits of that would be good for group discussion 08:24:04 we can also have SIG status report for sure 08:25:19 VPTG discussion will host online this week (4/2) 08:25:30 will have more detail from there 08:26:01 Cool 08:26:44 Wonder if users/ops will join too:) 08:26:57 I don't know 08:27:36 I can tell my friends that use Multiarch OpenStack once we know when it'll be 08:27:46 they're $my TZ 08:28:23 thanks tonyb! Some broadcast is also what we need:) 08:29:46 ricolin: Well we can spamm the twittreverse and I can ask the osops moderators to help out 08:30:03 ricolin: there were a couple of multi-arch consumers at the last meetup 08:30:41 tonyb, local meetup? 08:31:07 ricolin: It was the London one in January 08:31:32 cool 08:31:54 so we should also host user/ops feedback!! 08:31:58 https://etherpad.openstack.org/p/LON-2020-OPS-AGENDA 08:32:16 #link https://etherpad.openstack.org/p/LON-2020-OPS-AGENDA 08:32:27 ricolin: That may be a little be of a stretch at this stage ... but sure 08:32:47 ricolin: if only we could somehow provide real beer and a virtual meetup 08:33:38 we can list beer in most have item to join this vPTG!! 08:34:27 hehe 08:35:08 guess that's all we need to discuss about VPTG in this meeting 08:35:23 #topic Follow on CI Resources 08:36:26 * tonyb is strangely silent 08:37:00 ianw, any more update on devstack patch? 08:37:10 no, i haven't been looking at that, sorry 08:37:21 kevinz and i spent some time looking at node failures in the linaro us cloud 08:37:32 turned out to be a lack of floating ips 08:37:38 so we switched it to be ipv6 only 08:37:56 that went in today, and seems to be working 08:38:07 #link https://review.opendev.org/716112 08:38:10 ianw: ooo another IPv6 only cloud that's kinda cool 08:38:26 you means that cause failure in earlier patch set of https://review.opendev.org/#/c/708317/ ? 08:39:06 ricolin: quite a few arm64 jobs, even in dib and kolla were returning NODE_FAILURE; they couldn't get a node to run on 08:39:14 now we should be able to run at capacity 08:39:37 we also discussed the linaro-london cloud, which is currently disabled, because the us cloud is better so we don't want to run jobs there 08:40:03 ianw, okay, I did saw some error message regarding that issue days ago, so just make sure it's the same thing 08:40:05 we decided to leave it for a while; if us capacity gets exceeded we can turn it back on, but while us is not at capacity it doesn't make sense 08:41:31 that's about all i have to report :) 08:42:15 If I wanted to play with tox -e py on arm64 .... which repos is zuul "watching" 08:42:15 ianw, I didn't see the wheels story in https://storyboard.openstack.org/#!/project/openstack/multi-arch-sig 08:42:27 ricolin: That's on me 08:42:39 oh! sorry ianw! 08:43:38 tonyb, just a reminder here:) 08:43:41 tonyb: umm, not sure, you could probably just add a arm64 nodeset to a tox job and it ... might ... just work 08:44:19 ianw: okay I was kinda hoping you'd say that 08:44:45 #action tonyb write up the "wheels story" (and no they don't go round and round) 08:44:50 it may run a some setup roles that may or may not be ready for that ... i'd be interested in fixing anything 08:45:06 #action tonyb try a basic tox job in a repo 08:45:27 thanks tonyb 08:47:46 from my part, the multi-arch image support for kubernetes/cloud-provider-openstack is ready, I'm working on multi-arch support for other OpenStack container images 08:48:04 like all the agent images we have 08:49:01 which will be something I plan to talk to teams in VPTG(or before) 08:49:57 For the CI resource pool, I didn't get any other company to take action to donate arm resources to CI as they once promised 08:50:10 but will keep tracking with them 08:50:48 ianw, tonyb anything happen recently regarding CI? 08:51:05 ricolin: Not for me 08:51:07 *any other things 08:51:16 the limited container images we had died 08:51:32 So thinking outside the box 08:52:31 sorry, not fully understand `the limited container images we had died` 08:52:38 try me again! 08:53:03 Hmm I need to think how we'd phrase this .... could we get the arm64 zuul to run jobs on a non opendev.org repo? 08:53:26 as in run something against github? 08:53:42 ricolin: Sorry I've been working to get ppc64le conatiners and diskimages built to so that tripleo can work with upstream sources in multiarch 08:54:04 ricolin: they were building just fine and we were abouit the publish them but then the builders failed 08:54:15 ianw: I was thinking rdoproject 08:54:23 tonyb, got it:) 08:54:48 ianw: *possibly* against a couple of opendev.org/openstack/*tripleo* repos longer term 08:55:28 tonyb: i mean, anything zuul is monitoring, could run an arm64 job; it's just a node type to choose from 08:56:36 ianw: sure for the opendev.org/openstack/*tripleo* ones yeah but zuul isnot monitoring anything in rdoproject 08:57:09 ianw: and I really don't know if one zull can monitor 2 places like that 'nor if that's something *everyone* would support/agree to 08:57:18 hence the "oustide the box" statement 08:59:26 #action ricolin update multi-arch container image support story task to storyboard 08:59:56 tonyb: it could be added; although as you say, the technical may be the least of it 09:00:14 ianw: Something to think about down the road 09:00:49 i think that because zuul is still setup by puppet mostly, https://opendev.org/opendev/system-config/src/branch/master/hiera/group/zuul-scheduler.yaml might be the place 09:02:06 Okay, shall we move the discussion back to multi-arch-sig channel?:) 09:02:39 yep 09:02:46 thanks ianw tonyb for join this meeting!:) 09:02:48 #endmeeting