Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-02-19
19:20:35 efried phew
19:20:49 efried consider my chops busted
19:24:04 openstackgerrit Merged openstack/nova stable/ocata: libvirt: bandwidth param should be set in guest migrate https://review.openstack.org/519635
19:24:20 openstackgerrit Merged openstack/python-novaclient master: Fix the docstring for the update method https://review.openstack.org/545819
19:29:26 openstackgerrit Matt Riedemann proposed openstack/nova master: Combine error handling blocks in _do_build_and_run_instance https://review.openstack.org/545960
19:31:54 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: unquiesce instance on volume snapshot failure https://review.openstack.org/545961
20:03:26 openstackgerrit Matt Riedemann proposed openstack/nova stable/pike: unquiesce instance on volume snapshot failure https://review.openstack.org/545966
20:06:17 openstackgerrit Merged openstack/nova stable/ocata: Use proper user and tenant in the owner section of libvirt.xml. https://review.openstack.org/525997
20:08:14 mriedem dansmith: want to get this simple fixture cleanup patch? https://review.openstack.org/#/c/539758/
20:08:21 mriedem that will unblock a few other approved changes
20:08:50 mriedem this is the series where bfv failing during scheduling always orphans your volumes
20:08:52 mriedem which sucks
20:10:40 mriedem FYI in case any cores want to shed 1300+ LOC https://review.openstack.org/#/c/544698/
20:16:40 dansmith mriedem: ack
20:22:40 openstackgerrit Merged openstack/os-vif master: zuul: Enable functional tests in gate https://review.openstack.org/530961
20:22:46 openstackgerrit Merged openstack/nova master: unquiesce instance on volume snapshot failure https://review.openstack.org/519464
20:32:25 openstackgerrit Matt Riedemann proposed openstack/nova stable/ocata: unquiesce instance on volume snapshot failure https://review.openstack.org/545973
20:59:25 melwitt dansmith, mriedem: on https://review.openstack.org/#/c/544698, "We dropped support for aggregates in newton" just means dropped support for the old "main db" aggregates, not dropped support for aggregates altogether, right?
20:59:42 mriedem melwitt: we just deleted the API
20:59:47 mriedem it was an admin-only extension,
20:59:49 mriedem we said, fuck it
21:00:17 dansmith melwitt: heh, right, sorry. I meant we dropped support for this stuff I'm removing :D
21:00:28 dansmith melwitt: if you think it's important I can rev it
21:00:44 melwitt when I first read it I was like O.o
21:01:24 melwitt I'll just put a note to self on there
21:01:27 mriedem dropped the migration compat code
21:01:30 mriedem like for flavors
21:01:54 melwitt yeah. I see it in the patch, was just trying to connect the commit message in case there was something really major I was under a rock about
21:23:21 mriedem dansmith: after reading the bug for https://review.openstack.org/#/c/543970/ can you make sure my comment is correct before I +W?
21:30:33 dansmith mriedem: replied for posterity, but in short: yep.
21:31:50 hrw sean-k-mooney: will be
21:31:55 mriedem cool; i'll start the backport party
21:35:58 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Lazy-load instance attributes with read_deleted=yes https://review.openstack.org/545987
21:37:38 openstackgerrit Matt Riedemann proposed openstack/nova stable/pike: Lazy-load instance attributes with read_deleted=yes https://review.openstack.org/545988
21:38:25 openstackgerrit Matt Riedemann proposed openstack/nova stable/ocata: Lazy-load instance attributes with read_deleted=yes https://review.openstack.org/545989
21:43:51 hrw sean-k-mooney: added myself to nova etherpad
21:45:52 mriedem not sure why we'd even be lazy loading instance.system_metadata in that evacuate cleanup path, the db api should manually join it https://github.com/openstack/nova/blob/stable/pike/nova/db/sqlalchemy/api.py#L2141-L2143
21:50:28 openstackgerrit Matt Riedemann proposed openstack/nova stable/ocata: libvirt: Don't VIR_MIGRATE_NON_SHARED_INC without migrate_disks https://review.openstack.org/519636
21:51:14 openstackgerrit Merged openstack/nova master: Drop extra loop which modifies Cinder volume status https://review.openstack.org/539758
21:51:45 openstackgerrit Merged openstack/nova master: Store block device mappings in cell0 https://review.openstack.org/544748
21:52:13 openstackgerrit Merged openstack/nova master: Add functional tests to ensure BDM removal on delete https://review.openstack.org/544747
21:53:30 dansmith mriedem: the compute drop passed grenade this time
21:53:37 dansmith failed something else, which I'm looking at but..
21:54:15 dansmith hmm, rabbit crash maybe?
21:54:24 dansmith http://logs.openstack.org/80/543580/1/check/legacy-tempest-dsvm-cells/e4a862c/logs/screen-n-cpu.txt.gz?level=TRACE#_Feb_19_20_06_46_946074
21:56:05 mriedem there are crash reports in http://logs.openstack.org/80/543580/1/check/legacy-tempest-dsvm-cells/e4a862c/logs/rabbitmq/rabbit@ubuntu-xenial-inap-mtl01-0002617609-sasl.txt.gz
21:56:29 dansmith ah sweet
22:05:08 mriedem mmedvede: can we get the pkvm CI to not vote at all on stable/ocata changes? it always fails immediately https://review.openstack.org/#/c/545973/
22:06:47 efried edleafe, melwitt, mriedem, jaypipes, cdent: Specless bp to add ?required=<trait list> to GET /resource_providers ?
22:07:29 mriedem mmedvede: looks like the tests randomly timeout because it takes forever to spawn the actual guest
22:07:30 mriedem 2018-02-19 21:14:56.466 31787 INFO nova.compute.manager [req-21095306-b40d-4762-a083-8510184d386d tempest-ListServerFiltersTestJSON-391319940 tempest-ListServerFiltersTestJSON-391319940] [instance: 5b94a837-eb53-4275-b893-8e51b8d10518] Took 243.80 seconds to spawn the instance on the hypervisor.
22:07:53 efried Would "adding ?required=<trait list> to GET /resource_providers" be considered for a specless blueprint?
22:08:06 edleafe efried: yeah, I think specless is ok. Just clarify in the BP that it will modeled on GET /allocation_candidates
22:08:08 mriedem efried: no
22:08:10 mriedem it's an api change
22:08:12 melwitt efried: I'd have expected an api change to be a spec, albeit a small one in that case
22:08:22 efried ight
22:08:35 edleafe oh geez, yeah, an API change. What they said.
22:14:01 jaypipes efried: I thought all API changes required a spec?
22:14:20 efried jaypipes: Apparently that's true; see above.
22:43:26 openstackgerrit Merged openstack/nova master: Remove deprecated aggregate DB compatibility https://review.openstack.org/544698
22:54:44 openstackgerrit Merged openstack/nova master: Lazy-load instance attributes with read_deleted=yes https://review.openstack.org/543970
23:03:55 mriedem mdbooth: you should take a look at this https://review.openstack.org/#/c/542646/
23:19:05 tetsuro is avolkov around?
23:20:12 tetsuro I’d like you to answer the question in https://review.openstack.org/#/c/539865/. … or anyone who can help me.
23:47:44 mnaser is there any way i can do something to get output from nova-compute about memory usage
23:47:53 mnaser i have a nova-compute process sitting on 13gb of ram.
23:48:21 mnaser (before i restart it, just to get helpful information out)
23:49:30 melwitt mriedem: these comments aren't related to the patch, are they? just a suggestion on some better cleanup we could do in a separate patch? https://review.openstack.org/#/c/520158/3/nova/compute/api.py@2826
23:50:14 mnaser wow... actually almost every nova-compute process is using 13g of memory..
#openstack-nova - 2018-02-20
00:00:02 openstackgerrit Eric Fried proposed openstack/nova-specs master: Filter Resource Provider List for Traits https://review.openstack.org/546009
00:00:21 efried edleafe, mriedem, melwitt, jaypipes, cdent: ^
00:04:28 melwitt mnaser: the only thing I know of is the hypervisor-stats type of things e.g. https://docs.openstack.org/nova/pike/admin/common/nova-show-usage-statistics-for-hosts-instances.html
00:04:45 melwitt mnaser: but you may be able to use something like this to dig deeper https://github.com/rogerhu/gdb-heap
00:05:10 mnaser melwitt: looks like a bunch of reading but i'll look into it because it seems a bit (very) high
00:05:19 mnaser melwitt: thank you for that information
00:08:37 melwitt the only other thing could be a GMR (guru meditation report)
00:09:32 melwitt I'm not sure that would help too much https://docs.openstack.org/nova/latest/reference/gmr.html
00:10:21 mmedvede mriedem: ok, I'll change that
00:10:49 melwitt but it should be simple to try, send USR2 to the nova-compute pid
00:39:02 openstackgerrit Merged openstack/nova master: Fix PatternPropertiesTestCase for py 3.6 https://review.openstack.org/545798
00:39:22 openstackgerrit Merged openstack/nova master: Do not multiply megabytes with 1024 to get gigabytes. https://review.openstack.org/536335
00:42:09 openstackgerrit Merged openstack/nova master: libvirt: add Linux distribution guest only description for inject_xxx options https://review.openstack.org/528441
00:51:38 openstackgerrit Merged openstack/nova master: Update docstring for get_traits virt driver method https://review.openstack.org/539505
00:51:53 openstackgerrit Merged openstack/nova master: Add resource_class to fields in ironic node cache https://review.openstack.org/539506
00:52:11 openstackgerrit Merged openstack/nova master: trivial: Don't use 'Test' prefix for non-TestCase classes https://review.openstack.org/539251
01:02:13 openstackgerrit Merged openstack/nova master: trivial: Resolve "X is renamed to Y" warnings https://review.openstack.org/539252
01:02:32 openstackgerrit Merged openstack/nova master: Don't modify objects directly https://review.openstack.org/539253
01:04:31 tetsuro could anyone review https://review.openstack.org/#/c/531347/? It got one +2 a month ago.
01:35:51 mriedem melwitt: yeah i guess, i probably hadn't noticed that when rebasing it last time
01:36:04 mriedem mnaser: my guess would be any caching done in the nova.compute.manager.resource_tracker
01:36:15 mriedem mnaser: we cache some placement-related stuff in there now
01:36:32 mriedem but it shouldn't be much..
01:37:13 mriedem https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L133
01:37:25 melwitt mriedem: okay, just wanted to confirm whether there's more to do on that patch before I +W
01:37:57 mriedem melwitt: as far as i know, not tracking the created snapshots and cleaning them up is latent behavior
01:38:13 melwitt mriedem: that's what it looked like to me too
01:38:52 mriedem mnaser: and the RT has an in-memory report client that has a cache of its own https://github.com/openstack/nova/blob/stable/pike/nova/scheduler/client/report.py#L233
01:51:36 openstackgerrit Merged openstack/nova master: Remove unused method _parse_node_instance_info https://review.openstack.org/539507
02:12:15 openstackgerrit Merged openstack/nova master: Add the ability to get absolute limits from Cinder https://review.openstack.org/520122

Earlier   Later