Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-03-03
16:57:37 mriedem we also test live migrating paused instances in the gate
16:57:45 mriedem that's been in nova for several years now
16:57:48 siva_krishnan johnthetubaguy: mriedem it doesn't work when post copy is enabled.
16:58:27 johnthetubaguy siva_krishnan: why?
17:00:21 superdan mriedem: whoa I have core there?
17:00:24 johnthetubaguy siva_krishnan: I am curious, how/when does it fail?
17:00:26 siva_krishnan johnthetubaguy: fails with this error http://paste.openstack.org/show/601362/
17:00:55 johnthetubaguy siva_krishnan: hmm, interesting, work a bug report I think
17:01:28 siva_krishnan johnthetubaguy: sure.
17:01:41 openstackgerrit Lee Yarwood proposed openstack/nova master: Make BlockDeviceMapping object support uuid https://review.openstack.org/242603
17:01:45 openstackgerrit Lee Yarwood proposed openstack/nova master: Add uuid column to BlockDeviceMapping https://review.openstack.org/242602
17:02:55 johnthetubaguy siva_krishnan: probably should not request post_copy if vm is paused, that should fix it
17:03:55 siva_krishnan yup
17:05:27 openstackgerrit Lee Yarwood proposed openstack/nova master: libvirt: Check if domain is persistent before detaching volume https://review.openstack.org/441204
17:09:28 mdbooth lyarwood: Swift work with the transient domain patch :)
17:10:00 mdbooth lyarwood: I wonder, though, if it's kinda weird we're getting the persistent state from guest, then passing it straight back.
17:10:41 mdbooth Intuitively I think guest.detach_device_with_retry() should just do the right thing?
17:10:42 lyarwood mdbooth: weird in the sense it might change in the meantime
17:10:52 mdbooth Nah, I think that's ok
17:11:18 lyarwood mdbooth: yeah I see your point, why allow the assumption
17:11:23 mdbooth I just wonder if the default behaviour of that persistent argument should be neither True nor False, but 'whatever will work'
17:11:50 mdbooth There may be a reason to override it somewhere, but if not we should probably also just remove the argument
17:12:39 mdbooth Anyway, patch looks good. That's a stylistic thing I guess.
17:12:40 kashyap mdbooth: Yeah, during the downstream bug triage, Lee already pointed the fix on IRC
17:12:45 kashyap So yeah, lyarwood++
17:13:47 mriedem1 lyarwood: i guess i'm going to have to do that here too https://review.openstack.org/#/c/349014/
17:14:57 openstackgerrit Danil Akhmetov proposed openstack/nova master: Use proper user and tenant in the owner section of libvirt.xml. https://review.openstack.org/399679
17:15:28 mdbooth mriedem lyarwood: Yeah, I think a sane default could prevent a proliferation of unhandled edge cases
17:15:40 lyarwood mriedem: yeah I think so
17:16:10 mriedem i'm fine with that
17:16:15 mriedem then i don't have to duplicate it
17:17:31 mriedem johnthetubaguy: am i overthinking the need for a unique constraint across volume_id and attachment_id here? https://review.openstack.org/#/c/437597/4/nova/db/sqlalchemy/models.py
17:18:01 mdbooth lyarwood: +1 with a suggestion from me
17:19:45 johnthetubaguy mriedem: hmm, good question, in theory its unique globally in the table I guess?
17:19:47 openstackgerrit Sujitha proposed openstack/nova master: Add description to policies in admin_actions. https://review.openstack.org/440829
17:19:57 mriedem the attachment id?
17:20:08 mriedem well, i wasn't sure if we could have multiple bdms with the same attachment id
17:20:24 mriedem the attachment is scoped to the volume and instance
17:20:54 mriedem and i don't think we'd have a case where we have like, 2 bdms with the same volume and attachment ids but difference device names
17:20:55 mriedem for example
17:21:07 mriedem that would mean the same attachment is mounted in two places which is bonkers
17:22:08 openstackgerrit Matthew Booth proposed openstack/nova master: Remove GlanceImageService https://review.openstack.org/441241
17:22:54 mriedem heh that was on my todo list
17:24:28 openstackgerrit Matthew Booth proposed openstack/nova master: fdatasync() downloaded images before use https://review.openstack.org/441246
17:25:28 lyarwood mdbooth: thanks for the review, would you mind if refactor is in a follow up patch to keep the backports cleaner?
17:25:51 mdbooth lyarwood: Patch looked good, so not at all
17:26:32 lyarwood mdbooth: cool thanks
17:30:58 mdbooth mriedem: I literally just grepped for 'GlanceImageService', and it's no longer present in the codebase anywhere, so unless we're doing some funky class name munging somewhere, I don't think anything can instantiate it.
17:46:47 openstackgerrit Lee Yarwood proposed openstack/nova master: libvirt: Check if domain is persistent before detaching volume https://review.openstack.org/441204
17:55:09 openstackgerrit Andreas Jaeger proposed openstack/nova master: Use Sphinx 1.5 warning-is-error https://review.openstack.org/441257
18:15:24 openstackgerrit Lee Yarwood proposed openstack/nova master: WIP libvirt: refresh connection_info during hard_reboot https://review.openstack.org/441269
18:16:51 openstackgerrit Matthew Booth proposed openstack/nova master: Remove GlanceImageService https://review.openstack.org/441241
18:16:52 openstackgerrit Matthew Booth proposed openstack/nova master: fdatasync() downloaded images before use https://review.openstack.org/441246
18:45:29 superdan melwitt: boom. down to 28 fails: http://logs.openstack.org/94/436094/11/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/f2d1af0/logs/testr_results.html.gz
18:45:59 superdan nine of those are os-hypervisors related I think
18:46:19 melwitt nice
19:15:23 mriedem mdbooth_: $GLANCE_V1_ENABLED in devstack is what i was thinking of i think
19:29:07 openstackgerrit Dan Smith proposed openstack/nova master: Target cell in super conductor operations https://review.openstack.org/438022
19:29:07 openstackgerrit Dan Smith proposed openstack/nova master: WIP: Get instance availability_zone without hitting the api db https://review.openstack.org/439754
19:29:08 openstackgerrit Dan Smith proposed openstack/nova master: Make scheduler target cells to get compute node instance info https://review.openstack.org/439891
19:29:08 openstackgerrit Dan Smith proposed openstack/nova master: Change MQ targeting to honor only what is in the context https://review.openstack.org/438023
19:29:11 openstackgerrit Dan Smith proposed openstack/nova master: Add cell field to Destination object https://review.openstack.org/438024
19:29:15 openstackgerrit Dan Smith proposed openstack/nova master: Teach simple_tenant_usage about cells https://review.openstack.org/441388
19:29:15 openstackgerrit Dan Smith proposed openstack/nova master: WIP: Make conductor ask scheduler to limit migrates to same cell https://review.openstack.org/438025
19:39:54 openstackgerrit Danil Akhmetov proposed openstack/nova master: Use proper user and tenant in the owner section of libvirt.xml. https://review.openstack.org/399679
19:40:40 mriedem cdent: jay wanted to do an os-traits release after https://review.openstack.org/#/q/project:openstack/os-traits+branch:master+topic:normalize was done right?
19:40:58 cdent mriedem: I believe so, yes
19:41:03 mriedem ok
19:41:29 cdent I'm not sure how much of the storage related stuff is done or needs to be done for that to happen
19:42:04 mriedem i'll get the release request teed up and jay can ack it
19:48:09 artom Maybe someone here knows - when running tox -e api-samples SomeTest, it doesn't generate the sameples in doc/
19:48:36 artom I could run tox -e api-samples with no test specified, but that's like 20 minutes. What class/test do I need to add to SomeTest for it to generate the doc/ samples
19:49:33 sdague artom: I don't think it takes a dedicated class
19:50:23 artom sdague, weird. I've added my new .json.tpl file, but when I run the test no doc/ sample is generated
19:50:50 artom Or maybe...
19:51:15 melwitt in the past, there was an env var you had to set to make it generate samples
19:52:47 artom Bleah, I have to create the version directory manually
19:52:52 artom Apologies for the noise
20:08:11 mriedem melwitt: tox -e api-samples has that built in
20:08:32 melwitt yeah, after I said that, I checked and say that in tox.ini
20:08:37 melwitt *saw
20:09:06 melwitt I'm clearly not in the know anymore on api-samples
20:30:34 openstackgerrit Chris Friesen proposed openstack/nova master: Add comment to instance_destroy() https://review.openstack.org/441460
20:30:50 cfriesen_ mriedem: ^ that comment for instance_destroy() we talked about yesterday
20:31:36 openstackgerrit Dan Smith proposed openstack/nova master: Target cell in super conductor operations https://review.openstack.org/438022
20:31:36 openstackgerrit Dan Smith proposed openstack/nova master: WIP: Get instance availability_zone without hitting the api db https://review.openstack.org/439754
20:31:37 openstackgerrit Dan Smith proposed openstack/nova master: Change MQ targeting to honor only what is in the context https://review.openstack.org/438023
20:31:38 openstackgerrit Dan Smith proposed openstack/nova master: Make scheduler target cells to get compute node instance info https://review.openstack.org/439891
20:31:41 openstackgerrit Dan Smith proposed openstack/nova master: Add cell field to Destination object https://review.openstack.org/438024
20:31:44 openstackgerrit Dan Smith proposed openstack/nova master: WIP: Make conductor ask scheduler to limit migrates to same cell https://review.openstack.org/438025
20:31:45 openstackgerrit Dan Smith proposed openstack/nova master: Make consoleauth target the proper cell https://review.openstack.org/441461
20:31:52 cfriesen_ artom: did you find that documented somewhere or did you have to figure it out?
20:32:16 mriedem cool
20:54:20 superdan melwitt: I think one of these fails is due to quota: http://logs.openstack.org/94/436094/11/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/f2d1af0/logs/testr_results.html.gz
20:54:29 superdan melwitt: down near the bottom, test_update_instance_port_admin_state
20:54:53 superdan runs in isolation, but not as part of the larger thing
20:55:22 superdan I'm wondering if we're reserving in cell0 and claiming in cell1 or something like that
20:56:20 melwitt superdan: ah, okay. yeah, I think it would work that way. the quota commit (done by compute) would probably end up in the compute cell's db, so that would end up not freeing the reservation at the top
20:56:30 superdan right
20:57:02 superdan is reservations where that gets made in the api?

Earlier   Later