| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-07-31 | |||
| 13:52:04 | bauzas | context being ^ | |
| 13:52:36 | gibi_ | cdent: with Jay's https://review.openstack.org/#/c/488510 and resource_tracker on drugs I can reproduce problems with resize in a stable way with https://review.openstack.org/#/c/487958 | |
| 13:53:03 | openstackgerrit | Merged openstack/python-novaclient master: Update reno for stable/pike https://review.openstack.org/488799 | |
| 13:53:04 | cdent | gibi_: excellent, looking | |
| 13:53:26 | gibi_ | cdent: it seems resource clean up is missing after both confirm and revert | |
| 13:54:27 | cdent | nice work | |
| 13:55:25 | cdent | good timing too, we can noodle about in the scheduler meeting | |
| 13:57:54 | openstackgerrit | Merged openstack/nova master: hyperv: stop serial console workers while deleting vm files https://review.openstack.org/387831 | |
| 13:59:36 | edleafe | cdent: gibi_: just added it to my local agenda | |
| 14:00:11 | gibi_ | edleafe: cool, I will join to the meeting | |
| 14:00:35 | edleafe | Scheduler meeting starting now! | |
| 14:04:09 | openstackgerrit | Merged openstack/nova master: Stop using mox stubs in test_console_auth_tokens.py https://review.openstack.org/477374 | |
| 14:04:39 | openstackgerrit | Merged openstack/nova master: Add oslo_concurrency=INFO to default log levels for nova-manage https://review.openstack.org/487179 | |
| 14:04:45 | openstack | Launchpad bug 1707319 in OpenStack Compute (nova) "Security group doesn't apply to existing port" [Undecided,In progress] - Assigned to hongbin (hongbin034) | |
| 14:04:45 | bauzas | mriedem: looks a pretty critic bug to me https://bugs.launchpad.net/nova/+bug/1707319 | |
| 14:04:57 | bauzas | mriedem: critical* even | |
| 14:05:26 | bauzas | mriedem: I'm trying to get when we regressed (if we regressed) | |
| 14:05:49 | mriedem | hongbin pinged me about it on saturday, i haven't looked | |
| 14:07:30 | bauzas | mriedem: the problem is when you pass an existing NIC | |
| 14:07:40 | bauzas | mriedem: we totally forget to check the SGs if so | |
| 14:07:55 | bauzas | mriedem: we only do that when you ask nova to create a port | |
| 14:15:39 | bauzas | mriedem: thinking out loud, are security groups a nova thing or a neutron thing ? | |
| 14:15:50 | bauzas | mriedem: looks like we proxied that to neutron, nope ? | |
| 14:17:36 | mriedem | we proxy when creating the port | |
| 14:19:06 | bauzas | mriedem: nevermind, got my answer https://docs.openstack.org/security-guide/networking/services-security-best-practices.html | |
| 14:19:11 | bauzas | mriedem: will invalid the bug | |
| 14:20:54 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Test resize with placement api https://review.openstack.org/487958 | |
| 14:21:27 | gibi_ | jaypipes: just added more printout to see the whole resource state when the test fails. The commit message no links to test results with that extra printouts . ^^ | |
| 14:22:27 | mriedem | bauzas: can you clarify? | |
| 14:22:44 | mriedem | https://docs.openstack.org/security-guide/networking/services-security-best-practices.html#security-groups just says nova proxies security group information to neutron | |
| 14:22:57 | kashyap | smcginnis: Hey, thanks for looking out for Jenkins for this -- https://review.openstack.org/#/c/489198. Or do you do it by a script? :-) | |
| 14:23:33 | bauzas | mriedem: I just made a comment | |
| 14:23:34 | smcginnis | kashyap: Hah, nope. I took a quick look and saw it had a timeout failure. | |
| 14:24:06 | gibi_ | jaypipes: s/no/now/ | |
| 14:24:13 | kashyap | smcginnis: Ah, thank you. | |
| 14:24:24 | smcginnis | kashyap: No problem. ;) | |
| 14:24:25 | bauzas | mriedem: tl;dr: since the port was created beforehand directly to Neutron, you should define the SG policies at that moment | |
| 14:25:02 | jaypipes | gibi_: ty sir | |
| 14:26:03 | mriedem | bauzas: did you check the code? are you sure we don't attempt to update the port? | |
| 14:26:04 | jaypipes | dansmith, cdent, gibi_: I'm wondering if we have any tests that use the libvirt driver and not the fake virt driver for resize/move operations? | |
| 14:26:15 | mriedem | because i know we update the port, i'm just not sure what we put in the body for the PUT | |
| 14:26:29 | dansmith | jaypipes: eh? don't we actually do moves in tempest? | |
| 14:26:37 | cdent | there are some moves in tempest | |
| 14:26:54 | mriedem | yes we have resize, cold migration, shelve/unshelve and live migration in tempest | |
| 14:27:14 | mriedem | resize is tested single node in single node jobs, otherwise those are all tested on multinode jobs | |
| 14:27:14 | bauzas | mriedem: nope, we don't update the port | |
| 14:27:14 | jaypipes | do we see the same issue with placement not being properly updated in those tempest tests? | |
| 14:27:16 | dansmith | maybe jaypipes means unit/functional tests? | |
| 14:27:21 | bauzas | mriedem: I checked the code | |
| 14:27:25 | bauzas | mriedem: we just ignore that | |
| 14:27:40 | gibi_ | jaypipes: notification sample test are doing resize with FakeDriver | |
| 14:27:41 | bauzas | mriedem: and that's at least existing from stable/newton | |
| 14:27:50 | bauzas | it's not a regression at all | |
| 14:27:51 | dansmith | jaypipes: I was looking at your patch and I think we probably need a debug statement when we do the remove so we know it's happening | |
| 14:28:07 | dansmith | i.e. so we can check it in the tempest run logs | |
| 14:28:08 | jaypipes | dansmith: totes | |
| 14:28:17 | bauzas | mriedem: so we *could* update the port | |
| 14:28:33 | gibi_ | jaypipes: also the server_group test has VM moves like migrate and evacuate | |
| 14:28:38 | bauzas | mriedem: but that's just two roundtrips IMHO | |
| 14:28:59 | gibi_ | jaypipes: e.g. https://github.com/openstack/nova/blob/master/nova/tests/functional/test_server_group.py#L367 | |
| 14:29:04 | cdent | jaypipes, dansmith, gibi_: what we don't have in those tests is any confirmation of the state of allocations | |
| 14:29:11 | bauzas | mriedem: since nova has SGs per instances, while neutron has per ports, it's a waste of HTTP calls IMHO | |
| 14:29:17 | cdent | and since we're probably not running at capacity it doesn't become an issue | |
| 14:30:08 | gibi_ | cdent: yes, I think the only test where we assert on capacity is the new one we are troubleshooting now | |
| 14:30:23 | dansmith | jaypipes: I don't see any "expected to find" logs in the multinode run, which would mean we were bailing out of that un-doubling | |
| 14:31:03 | mriedem | gibi_: i think cdent is talking about tempest dsvm runs | |
| 14:31:09 | mriedem | where we aren't using fake drivers | |
| 14:32:06 | dansmith | we wouldn't confirm allocations in a tempest job anyway | |
| 14:32:09 | gibi_ | ahh, then please disregard my comment | |
| 14:32:10 | openstackgerrit | Merged openstack/nova master: Remove redundant free_vcpus logging in _report_hypervisor_resource_view https://review.openstack.org/487216 | |
| 14:32:31 | jaypipes | I'm going to add more log statements to the patch. give me a few minutes | |
| 14:32:36 | openstackgerrit | Merged openstack/nova master: Remove unnecessary code https://review.openstack.org/488299 | |
| 14:32:40 | kashyap | lyarwood: I know you gave +2, but do you prefer me to remove the 'Conflicts' note in the upstream backport? | |
| 14:33:06 | lyarwood | kashyap: no it's fine, that's more of a supernit tbh | |
| 14:33:47 | kashyap | lyarwood: Yeah, no problem. I aim to please, so thought I'd check :-) | |
| 14:38:33 | jaypipes | cdent: gimme a few. pushing a new rev with more log statements. | |
| 14:38:47 | cdent | roger | |
| 14:38:58 | mdbooth | sdague: A while back mriedem asked for a functional test for https://review.openstack.org/#/c/462521/ . Happy to oblige, but I'm not so familiar with what infra we already have in functional. Any pointers? | |
| 14:41:50 | mriedem | mdbooth: gibi has a 2 node resize functional test here https://review.openstack.org/#/c/487958/4 | |
| 14:42:03 | mdbooth | mriedem: Looking, thanks. | |
| 14:42:14 | bauzas | kudos to sdague | |
| 14:42:39 | bauzas | I looked today at launchpad and for the first time of my career, saw 0 new bugs | |
| 14:42:52 | mriedem | mdbooth: this one fakes out a resource claim failure on a compute https://github.com/openstack/nova/blob/master/nova/tests/functional/regressions/test_bug_1671648.py | |
| 14:43:10 | bauzas | I'm pretty impressed by his ability to triage bugs so fast :) | |
| 14:43:17 | bauzas | sdague: <3 | |
| 14:44:43 | mdbooth | mriedem: Should be able to find a good starting point there. Thanks! | |
| 14:46:12 | openstackgerrit | Jay Pipes proposed openstack/nova master: remove source provider allocs in confirm_resize() https://review.openstack.org/488510 | |
| 14:46:12 | openstackgerrit | Jay Pipes proposed openstack/nova master: placement: don't allocate on compute nodes https://review.openstack.org/488595 | |
| 14:46:25 | jaypipes | cdent, gibi, dansmith, edleafe: ok, have at it. ^ | |
| 14:46:32 | cdent | ✔ | |
| 14:47:19 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Test resize with placement api https://review.openstack.org/487958 | |
| 14:52:16 | gibi_ | here is the full debug log with jaypipes new patch set http://paste.openstack.org/show/617024/ | |
| 14:52:59 | gibi_ | my copy paste buffer is too small | |
| 14:53:04 | gibi_ | disregard the above link | |
| 14:53:22 | jaypipes | gibi_: :) | |
| 14:53:27 | jaypipes | gibi_: I was gonna say... | |
| 14:54:38 | gibi_ | is there a limit on paste.openstack.org about the size of the log? | |
| 14:55:35 | clarkb | gibi_: yes its like 1MB or something. If you have to paste very large amounts of text I think gist.github.com allows large pastes anonymously | |
| 14:55:46 | gibi_ | clarkb: thanks | |
| 14:56:00 | gibi_ | it seems pastebin doesn't have such limit either | |