Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-09-29
15:44:16 gmann tosky: yeah, there is no grenade job for ceph
15:45:12 tosky I would say than there is no reason for the W-1
15:45:15 tosky it could be added later
15:48:20 gmann lyarwood: we can go with three options: 1. keep the run_test.sh script like PS2 (https://review.opendev.org/#/c/742056/2) and later convert that script to ansible roles, 2. keep ceph as legacy job(or zuulv3 with run_test.sh script) and rest convert to zuulv3, 3. keep legacy job and convert it completely along with ceph storage. which one you prefer ?
15:48:43 gmann tosky: it lose the coverage of ceph storage which is tested with legacy job
15:49:52 tosky gmann: shouldn't https://review.opendev.org/752557 get W-1 then, instead of https://review.opendev.org/#/c/742056/ ?
15:50:26 gmann tosky: both.
15:51:15 gmann current legacy job (grenade and live migration) run the tests on ceph too
15:51:42 gmann with runtime setup of ceph on multi nodes
15:54:15 tosky oh
15:54:29 tosky without devstack-plugin-ceph then?
15:54:33 tosky I guess it's a thing from the past
16:05:39 gmann no, by using devstack-plugin-ceph with cloning and running the setup from there.
16:06:03 gmann but we can use devstack-plugin-ceph with separate ceph job as added as TODO in those patches
16:36:13 openstackgerrit Merged openstack/python-novaclient master: Remove the unused coding style modules https://review.opendev.org/754961
17:14:56 lyarwood gmann / tosky ; tbh I think we can leave ceph as a TODO for the W cycle
17:15:23 lyarwood gmann / tosky ; for both grenade and LM as it's going to take lots of work in the plugin to get multinode jobs working
17:15:48 lyarwood gmann / tosky ; we previously hacked ceph support into the legacy jobs, I'd like to do it correctly this time
17:15:58 gmann lyarwood: and keep running ceph as part of legacy job?
17:16:05 gmann for victoria
17:16:43 lyarwood gmann: well I wasn't going to backport things to stable/victoria now unless we had to for this tbh
17:17:18 lyarwood gmann: if we are then we can backport the eventual multinode ceph stuff as well
17:17:41 lyarwood gmann: but I really can't think of a ceph specific bug caught by either job in the last few cycles FWIW
17:18:08 gmann lyarwood: ohk, as victoria branch is cut now so anyways these patches are going in W.
17:20:54 gmann dansmith: are you ok for ceph as TODO in W cycle ? Victoria gate will be no change and keep ceph testing in legacy job same way.
17:22:35 lyarwood to be clear, it's the live migration, grenade and evacuation tests on ceph that are going to be a TODO early in this cycle
17:22:47 gmann yeah.
17:22:55 lyarwood we just need a multinode job in the plugin itself and everything else is trivial tbh
17:23:11 tosky but legacy is going to break
17:23:19 gmann lyarwood: cinder has multinode ceph job but not grenade
17:23:25 tosky as devstack-gate won't support victoria
17:24:15 lyarwood gmann: if there's already a job in cinder then we can just use that, I couldn't see one in the plugin itself
17:24:28 lyarwood tosky: how so? stable/victoria is already cut
17:24:41 lyarwood tosky: are we really changing things in CI after RC?
17:24:43 tosky lyarwood: devstack-gate is branchless
17:24:48 lyarwood ah crap
17:25:00 tosky and it needs explicit support of each new release
17:25:01 gmann lyarwood: let me try those but as separate patch. i am +A on your patch and removed -W from mine
17:25:13 tosky that was the point of the legacy removal :)
17:25:28 lyarwood okay well it looks like we are backporting all of this to stable/victoria after all
17:25:59 gmann tosky: lyarwood i do not think openstack is 100% on zuulv3 and we can remove d-g support for victoria
17:26:09 tosky gmann: we are almost there
17:26:09 dansmith gmann: sorry just catching up
17:26:17 tosky and right now nova risk being the blocker
17:26:28 tosky check my work in progress etherpad
17:26:29 dansmith gmann: I think that losing the live migration on ceph coverage for really any amount of time, is problematic
17:26:35 gmann i expect d-g to be removed in W or X.
17:27:08 gmann tosky: yeah but we can remove dog support if it is 100% done and no dependency at all.
17:27:19 tosky isn't the ceph job a matter of including the devstack-plugin-ceph plugin and delegating the configuration to it?
17:27:21 gmann tosky: and do not forget the 3rd party CI/CD.
17:27:38 tosky gmann: 3rd party CI/CD have been told about that
17:27:45 lyarwood tosky: no, we need a multinode / multi compute job for these tests
17:27:57 lyarwood tosky: the standard job is just single host
17:28:28 lyarwood cinder-plugin-ceph-tempest-mn-aa looks like it's multinode but I don't think it's deploying two computes
17:28:52 tosky lyarwood: I mean, if you get the standard job, derive a the new job from it, change the nodeset and add devstack-plugin-ceph, what else will be missing?
17:29:05 dansmith the post-config change right?
17:29:19 lyarwood I'm pretty sure the plugin doesn't import the ceph keyring
17:29:27 lyarwood the old LM job would scp things around iirc
17:29:33 lyarwood we could hack something up quickly for that
17:29:44 lyarwood I just wanted to write a role for it etc
17:29:47 lyarwood into the plugin itself
17:30:21 tosky just as a general picture regarding the zuul v3 migration: right now, there are just:
17:30:41 dansmith I'm fully in support of the v3 conversion, and cleaning this all up, but it seems a little foolish to drop coverage of the most common deployment mechanism's live-migration support because of it,
17:31:08 dansmith even if it hasn't caught anything lately.. Murphy's probably due for an appearance if anything :)
17:31:17 tosky a non-voting cinder job (being worked on by me); a heat job (there is a WIP patch); a murano-apps and a murano-dashboard jobs (apparently abandoned)
17:31:40 tosky 2 neutron jobs (being worked on, almost ready); the nova jobs
17:31:47 tosky and that's basically it for the openstack/ namespace
17:31:49 tosky so...
17:31:56 openstackgerrit Ghanshyam Mann proposed openstack/nova master: Add nova-grenade-multinode-ceph job https://review.opendev.org/755092
17:32:03 tosky not that I want to push for it, but that's what I'm trying to do :)
17:32:07 lyarwood okay well add these ceph jobs to the list
17:32:47 lyarwood gmann: lets take https://review.opendev.org/#/c/752557/ out of the gate and I'll work on the ceph part tonight
17:32:47 gmann lyarwood: dansmith ^^ let's try. but yes we might need to setup the keyring things on subnode
17:33:05 gmann lyarwood: ok
17:36:18 dansmith gmann: I hope it's that easy!
17:39:43 openstackgerrit Ghanshyam Mann proposed openstack/nova master: Add nova-grenade-multinode-ceph job https://review.opendev.org/755092
19:06:49 openstackgerrit Stephen Finucane proposed openstack/nova-specs master: Add modernize-os-hypervisors-api spec https://review.opendev.org/755109
#openstack-nova - 2020-09-30
03:05:15 openstackgerrit wu.shiming proposed openstack/os-vif master: Fix hacking min version to 3.0.1 https://review.opendev.org/755168
08:11:42 ralonsoh hi folks, do you know what is this message? https://a5e1b7092bd931de9d7c-99461a827f8c9e81159099d5f417814c.ssl.cf1.rackcdn.com/738163/21/check/neutron-ovn-tempest-full-multinode-ovs-master/462d24b/compute1/logs/screen-n-cpu.txt
08:12:06 ralonsoh --> http://paste.openstack.org/show/798558/
08:12:29 ralonsoh sean-k-mooney, stephenfin ^^
08:12:34 ralonsoh if you know that
08:12:39 lyarwood bauzas: https://review.opendev.org/#/c/752486/ & https://review.opendev.org/#/c/749233/ & https://review.opendev.org/#/c/729178/ & https://review.opendev.org/#/c/744552/ - would you mind taking a look at these stable backports today if you have time?
08:24:14 bauzas lyarwood: rushing to finish my talk recording this morning, hopefully would be able to review later today
08:24:25 bauzas 7 mins left for recording
08:24:49 bauzas and I can't count the number of rushes i already had
08:26:04 lyarwood bauzas: ack np good luck!
08:37:51 frickler ralonsoh: that looks like https://bugs.launchpad.net/nova/+bug/1863021 but I assumed that that bug had long been fixed everywhere
08:37:51 openstack Launchpad bug 1863021 in OpenStack Object Storage (swift) "[SRU] eventlet monkey patch results in assert len(_active) == 1 AssertionError" [Undecided,In progress] - Assigned to Chris MacNaughton (chris.macnaughton)
08:39:15 frickler maybe not in ovs?
08:39:22 stephenfin ralonsoh: frickler: Yeah, that should have been fixed with github.com/openstack/nova/commit/2078ef850c9
08:39:55 stephenfin although that's using a subprocess so maybe the fix is needed elsewhere
08:41:03 ralonsoh frickler, stephenfin thanks a lot
09:18:11 openstackgerrit Stephen Finucane proposed openstack/nova-specs master: Add modernize-os-hypervisors-api spec https://review.opendev.org/755109
09:20:06 lyarwood stephenfin: haha is that poop emoji supposed to be in there
09:20:16 stephenfin yes
09:20:22 lyarwood hahaha
09:37:13 tobias-urdin lyarwood: can i borrow your brain for 2 minutes? :) trying to troubleshoot an issue, am i assuming correctly that for a volume-backed instance being resized, and scheduled to another node than it resides on, _is_storage_shared_with() would return false since (source host != dest host) and remotefs.create_dir() would be executed over a SSH connection even if it's volume-backed, so the code
09:37:19 tobias-urdin path is not only for instances with ephemeral disks
09:37:22 tobias-urdin https://github.com/openstack/nova/blob/stable/train/nova/virt/libvirt/driver.py#L9860

Earlier   Later