| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-01-20 | |||
| 14:52:09 | sean-k-mooney | fyi ill be afk for a while but ill update the spec when i get back | |
| 14:52:16 | gibi | sean-k-mooney: ack | |
| 14:52:22 | sean-k-mooney | ah ok ya i guess using testtool makes sense there | |
| 14:53:14 | sean-k-mooney | oh needed to exted time frame so its the min bandwith tests | |
| 14:53:18 | sean-k-mooney | weird | |
| 14:53:44 | gibi | I see in the api log that fault is lazy loaded | |
| 14:53:48 | gibi | for this request | |
| 14:54:00 | gibi | so somehow the fault is missing in the db | |
| 15:12:49 | lyarwood | https://review.opendev.org/c/openstack/nova/+/754702/ - there's a first, subnode deployment failed but we still tried to run multinode tests | |
| 15:44:04 | kashyap | Unrelated ... TIL, Red Hat announced gratis RHEL (you can now run upto 16 production instances): https://www.redhat.com/en/blog/new-year-new-red-hat-enterprise-linux-programs-easier-ways-access-rhel | |
| 15:44:35 | kashyap | lyarwood: --^ Wonder if that allows one to start testing w/ proper RHEL in the upstream CI | |
| 15:45:56 | lyarwood | kashyap: I wouldn't expect it would no | |
| 15:46:24 | lyarwood | kashyap: and in any case testing RHEL in master CI seems a little pointless if we have centOS streams | |
| 15:46:45 | kashyap | lyarwood: Ah, right; if we're testing w/ CentOS | |
| 15:46:56 | kashyap | ... Streams, it's indeed not worth it | |
| 15:47:13 | kashyap | I brought it up because some upstreams (like QEMU) will start using it now | |
| 15:48:12 | lyarwood | I'd like to introduce *some* centOS streams coverage into the Nova gate if at all possible but that's about it | |
| 15:51:43 | kashyap | lyarwood: Right; any non-trivial enterprise distro testing is upto relevant downstreams (as it's part of their value prop). | |
| 15:52:30 | kashyap | I, for one, would be happy if a Fedora job that is sustained w/o bit-rot. I don't care about CentOS or Streams, personally. | |
| 15:52:48 | lyarwood | we have that in the experimental queue | |
| 15:53:14 | lyarwood | but centOS is a supported distro unlike Fedora | |
| 15:53:27 | lyarwood | so I'd rather have it in the check and gate queues | |
| 15:54:26 | kashyap | lyarwood: Yeah; I know you did some work on that; and I use it occasionally. To which, I thank you! | |
| 15:54:43 | lyarwood | a few of us normally try to support the new fedora releases | |
| 15:54:54 | kashyap | You're saying CentOS as "supported" as in, LTS, I take it | |
| 15:55:05 | kashyap | lyarwood: Right; I did it in the past, and didn't do it after that. | |
| 15:55:19 | kashyap | I also know ianw++ does some tireless maint work | |
| 15:55:23 | sean-k-mooney | lyarwood: fedora is techniclaly not in the PTI | |
| 15:55:48 | sean-k-mooney | if your talking about devstack then its only partly supported | |
| 15:56:37 | sean-k-mooney | https://github.com/openstack/devstack/blob/master/doc/source/overview.rst#base-os | |
| 15:58:00 | kashyap | lyarwood: BTW, upstream libvirt tests CentOS7, 8, Streams, and a crap-ton of other distros: https://gitlab.com/libvirt/libvirt/-/pipelines/244260219 | |
| 15:58:41 | kashyap | It's at least good that they (including QEMU) test on RHEL; to catch regressions that can sneak in through rebases. | |
| 15:58:45 | sean-k-mooney | kashyap: by the way the previous developer licencing explcitly disallowed rhel | |
| 15:58:49 | kashyap | (And upper layers benefit from that.) | |
| 15:59:05 | lyarwood | sean-k-mooney: huh I was going on the runtimes for each release | |
| 15:59:07 | kashyap | sean-k-mooney: Yep; that's fixed now. The developer subscription is now far more useful. | |
| 15:59:37 | sean-k-mooney | lyarwood: fedora is not in the support runtime for any release | |
| 15:59:52 | sean-k-mooney | right now centos stream is not either technically | |
| 16:00:15 | sean-k-mooney | although i think that will change in the y release | |
| 16:00:27 | sean-k-mooney | i know rdo are moving to stream now | |
| 16:00:41 | sean-k-mooney | but technially that is not a supproted runtime | |
| 16:00:53 | lyarwood | yup I'd still like to move ahead with some centOS coverage of some form this cycle if possible | |
| 16:01:54 | sean-k-mooney | yes we could do that though we proably shoudl not use stream | |
| 16:02:17 | sean-k-mooney | at least until the offical runtime list is either ammened or next cycle | |
| 16:02:26 | lyarwood | yeah I'll just use whatever is in the base job from devstack/tempest (/me forgets which) | |
| 16:02:39 | sean-k-mooney | lyarwood: well both are avaiable | |
| 16:02:49 | sean-k-mooney | and i fixed the only stream issue earlier in the cycle | |
| 16:03:03 | sean-k-mooney | so we can use both if needed | |
| 16:03:15 | sean-k-mooney | its just up to use to choose the correct nodeset | |
| 16:03:41 | sean-k-mooney | kashyap: the limit on 16 instance basically means we cant use the "No-cost RHEL for small production workloads" clause | |
| 16:04:08 | kashyap | sean-k-mooney: What do you mean can't use? It is for CI/developer purposes | |
| 16:04:15 | kashyap | That's their intentional limit | |
| 16:04:30 | kashyap | sean-k-mooney: Many upstreams benefit from that | |
| 16:04:46 | kashyap | It is not supposed to solve every use case of every project. But for vast majority; this is a good move. | |
| 16:04:58 | sean-k-mooney | that document does not explcitly state you arfree to use it for CI | |
| 16:06:13 | sean-k-mooney | it describe 2 catagories "No-cost RHEL for customer development teams" upstream ci is not a customer | |
| 16:07:30 | sean-k-mooney | the other one is "No-cost RHEL for small production workloads" | |
| 16:12:57 | sean-k-mooney | which is adderssed by "Individual Developer subscription for RHEL can be used in production for up to 16 systems." | |
| 16:54:58 | kashyap | The annoucement clearly spells out "community projects maintaining large CI/Build systems". | |
| 17:02:23 | dansmith | gibi: lyarwood: If you're around, could we get this landed? https://review.opendev.org/c/openstack/nova/+/771234 | |
| 17:02:41 | dansmith | we're currently copying glance-internal properties into snapshots, which we need to stop so glance can stop letting api users set internals :P | |
| 17:04:24 | openstackgerrit | Merged openstack/nova master: tox: Stop linting as part of docs target https://review.opendev.org/c/openstack/nova/+/735176 | |
| 17:04:57 | openstackgerrit | Merged openstack/nova master: doc: require openstack client change for every new API microversion https://review.opendev.org/c/openstack/nova/+/717727 | |
| 17:12:03 | lyarwood | dansmith: sorry was on a call, looking | |
| 17:16:14 | gibi | dansmith: looks good to me | |
| 17:17:01 | dansmith | gibi: thanks | |
| 17:17:20 | lyarwood | dansmith: did you want to backport this? | |
| 17:17:45 | dansmith | lyarwood: we don't really need to, as I don't think the glance part will get backported | |
| 17:17:48 | lyarwood | dansmith: if you do could you add a bug otherwise this LGTM | |
| 17:17:57 | lyarwood | dansmith: an ack then I'll +W | |
| 17:17:59 | lyarwood | ah* | |
| 17:17:59 | dansmith | thanks | |
| 17:18:26 | dansmith | if we backported the enforcement in glance, we'd start breaking nova if you yum updated glance before, so I don't think it's a good idea | |
| 17:19:37 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Replace blind retry with libvirt event waiting in detach https://review.opendev.org/c/openstack/nova/+/770246 | |
| 17:30:09 | openstackgerrit | Merged openstack/nova master: libvirt: Only add a USB controller if it's necessary https://review.opendev.org/c/openstack/nova/+/756549 | |
| 17:53:34 | openstackgerrit | Merged openstack/nova master: libvirt: Remove support for '[libvirt] use_usb_tablet' https://review.opendev.org/c/openstack/nova/+/756550 | |
| 17:54:12 | openstackgerrit | Merged openstack/nova master: Make PCI claim NUMA aware during live migration https://review.opendev.org/c/openstack/nova/+/748453 | |
| 17:55:12 | openstackgerrit | Merged openstack/nova master: api-ref: Add notes about volume attach and detach being async https://review.opendev.org/c/openstack/nova/+/768104 | |
| 17:55:47 | openstackgerrit | Merged openstack/nova master: Docs: correct cpu_thread_policy explanation https://review.opendev.org/c/openstack/nova/+/765172 | |
| 18:21:27 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: WIP: libvirt: start tracking NUMACell.socket for hosts https://review.opendev.org/c/openstack/nova/+/766816 | |
| 18:23:22 | stephenfin | lyarwood: This can wait til after M2, but you should probably look at https://review.opendev.org/c/openstack/python-openstackclient/+/771699 at some point | |
| 18:23:53 | lyarwood | stephenfin: ack | |
| 18:23:59 | stephenfin | Suggestions for how to make that less horrific for users are welcome, though I think that's just part of the complexity of BDM v2, unfortunately :-( | |
| 18:24:11 | lyarwood | stephenfin: speaking of M2 would you mind hitting https://review.opendev.org/c/openstack/nova-specs/+/771474 again | |
| 18:24:18 | stephenfin | sure | |
| 18:55:14 | openstackgerrit | Artom Lifshitz proposed openstack/os-traits master: WIP: Add COMPUTE_SOCKET_PCI_NUMA_AFFINITY trait https://review.opendev.org/c/openstack/os-traits/+/771705 | |
| 18:55:38 | artom | Oh we get notified in both channels eh | |
| 19:01:46 | sean-k-mooney | all the placment and os traits related stuff is now here | |
| 19:02:04 | artom | So is -placement ded? | |
| 19:02:11 | sean-k-mooney | kind of | |
| 19:02:16 | artom | zombie? | |
| 19:02:29 | sean-k-mooney | i mean i think most of us are still there too | |
| 19:02:46 | sean-k-mooney | but gibi is ptl of both currently and three really isnt much of a placment core team | |
| 19:02:53 | sean-k-mooney | at least that is active | |
| 19:32:01 | openstackgerrit | Lee Yarwood proposed openstack/nova master: docs: Add reference docs for internal block device structures https://review.opendev.org/c/openstack/nova/+/771711 | |
| 19:32:14 | lyarwood | stephenfin: ^ for the morning, spent some time this afternoon cleaning an old email from mdbooth up and turning it into a reference doc | |
| 20:21:05 | openstackgerrit | sean mooney proposed openstack/nova-specs master: libvirt: support vdpa kernel networking https://review.opendev.org/c/openstack/nova-specs/+/764999 | |
| 20:22:16 | dansmith | lyarwood: there's another failure in the multistore job I'm seeing evidence of, which actually might be caused by us snapshotting an image which includes glance internals, which prevents us from doing multistore things, thinking they're already done | |
| 20:22:33 | dansmith | lyarwood: which hopefully that do-not-inherit patch will fix | |
| 20:22:59 | dansmith | hadn't seen this one failure until just now | |