Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-09-13
15:20:24 stephenfin Not quite
15:20:33 openstackgerrit Merged openstack/nova master: Find instance in another cell during floating IP re-association https://review.opendev.org/656594
15:21:08 stephenfin It's only if you have '[libvirt] cpu_model' and '[libvirt] cpu_flags' (or maybe just the first - I'll check) set on the host
15:21:31 stephenfin which I _think_ most people will have because (a) live migration and (b) spectre/meltdown
15:21:55 mriedem you mean cpu_models and cpu_model_extra_flags yeah?
15:22:08 stephenfin that's the one
15:22:09 mriedem and yeah those are both [] in this run
15:23:05 stephenfin Because we're not actually using the libvirt driver in the functional tests, it should be possible set both of those in a functional tests
15:23:07 stephenfin I'll check
15:23:47 mriedem so cpu_model was renamed to cpu_models in train, right?
15:24:05 stephenfin it was. It's now a list
15:24:08 mriedem because i see https://docs.openstack.org/nova/stein/configuration/config.html#libvirt.cpu_model in stein but not in train and https://docs.openstack.org/nova/latest/configuration/config.html#libvirt.cpu_models in train but not stein
15:24:16 mriedem so,
15:24:45 mriedem if cpu_model isn't deprecated with a rename how is the upgrade handled? just checking for cpu_models first and if not found use cpu_model and log a warning to use the new option?
15:25:00 stephenfin deprecated_name
15:25:07 mriedem bnemec: is there any kind of flag in oslo.config that says "kill the service on startup if using config options that aren't actually defined"?
15:25:27 stephenfin ListOpt works just like StrOpt if you only list one item
15:25:28 mriedem oh there it is
15:25:32 mriedem bnemec: ignore
15:29:38 mriedem so is there a model i could use in this devstack job that we could expect to be in a gate run, like IvyBridge? and i could probably just set cpu_model_extra_flags to pcid?
15:30:34 mriedem or vmx or something?
15:31:21 sean-k-mooney vmx would need nested virt but ya pcid would proably work
15:31:22 stephenfin I'm using this https://gist.github.com/stephenfin/31a44d9cc40d9ce7abffadb5a6830cb1#file-local-conf-master-L25-L28
15:31:36 stephenfin and that's on 6+ year old hardware so it should do the trick
15:31:56 mriedem ok i'll update my devstack patch
15:33:30 sean-k-mooney pcid was added in the first gen xeon i think back like 11 years ago
15:33:35 sean-k-mooney so ya that is safe
15:33:46 sean-k-mooney well
15:33:53 sean-k-mooney as long as the geate is useing it
15:34:01 sean-k-mooney it will proably work however
15:34:09 mriedem we'll see https://review.opendev.org/#/c/681938/
15:36:05 mriedem updated release cycle highlights https://review.opendev.org/681943
15:45:20 aspiers I've just drafted a blog post on SEV. Feedback welcome: https://blog.adamspiers.org/?p=1871&preview=1&_ppp=1bfc4e4f7e
16:18:56 openstackgerrit Matt Riedemann proposed openstack/nova master: api-ref: remove mention about os-migrations no longer being extended https://review.opendev.org/682102
16:19:35 openstackgerrit Matt Riedemann proposed openstack/nova master: api-ref: remove mention about os-migrations no longer being extended https://review.opendev.org/682102
16:24:36 stephenfin aspiers: Can you check something for me?
16:30:34 aspiers stephenfin: I'll try
16:31:19 stephenfin sorry, got distracted
16:31:20 stephenfin aspiers: https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L697
16:31:35 stephenfin Isn't that line wrong? Should it read:
16:31:49 stephenfin self._host.get_capabilities().host.cpu.model (not .arch) ?
16:31:54 stephenfin *shouldn't it
16:32:36 stephenfin (Actually, "shouldn't it" is colloquial. Oops :))
16:39:11 sean-k-mooney https://review.opendev.org/#/c/634827/ failed withthe 6 not greathan 7 notificaiton thing again
16:40:10 stephenfin sean-k-mooney: https://review.opendev.org/682025
16:40:13 sean-k-mooney so the numa serise will block until https://review.opendev.org/#/c/682025/ is mergd
16:40:17 sean-k-mooney ya
16:40:34 sean-k-mooney just pointing out there is no point recheck untill that has merged
16:41:03 sean-k-mooney wow that was the fix...
16:44:18 stephenfin yup
16:44:32 stephenfin tbf, I've made that same mistake myself many times
16:44:39 openstackgerrit Stephen Finucane proposed openstack/nova master: DNM: Reproducer for bug #1843836 https://review.opendev.org/682111
16:44:41 openstack bug 1843836 in OpenStack Compute (nova) "Failure to schedule if flavor contains non-CPU flag traits" [Undecided,In progress] https://launchpad.net/bugs/1843836 - Assigned to Stephen Finucane (stephenfinucane)
16:44:43 stephenfin mriedem_afk: ^
16:44:58 sean-k-mooney i think it was enqued directly to gate so it should merge soon i hope
16:45:43 stephenfin mriedem_afk: there's a WIP functional reproducer for the bug. I'll clean it up Monday once I'm done investigating the other issue that reproducer highlighted
16:45:50 stephenfin *potential issue
16:45:55 stephenfin sean-k-mooney: It sure was, yup
16:47:49 aspiers stephenfin: that github link is basically a DoS attack on my browser X-D
16:48:03 stephenfin aspiers: Hahahaha
16:48:17 stephenfin yeah...don't try to load it on a phone
16:48:49 aspiers no that was on my laptop
16:49:08 stephenfin if there was ever a sign that a file had grown too large...
16:49:15 aspiers right?
16:49:30 stephenfin funnily enough, libvirt source files tend to be huuuuge as well
16:49:57 sean-k-mooney thats not uncommon in c
17:00:15 aspiers stephenfin: anyway yes it looks wrong to me
17:01:22 stephenfin \o/
17:01:32 stephenfin Another one to fix on Monday
17:01:42 stephenfin Thanks for taking a look, aspiers
17:17:55 aspiers BTW I'm off next week
17:18:11 aspiers stephenfin: In case you missed it, I've just drafted a blog post on SEV. Feedback welcome: https://blog.adamspiers.org/?p=1871&preview=1&_ppp=1bfc4e4f7e
17:23:37 stephenfin aspiers: Won't get to it today but I'll try early next week. Enjoy your time off!
17:25:59 aspiers cheers :)
17:28:23 artom Where's the part you call us all poopyheads for not reviewing your code enough?
17:49:13 sean-k-mooney damit... mriedem_afk https://review.opendev.org/#/c/682025/ is going to fail
17:50:34 sean-k-mooney subunit parser is exploding
17:51:33 mriedem drats
17:51:55 sean-k-mooney https://storage.gra1.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_897/682025/1/gate/openstack-tox-py37/89752ac/testr_results.html.gz
17:53:02 artom What even causes those?
17:53:32 sean-k-mooney large amount of output to standard out or error
17:53:59 sean-k-mooney like all the logging of the host capablities
17:54:48 mriedem those seem to have spiked against recently so something must have merged that is causing us to tip over
17:54:57 mriedem http://status.openstack.org/elastic-recheck/#1813147
17:55:07 mriedem #2 top most fail in the check queue right now
17:55:49 sean-k-mooney its happening in other jobs too not just py35 but ya
17:56:13 sean-k-mooney before it was due to a bunch of deprecation warnings
17:57:35 mriedem the name of the bug is old
17:58:18 sean-k-mooney can we force submit that change? im running the unit test locally to see if it explodes but i dont expect it to
17:58:37 mriedem as in get it re-queued?
17:58:44 mriedem i think you'd have to rebase
17:58:50 mriedem which will put it back through check
17:58:59 sean-k-mooney as in ask infra to override zuul
17:59:00 mriedem i asked for it to be promoted to gate earlier in infra
17:59:08 sean-k-mooney but we could requeu it too
17:59:45 mriedem jesus look at this https://4c970d31e6335324d673-0d0b23b7f62d479822a1b52b605369e3.ssl.cf1.rackcdn.com/682025/1/gate/nova-tox-functional-py36/5f76307/job-output.txt
17:59:49 mriedem search for inprogress
17:59:51 mriedem see the wall of 00s
18:00:02 artom Those subunit things appear to start on september... 4th?
18:00:06 mriedem wtf is that

Earlier   Later