Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-08-09
13:28:53 mriedem johnthetubaguy: ^?
13:32:14 mriedem another queens backport https://review.opendev.org/#/c/672161/2
13:35:08 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Fix type error on call to mount device https://review.opendev.org/669629
13:36:11 mriedem and https://review.opendev.org/#/c/669629/
13:38:03 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/659247
13:38:28 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/659247
13:46:24 finucannot aspiers: So apparently I can't access lxml documentation because it's Friday and that's not allowed (try navigating to lxml.de)
13:48:26 mordred the code is self-documenting?
13:48:46 finucannot I guess it _has_ to be
13:49:55 mriedem you can get to https://lxml.de/tutorial.html
13:50:21 mriedem https://lxml.de/api/index.html
13:50:29 finucannot Ah, sweet
13:50:32 finucannot mriedem++
13:50:33 mriedem that'll be $5 please
13:51:48 finucannot Darn, I'm fresh out of dollar bills. Crying shame, that
13:52:34 mriedem i'll take 5 eu
13:53:00 finucannot Britain refuses to take even one
13:53:21 mriedem boris has it under control
13:54:14 mriedem polite muttering i'm sure
14:04:33 aspiers mriedem: no swear-words in this channel please
14:05:29 frickler mriedem: I commented on https://bugs.launchpad.net/nova/+bug/1669468 earlier with what I think would be the reason for about 50% failure rate, but I'm not sure at which place to fix it
14:05:29 openstack Launchpad bug 1669468 in OpenStack Compute (nova) "tempest.api.compute.servers.test_novnc.NoVNCConsoleTestJSON.test_novnc fails intermittently in neutron multinode nv job" [Medium,Confirmed]
14:15:02 melwitt frickler: I think your assessment sounds probable. I think I could find where to fix it, I've done a few changes to console settings in devstack in the past
14:16:10 mriedem sorry, on a call now but will take a look later
14:19:21 cdent good article belmoreira . I made a couple of suggested edits. It seems like "The provider_tree "deepcopy" should be re-evaluated" is the big takeaway. Based on my recent posting it _might_ make sense to add a __copy__ to provider tree
14:27:00 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/659247
14:38:25 openstackgerrit Merged openstack/nova stable/queens: Stabilize unshelve notification sample tests https://review.opendev.org/674636
14:45:22 mriedem need a stable core for this queens one as well https://review.opendev.org/#/c/666862/
14:47:44 dansmith mriedem: I can just +W that because the only conflict was a test yeah?
14:47:57 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Fix double word hacking test https://review.opendev.org/665790
14:48:07 dansmith (as in without looking very hard)
14:58:29 mriedem dansmith: yeah
15:03:47 finucannot aspiers, sean-k-mooney: We're touching an awful lot of the same code between the amd-sev-libvirt-support, image-metadata-prefiltering and cpu-resources specs. I forsee a lot of merge conflicts, but also a chance to share best practices
15:04:03 finucannot aspiers, sean-k-mooney: Especially around the scheduler pre-filter stuff. Lots of comments left on both series
15:08:52 mriedem dansmith: since you just went through these for rocky, if you have a minute https://review.opendev.org/#/q/topic:bug/1764556+branch:stable/queens
15:08:55 mriedem i'm trying to flush queens for the release
15:13:08 aspiers finucannot: I feel like libvirt/driver.py and test_driver.py are horrendously big and could do with splitting up somehow
15:13:22 melwitt otherwise I can do those, I meant to but was watching the rocky change land first. I should have pre-reviewed
15:14:06 finucannot aspiers: Aye, it regularly crashes the flake8 process in syntastic (Vim plugin). I don't even know where to start though
15:14:10 artom mriedem, you're going for a royal flush?
15:14:21 mriedem no, 4 of a kind
15:14:33 mriedem shoot the moon
15:14:52 artom groaning \o/
15:15:16 mriedem this is the current +2'ed list of queens changes https://review.opendev.org/#/q/status:open+project:openstack/nova+branch:stable/queens+label:Code-Review=2
15:15:18 mriedem if that helps
15:21:58 dansmith mriedem: that service too old set looked like a clean pick from the previous yeah?
15:22:28 dansmith makes sense I guess since most of the change is the added and then modified tests
15:23:34 mriedem yup no conflicts
15:29:37 belmoreira thanks cdent
15:32:19 mriedem melwitt: are you working on a patch for bug 1669468? otherwise i can throw one up quick to see if it explodes.
15:32:19 openstack bug 1669468 in OpenStack Compute (nova) "tempest.api.compute.servers.test_novnc.NoVNCConsoleTestJSON.test_novnc fails intermittently in neutron multinode nv job" [Medium,Confirmed] https://launchpad.net/bugs/1669468
15:33:04 melwitt mriedem: I was planning to but haven't looked yet. if you have something to try, go ahead
15:33:39 mriedem btw i've got most of the tests in https://review.opendev.org/#/c/675218/ fixed but there are 2 failing for some yet unknown reason https://review.opendev.org/#/c/675218/
15:34:33 melwitt ack
15:41:22 openstackgerrit Merged openstack/nova stable/queens: fix cellv2 delete_host https://review.opendev.org/672690
15:41:29 openstackgerrit Merged openstack/nova stable/queens: Add functional regression test for bug 1837955 https://review.opendev.org/673567
15:41:29 openstack bug 1837955 in OpenStack Compute (nova) queens "MaxRetriesExceeded sometime fails with messaging exception" [Medium,In progress] https://launchpad.net/bugs/1837955 - Assigned to Matt Riedemann (mriedem)
15:42:13 openstackgerrit Merged openstack/nova stable/queens: libvirt: flatten rbd images when unshelving an instance https://review.opendev.org/668123
15:46:05 mriedem frickler: melwitt: this might have been the regression https://review.opendev.org/#/c/672493/6
15:46:33 mriedem heh, and test_novnc failed on that change http://logs.openstack.org/93/672493/6/check/tempest-multinode-full/2e30174/testr_results.html.gz
15:46:42 mriedem but the tempest-multinode-full job is non-voting in devstack so no one noticed
15:47:22 melwitt ah, yeah looks like it
15:47:26 mriedem that explains the recent up-tick in this failing all of a sudden again
15:52:41 cdent ah, melwitt, we're around at the same time. I was tasked yesterday with checking with you (and others) about the urgency/priority of consumer types so we can determine if we need to push harder o nit
15:55:43 melwitt cdent: I wouldn't say it's "urgent" but definitely something important. the earliest we could integrate it in nova is U though
15:56:08 melwitt and I expect the integration to be pretty involved
16:03:09 openstackgerrit Merged openstack/nova stable/queens: Cleanup when hitting MaxRetriesExceeded from no host_available https://review.opendev.org/673576
16:05:17 finucannot mriedem: What's the difference between 'filter_by(foo == False)' and 'filter_by(foo == false())' in the context of SQLAlchemy?
16:05:45 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: doc: Fix a parameter of NotificationPublisher https://review.opendev.org/670226
16:07:11 mriedem my understanding is that false() can be implemented by underlying db translation layers, e.g. ibm_db_sa, to translate what false means, so if the boolean type is an int (0 or 1) rather than False/True
16:07:17 mriedem but, it's a question for zzzeek really
16:07:39 mriedem https://docs.sqlalchemy.org/en/13/core/sqlelement.html?highlight=false#sqlalchemy.sql.expression.false
16:07:51 mriedem "A backend which does not support true/false constants will render as an expression against 1 or 0:"
16:08:11 finucannot I was looking at that (linked to it in a review) but didn't get whether we needed it or not
16:08:24 finucannot I guess MySQL, Postgres and SQLite all support the constants
16:08:29 mriedem for mysql/pg i don't think it's a problem
16:08:30 mriedem yeah
16:08:42 finucannot fair. Much of a muchness so
16:08:46 mriedem https://www.postgresql.org/docs/9.1/datatype-boolean.html
16:11:31 mriedem how do i know? because older versions of db2, when i needed to care, used 0/1 for booleans
16:11:52 mriedem boolean booleans were for gen-x'ers and their ilk
16:12:48 finucannot flashbacks to writing embedded C with Acme crap proprietary compilers back in uni
16:15:02 mriedem but did you wear a uni in uni? https://photos.app.goo.gl/fu4qRMpbNGtbTDfU6
16:16:40 mriedem dansmith: i think https://bugs.launchpad.net/nova/+bug/1839621 is where you come in and say # in the password has to be encoded
16:16:40 openstack Launchpad bug 1839621 in OpenStack Compute (nova) "Inappropriate split of transport_url string" [Undecided,New]
16:17:39 finucannot Much to my dismay, no, I did not
16:22:56 cdent thanks melwitt, I got your response out of the logs. Had a power outage just then.
16:24:54 cdent melwitt: so from your perpective getting the funcitonality done in placement before U starts up would be helpful
16:27:36 melwitt cdent: I don't think it's critical, but I feel like the question would be better answered by someone like tssurya, who is planning to consume the changes
16:27:59 cdent melwitt: she was involved in the earlier discussions and was hoping for more input from you
16:29:05 melwitt heh. well, there are some quirks with counting quota usage from placement today which need consumer types to be resolved. but counting from placement is opt-in, and I'd be looking to operators to know how urgently they want the quirks resolved
16:30:08 melwitt and AFAIK only CERN is planning to opt-in to counting quota usage from placement, the majority of folks will be running the default I expect
16:31:01 melwitt so, I consider it as medium priority unless operators signal otherwise
16:31:39 cdent I think in that case we can let CERN drive the bus then, and if tssurya needs an assist from other placement folk we can work that out. if this never ending performance improvement train in placement ever stops there may be some cycles spare
16:31:45 openstackgerrit Stephen Finucane proposed openstack/nova master: Add 'Instance.get_vifs' helper method https://review.opendev.org/675665
16:32:23 melwitt cdent: sounds reasonable to me
16:34:14 dansmith mriedem: yep, piled on
16:37:09 melwitt cdent: to be clear, on my side my plan is to start working on integration whenever the placement side is ready. we can't do it this cycle, it's too late, but whenever it's ready I would write up a spec to consume it
16:38:31 cdent melwitt: one factor is that none of the regular active contributors to placement has any particular need for the feature, so it's good (and appropriate) that tssurya is willing and able to work on it
16:40:08 melwitt cdent: yeah, makes sense. just wanted to be clear that it wouldn't get dropped on the floor once it's done. I'd write up the spec and do the integration once ready, but that said it's medium priority in my mind unless CERN turns up the priority

Earlier   Later