Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2023-08-22
11:45:00 auniyal sean-k-mooney ack
11:54:08 opendevreview Pierre Riteau proposed openstack/nova master: Fix a couple of typos https://review.opendev.org/c/openstack/nova/+/892300
12:28:36 opendevreview sean mooney proposed openstack/nova master: introduce global greenpool https://review.opendev.org/c/openstack/nova/+/873061
12:51:21 opendevreview sean mooney proposed openstack/nova master: introduce global greenpool https://review.opendev.org/c/openstack/nova/+/873061
13:34:59 dansmith gibi: can you re-+2 this? I forgot to update the log level in the test since you last saw: https://review.opendev.org/c/openstack/nova/+/891340
13:43:42 gibi dansmith: done
13:43:46 dansmith thanks
15:12:31 sean-k-mooney hum interesting https://review.opendev.org/c/openstack/nova/+/873061?tab=change-view-tab-header-zuul-results-summary
15:12:41 sean-k-mooney so the devstack jobs all passed
15:14:14 sean-k-mooney all the leaked greentrheads are in 3 moudles nova.tests.unit.virt.libvirt nova.tests.unit.test_service.TestWSGIService and nova.tests.unit.test_wsgi.TestWSGIServer
15:15:43 sean-k-mooney the functionaltests are similar so its proably that the relevent code that is leakign the greenthreads is shared.
15:50:33 gibi nova weekly meeting will start here in 10 minutes. I will be your host today and I try to do a quick meeting.
16:00:59 gibi #startmeeting nova
16:00:59 opendevmeet Meeting started Tue Aug 22 16:00:59 2023 UTC and is due to finish in 60 minutes. The chair is gibi. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:59 opendevmeet Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:59 opendevmeet The meeting name has been set to 'nova'
16:01:04 auniyal \o
16:01:05 gibi o/
16:01:10 sean-k-mooney o/
16:01:11 dansmith o/
16:01:11 gmann o/
16:01:18 dansmith when does bauzas get back? 2024?
16:01:25 sean-k-mooney i think monday
16:01:31 sean-k-mooney but i didnt check whcih year
16:01:39 dansmith "Monday, 2024"
16:02:23 gibi yepp 28th of Aug he is back
16:02:49 gibi so you will get rid of me being the chair after this meeting :)
16:03:16 gibi lets start
16:03:25 gibi #topic Bugs (stuck/critical)
16:03:32 gibi I don't see any critical bug
16:03:47 gibi #link https://bugs.launchpad.net/nova/+bugs?search=Search&field.status=New 43 new untriaged bugs (+3 since the last meeting)
16:03:52 gibi #info bug baton is on PTO (bauzas please pass it forward after the summer period)
16:04:09 gibi is there any bugs we need to discuss?
16:04:41 andrewbonney If anyone could assess https://bugs.launchpad.net/nova/+bug/2023035 it would be appreciated as it's blocking us upgrading to 2023.1 without using old config workarounds
16:05:16 opendevreview Merged openstack/nova master: Log excessive lazy-loading behavior https://review.opendev.org/c/openstack/nova/+/891340
16:05:44 dansmith \o/
16:06:39 gibi andrewbonney: I see you know about the workaround
16:07:16 andrewbonney Yeah, just doesn't feel ideal if it's easily patchable
16:07:20 gibi andrewbonney: I did not look deeper but it seems the one of the recent comment shows what is the difference between the two host cpus
16:07:40 dansmith isn't this a libvirt thing anyway?
16:08:05 gibi could be
16:08:51 dansmith I guess it says libvirt version is the same on both ends, so hmm
16:08:54 gibi `cpu_mode = host-model` so I guess it is difference between the model
16:09:03 dansmith maybe we could get kashyap to look at this
16:09:31 andrewbonney We'd be happy to provide any other debug if it's useful
16:09:31 sean-k-mooney there are many factor that can cause it to fail
16:10:19 sean-k-mooney do you have diffent kernel version on the diffent hosts
16:10:41 sean-k-mooney this kid of sound like perhaps some of the mitigations that are enabeld are not the same on both
16:10:50 andrewbonney The testing we did found that the running XML for a guest failed in the virsh hypervisor-cpu-compare on the same host it was running on
16:11:08 andrewbonney Where cpu-compare worked
16:11:57 sean-k-mooney and do you have the patchs that we mreged recently to update how we did the cpu compatiblity checks
16:12:27 andrewbonney This was using b9089ac from 2023.1
16:12:27 sean-k-mooney namely https://review.opendev.org/c/openstack/nova/+/869950
16:12:44 andrewbonney Yes, we think that caused it
16:12:47 sean-k-mooney ok that should be in that then
16:13:24 gibi OK. I will ask kashyap to look at it when he is back
16:13:25 sean-k-mooney well that is usign the new api
16:13:37 sean-k-mooney that the libvirt folks told use to use
16:13:41 dansmith andrewbonney: is it simple to just revert that patch from what you have to see?
16:14:12 andrewbonney Yes I could do that tomorrow and update the bug, but given what we tested with virsh command line I think it proves it already
16:14:34 dansmith but yeah, we probably need to dig deeper if this is the new api we're supposed to be using and it's not doing what we want.. not sure if that means we need to do something different or not
16:14:41 andrewbonney It felt like the guest XML that was being put into the new API was wrong, but I don't know libvirt enough to be sure
16:14:42 dansmith but kashyap would probably be a good person to chase that down
16:14:57 andrewbonney Thanks. Don't want to take up all your meeting time :)
16:15:04 gibi OK, moving on
16:15:12 gibi any other bug to discuss?
16:16:05 gibi #topic Gate status
16:16:22 dansmith definitely improving
16:16:40 gibi glad to hear that
16:16:50 dansmith one of my patches merged today with a single +W and no rechecks
16:16:54 dansmith which was pretty shocking :)
16:17:30 gibi we also merged one functional improvement https://review.opendev.org/c/openstack/nova/+/892126
16:17:39 gibi hopefully that will help too
16:17:48 dansmith yeah, which sounds like may have more impact than it appears, which is cool
16:18:07 dansmith I'll be looking for functional failures in the gate queue once that has soaked for a while
16:18:11 gibi I will check the functional results later this week to see if the common issues are disappeared or not
16:18:16 gibi yeah
16:18:17 gmann yeah, hopefully it does not go worst during release time, current gate stability is much better
16:18:59 dansmith yep, and if we can get some +Ws on these, they'll further reduce some of our runtime loading: https://review.opendev.org/q/topic:reduce-lazy-loads
16:19:22 dansmith during periods of high load, it can take a bit to execute a lazy load, and it's all unnecessary
16:20:25 dansmith but yeah, summary is we're in a much better place than we were.. not enough to consider it done, but enough to not be very fearful of release time I think :)
16:20:27 gibi yepp, those are easy wins
16:20:54 gibi #info Please look at the gate failures and file a bug report with the gate-failure tag.
16:20:59 sean-k-mooney i can try and take a look at them
16:21:03 gibi moving on
16:21:14 gibi #topic Release Planning
16:21:36 gibi we have 2 weeks before Feature Freeze
16:22:05 gibi I expect bauzas will start a feature tracking pad as soon as he is back
16:22:28 gibi any features we need to discuss here?
16:24:03 gibi then moving on
16:24:14 gibi #topic Stable Branches
16:24:29 gibi elodilles_pto is off
16:25:02 gibi is there any stable topic to discuss?
16:26:23 gibi #topic Open discussion
16:26:28 gibi nothing on the agenda
16:26:41 opendevreview Spencer Harmon proposed openstack/nova master: libvirt tsc_frequency https://review.opendev.org/c/openstack/nova/+/891924
16:26:42 gibi any other topics to bring up?
16:27:09 kgube Yes
16:27:18 gibi go ahead
16:27:27 kgube I sould have mentioned this with the features maybe

Earlier   Later