| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-06-25 | |||
| 09:00:05 | opendevreview | Lee Yarwood proposed openstack/nova master: compute: Avoid calling detach with src connection_info during LM rollback https://review.opendev.org/c/openstack/nova/+/797725 | |
| 09:44:11 | opendevreview | Elod Illes proposed openstack/nova stable/ussuri: [CI] Fix gate by using zuulv3 live migration and grenade jobs https://review.opendev.org/c/openstack/nova/+/795432 | |
| 09:48:48 | elodilles | lyarwood: as the ussuri gate fixing supermegasquash did not pass I decided to update the commit message (maybe it's a bit easier to track what happened in it) | |
| 09:49:00 | opendevreview | Jorhson Deng proposed openstack/nova master: process instance in soft-deleting task state during _init_instance https://review.opendev.org/c/openstack/nova/+/761264 | |
| 09:49:32 | elodilles | lyarwood: it's really just a commit message update. so if you will have time then please review it o:) | |
| 09:49:36 | elodilles | melwitt: ^^^ | |
| 10:08:43 | lyarwood | elodilles: ack will do, melwitt is out this week btw | |
| 10:08:44 | stephenfin | elodilles: mel is out all this week | |
| 10:08:49 | stephenfin | jinx | |
| 10:10:18 | stephenfin | gibi: Can we update the os-vif stable core team to include os-vif core now? We discussed during the team meeting a few weeks back and had no disagreements from what I can recall | |
| 10:10:34 | stephenfin | assuming you have the power to do so | |
| 10:14:05 | gibi | stephenfin: let me see | |
| 10:15:26 | elodilles | lyarwood stephenfin : ok, thanks for the info :) | |
| 10:20:16 | gibi | stephenfin: I did not find the meeting where we discussed that change | |
| 10:22:28 | stephenfin | hmm, maybe it was here so | |
| 10:22:36 | gibi | and also I dont see a os-vif stable group in gerrit | |
| 10:22:38 | stephenfin | we started discussing it here https://meetings.opendev.org/irclogs/%23openstack-nova/%23openstack-nova.2021-06-02.log.html#t2021-06-02T12:42:10 | |
| 10:23:41 | stephenfin | we also discussed it here https://meetings.opendev.org/irclogs/%23openstack-nova/%23openstack-nova.2021-06-14.log.html#t2021-06-14T11:09:39 | |
| 10:23:53 | stephenfin | now did we discuss it in the interim :) | |
| 10:24:13 | gibi | OK so it is a bit more complicated | |
| 10:24:30 | gibi | as there is no stable group for os-vif in gerrit | |
| 10:25:00 | gibi | or I dont see it | |
| 10:25:27 | gibi | elodilles: do you know who has approve rights now on os-vif stable? | |
| 10:27:13 | stephenfin | gibi: looks like it's nova-stable-maint https://github.com/openstack/project-config/blob/master/gerrit/acls/openstack/os-vif.config | |
| 10:27:31 | stephenfin | plus stable-maint-core | |
| 10:27:44 | stephenfin | so if we wanted to do this, we'd simply add os-vif-core to that list | |
| 10:27:55 | stephenfin | gibi: will I post to openstack-discuss about this, or do you want to? | |
| 10:30:06 | gibi | stephenfin: go ahead and post this on ML and then I reply that we should do it | |
| 10:30:45 | gibi | can we simply propose the acl change in gerrit review against the linked project config? | |
| 10:31:46 | stephenfin | I think so, yes | |
| 10:32:21 | opendevreview | Lee Yarwood proposed openstack/nova master: DNM: Test libvirtd with increased keepalive_interval https://review.opendev.org/c/openstack/nova/+/798069 | |
| 10:32:43 | gibi | stephenfin: cool, then could you propose the patch too/ | |
| 10:32:44 | gibi | ? | |
| 10:33:00 | stephenfin | make sense. Will do | |
| 10:34:04 | gibi | thanks | |
| 10:38:39 | lyarwood | gibi: ^ the LM with volumes attached failures continue even after the os-vif workaround in devstack, I'm going to see if playing around with the keepalive_interval helps in libvirtd | |
| 10:46:49 | lyarwood | gibi / stephenfin ; https://review.opendev.org/q/topic:nova-centos-8-stream - could I get some reviews on this series, tl;dr we are trying to use more upstream zuul infra downstream and given centos 8 stream is a supported base OS we wanted to get it covered in the upstream gate | |
| 10:47:10 | stephenfin | sure, I'll look shortly | |
| 10:51:50 | lyarwood | ta | |
| 12:10:05 | opendevreview | Merged openstack/nova master: db: Use module-level imports for sqlalchemy https://review.opendev.org/c/openstack/nova/+/786295 | |
| 12:20:37 | gibi | lyarwood: ack | |
| 12:20:56 | gibi | lyarwood: did we merged the os-vif WA? | |
| 12:22:32 | sean-k-mooney | gibi: i think so. | |
| 12:22:59 | sean-k-mooney | gibi: yes https://review.opendev.org/c/openstack/devstack/+/796826 | |
| 12:23:02 | sean-k-mooney | that is merged | |
| 12:23:18 | sean-k-mooney | gibi: is it helping | |
| 12:23:40 | sean-k-mooney | thats only on master actully we could backport that | |
| 12:24:18 | sean-k-mooney | gibi: i spoke to otherwiseguy about this last night and they have repoduced the issue and a plocibel fix | |
| 12:24:47 | sean-k-mooney | gibi: https://gist.github.com/otherwiseguy/af8f6ede8a09ccf5cef259c93bb00792 | |
| 12:25:12 | sean-k-mooney | that evenlet.sleep will fix it https://gist.github.com/otherwiseguy/af8f6ede8a09ccf5cef259c93bb00792#file-test_eventlet_reconnect-py-L17 | |
| 12:25:58 | gibi | ack | |
| 12:27:18 | gibi | lyarwood: does the os-vif WA helped decreasing the LM issue but not fully avoidet it? | |
| 12:27:28 | gibi | decreasing the frequency | |
| 12:28:56 | sean-k-mooney | gibi: i think lyarwood is a way for a while just an fyi but he will see that when he gets back | |
| 12:31:44 | gibi | sean-k-mooney: thanks | |
| 12:44:42 | opendevreview | Luigi Toscano proposed openstack/nova stable/ussuri: Remove broken legacy zuul jobs https://review.opendev.org/c/openstack/nova/+/795374 | |
| 12:47:04 | tosky | now that https://review.opendev.org/c/openstack/nova/+/795432/ seems to pass (just need a new +2!), I suspect 795374 may work too | |
| 12:57:08 | opendevreview | Merged openstack/nova master: db: Fold in indexes https://review.opendev.org/c/openstack/nova/+/786296 | |
| 13:08:59 | ganso | lyarwood, elodilles: Could you please re-workflow https://review.opendev.org/c/openstack/nova/+/795432 now that the CI has passed? | |
| 13:23:09 | elodilles | ganso: I'll let lyarwood approve that, as I was who updated the commit message | |
| 14:54:45 | gibi | I hate that I can only choose between a) duplicate the qos functional test to cover pps, b) reimplement a lot of production logic in the test to have dynamic asserts that works both for the old resource request format and the new one. | |
| 14:56:45 | sean-k-mooney | c) dont have tests and just assume its broken? | |
| 14:57:01 | sean-k-mooney | cause that my default when a or b is not done :) | |
| 14:57:42 | gibi | it is in the state of broken now, as I have a buch of TODOs to fix before the new resource request format and logic start fully working. So I've already achieved c) :D | |
| 14:58:35 | gibi | and with simple changes in the production logic I can make some scenarios work, but when I want to add test coverage to it I end up hating any solution I come up with | |
| 14:58:55 | sean-k-mooney | can you point me at the test quickly | |
| 14:59:08 | sean-k-mooney | im going to finsih shortly but maybe something will occure to me | |
| 15:00:01 | sean-k-mooney | this one https://review.opendev.org/c/openstack/nova/+/787205 | |
| 15:00:26 | sean-k-mooney | oh you have them spread over multiple patches | |
| 15:03:04 | sean-k-mooney | o/ | |
| 15:03:05 | gibi | sean-k-mooney: I can make https://review.opendev.org/c/openstack/nova/+/792394/9/nova/tests/functional/test_servers.py#6770 scenario to work locally with some modification in nova's neutron client code. But when I want to make that test actually pass I have to blow up the original test case to modify the asserts | |
| 15:03:12 | gibi | sean-k-mooney: have a nice weekend | |
| 15:04:15 | sean-k-mooney | ah ya i was looking at that | |
| 15:04:31 | sean-k-mooney | but did not see really what it was doiing | |
| 15:04:52 | sean-k-mooney | when you said neutron cleint code do you mean realy code or fixtures | |
| 15:05:10 | sean-k-mooney | i assume /nova/network/neutron.py | |
| 15:05:14 | gibi | that one ^^ | |
| 15:05:43 | sean-k-mooney | ok if it was just the fixture i woudl have jsut said make two | |
| 15:06:14 | gibi | I have multiple fixtures to support old and new resource request format | |
| 15:06:21 | gibi | that is a relatively small duplication | |
| 15:06:32 | gibi | https://review.opendev.org/c/openstack/nova/+/794306/2 | |
| 15:06:59 | gibi | and here https://review.opendev.org/c/openstack/nova/+/792394/9/nova/tests/fixtures/neutron.py | |
| 15:07:51 | gibi | the test logic, especially the the asserts needs to be changed as a) there will be more allocations b) the bindig profile allocation key changes format | |
| 15:09:01 | gibi | I could pull out all the asserts to functions and override the functions in the child test class according to the new needs | |
| 15:10:09 | gibi | it just probably needs a lot of refactoring, and a way to foresee how all assert will look like in the with the new resource request format in place | |
| 15:10:29 | gibi | to know what direction to refactor to | |
| 15:11:25 | sean-k-mooney | it might be simpler to just duplicate things | |
| 15:11:37 | sean-k-mooney | and if you have time later see if the commonality can be abstrated | |
| 15:12:03 | gibi | yeah that is one way to go | |
| 15:12:12 | sean-k-mooney | rather then trying to do both up front | |
| 15:12:45 | gibi | if i copy paste, then I need some reminder not to forget to abstract things out later | |
| 15:12:51 | gibi | as I will be lazy | |
| 15:13:14 | gibi | so not a reminder like a # TODO, but a reminder like a pain in the ass | |
| 15:13:58 | sean-k-mooney | have a test that emails you every time it runs :P | |
| 15:14:28 | gibi | :D | |
| 15:14:50 | sean-k-mooney | duplication when its not extream i think is ok | |
| 15:16:10 | gibi | yeah, in test I allow duplication if in return it helps seeing the test scenario cleanly | |
| 15:16:38 | gibi | I'm affraid there will be extreme copy past now to have all the server create scenarios and then all the server move scenarios covered | |
| 15:16:48 | gibi | ~1000 LOC | |
| 15:17:20 | gibi | anyhow you are right If I cannot predic which direction I should refactor then probably I need to duplicate first | |