Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-11-22
15:21:31 openstackgerrit Stephen Finucane proposed openstack/nova master: Don't create instance_extra entry for deleted instance https://review.openstack.org/412771
15:22:06 kashyap bauzas: If you don't mind a bit of typing: `systemctl restart foo bar baz whizz` should restart them all
15:22:22 bauzas kashyap: yeah I knew about that
15:22:25 bauzas but meh
15:22:27 bauzas okay
15:24:23 openstackgerrit Matthew Booth proposed openstack/nova master: libvirt: Re-initialise volumes, encryptors, and vifs on hard reboot https://review.openstack.org/400384
15:25:11 bauzas kashyap: given we have a 'systemctl list-units --failed' comment, I was hoping to see a 'systemctl start --failed'
15:25:27 bauzas but meh, I can gawk that
15:29:32 bauzas kashyap: oh hell, see that https://docs.openstack.org/devstack/latest/systemd.html#operating-on-more-than-one-unit-at-a-time
15:29:42 bauzas bauzas, RTFM, man !
15:30:17 kashyap bauzas: Ah
15:30:18 openstackgerrit Jianghua Wang proposed openstack/nova master: XenAPI: get vGPU stats from hypervisor https://review.openstack.org/512965
15:34:35 alex_xu mriedem: the second time we break the API due to the schema copy :(
15:37:14 mriedem alex_xu: yeah, a functional api sample test for that one at 2.36 would have caught it
15:37:31 mriedem anyway, i think i'll just document it in the api-ref and do a release note and we're good
15:37:40 mriedem the force parameter doesn't matter since pike anyway
15:38:45 alex_xu mriedem: what means for doesn't matter
15:39:02 mriedem force = update quota regardless of current use + reserved
15:39:05 mriedem since counting quotas in pike,
15:39:09 mriedem we don't have quota reservations anymore
15:39:47 mriedem although, maybe force still applies here, maybe i'm not reading this correctly
15:40:31 alex_xu mriedem: I thought force = update quota regardless of current use
15:40:34 stephenfin mdbooth: Could you briefly explain what 'is_implemented' was supposed to do here? https://review.openstack.org/#/c/511217/1/nova/virt/block_device.py
15:40:39 mriedem yeah maybe it does
15:40:51 mriedem so if usage is 10 and i change quota to 5, i'd have to force it
15:41:03 mriedem so maybe we should fix this in 2.56 with the file injection changes i'm making
15:42:10 alex_xu mriedem: yea, it can be
15:42:50 mriedem or we fix it in it's own microversion, i'm not sure if it's good to lump that bug fix into the file injection deprecation changes, which are going to be pretty large
15:42:57 alex_xu mriedem: in a single microversion also easy, since the fix is easy
15:42:58 mriedem this is the wip so far on that bp https://review.openstack.org/#/c/522027/
15:43:21 alex_xu mriedem: I prefer it's own microverion more
15:44:00 mriedem yeah me too i think
15:44:22 mriedem ok i can work up a spec on that i suppose and work it under my file injection change
15:44:49 alex_xu mriedem: ok
15:45:55 cfriesen kashyap: thanks...based on the comments at https://bugzilla.redhat.com/show_bug.cgi?id=1439078 it appears that virConnectCompareCPU() will not catch the case where we're using a cpu_mode of 'host-passthrough' and HT is disabled on the source and enabled on the dest.
15:45:55 openstack bugzilla.redhat.com bug 1439078 in qemu-kvm-rhev "After migration,VM crash in dst host with "qemu-kvm: error: failed to set MSR 0x38f to 0x7000000ff"" [Unspecified,Closed: notabug] - Assigned to dgilbert
15:46:00 mdbooth stephenfin: IIRC the way those classes work is by raising an exception in the constructor if the bdm isn't of the correct type
15:46:31 alex_xu mriedem: btw, there is patch list from bugsmash, some new contributor are submitted patches https://etherpad.openstack.org/p/OpenStack-Bug-Smash-Queens-Wuhan-Bugs-List
15:46:58 mdbooth stephenfin: So is_implemented() is returning True if there is any class which can instantiate an object for the given bdm
15:47:00 kashyap cfriesen: Right, it won't magically catch it
15:47:44 stephenfin mdbooth: And we don't need/want to do that anymore because?
15:47:49 mriedem alex_xu: of course i saw the group photo on wechat
15:48:39 mdbooth stephenfin: Oh, wow, this is my patch
15:48:50 mdbooth stephenfin: I don't recognise this at all :O
15:49:12 stephenfin mdbooth: This is what happens when I skim old bug patches :)
15:49:19 alex_xu mriedem: hah
15:49:20 stephenfin slowly whittling them down
15:49:34 mdbooth Sec, let me get some context
15:50:56 mdbooth stephenfin: Ah... we don't want to do that any more because none of the other almost identical things do that
15:51:04 mdbooth And it's ugly as hell...
15:51:31 mdbooth And inefficient, and we don't need to do that
15:51:33 alex_xu mriedem: thanks for the email
15:52:20 cfriesen kashyap: so is this a nova bug (don't allow live migration between non-HT and HT hosts when using 'host-passthrough') or is it a documentation issue (when using 'host-passthrough' always set HT the same on all the hosts in a given AZ)
15:52:32 mdbooth stephenfin: See is_swap_or_ephemeral, new_format_is_swap, new_format_is_ephemeral
15:54:40 stephenfin That's as good a reason as any
15:54:47 stephenfin mdbooth: +2d
15:55:33 stephenfin mdbooth: This one looks good too, only for the broken UTs https://review.openstack.org/#/c/511218/
16:08:36 openstackgerrit Elod Illes proposed openstack/nova master: Transform instance-live_migration_pre notification https://review.openstack.org/482070
16:13:22 stephenfin sahid: Can you take a look at this? Looks innocuous enough https://review.openstack.org/#/c/522161/
16:18:18 gibi stephenfin, mriedem: Could one of you check back to https://review.openstack.org/482070 it needed a small extra change top of the assert mriedem asked for so I haven't fast approved it yet
16:26:14 openstackgerrit Merged openstack/nova master: rt: Make resource tracker always invoking get_inventory() https://review.openstack.org/518294
16:27:16 melwitt mriedem: skipping the cells meeting today? I think dansmith is out today
16:27:57 mriedem yes, i'm out today too
16:28:08 melwitt k, cool
16:28:51 tssurya okay :)
16:29:25 sahid stephenfin: ok i will
16:37:57 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: Deprecate file injection https://review.openstack.org/522027
16:47:20 openstackgerrit Stephen Finucane proposed openstack/nova master: Added PCI NUMA policies https://review.openstack.org/390520
17:14:10 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove old-style quotas code https://review.openstack.org/511689
17:16:08 melwitt stephenfin: did you see my comment on the review? I also forgot to remove the command from the nova-manage doc
17:16:45 stephenfin melwitt: I did not. I agree though - let's keep it separate
17:17:15 stephenfin melwitt: You want to do that or will I?
17:17:34 stephenfin (remove the nova-manage change from the patch)
17:18:00 openstackgerrit Merged openstack/nova master: Use Migration object in ComputeManagerMigrationTestCase https://review.openstack.org/502126
17:20:59 melwitt stephenfin: I can do it. sec
17:25:33 openstackgerrit melanie witt proposed openstack/nova master: Remove old-style quotas code https://review.openstack.org/511689
17:27:54 openstackgerrit Merged openstack/nova master: Moving more utils to ProviderUsageBaseTestCase https://review.openstack.org/499539
17:28:31 openstackgerrit Matt Riedemann proposed openstack/nova-specs master: Fix force parameter for os-quota-sets API https://review.openstack.org/522305
17:28:34 mriedem alex_xu: bauzas: jaypipes: ^
17:28:39 mriedem johnthetubaguy: sdague: ^
17:31:20 sdague mriedem: hmmm... this being an admin side API I would be more inclined to fix it and backport it.
17:31:27 johnthetubaguy talking a look
17:32:09 mriedem sdague: that would certainly be easier
17:33:09 mriedem feel free to leave a comment in there; it was regressed in newton, and we still have stable/newton, so we could backport a fix if we decided to go that route
17:33:16 mriedem it's literally a 1 line code fix
17:35:15 sdague mriedem: comment left
17:36:51 mriedem thanks
17:40:05 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove 'nova-manage quota refresh' command https://review.openstack.org/521829
17:40:05 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove 'nova-manage host' and 'nova-manage agent' https://review.openstack.org/521837
17:40:06 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove 'nova-manage shell' command https://review.openstack.org/521835
17:40:06 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove 'nova-manage account' and 'nova-manage project' https://review.openstack.org/521833
17:40:07 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove 'nova-manage logs' command https://review.openstack.org/522133
18:08:54 cfriesen got a DB question...why don't we have "cascade='delete'" for "flavor_extra_specs" and "flavor-projects" tables the way we do for aggregates?
18:33:05 frickler nova-conductor seems to sometimes mangle errors it receives from nova-compute and replace them with "CantStartEngineError: No sql_connection parameter is established", is that a known issue?
18:38:27 frickler http://paste.openstack.org/show/627134/
18:51:32 frickler seems to happen quite regularly in CI, too: http://logstash.openstack.org/#dashboard/file/logstash.json?query=message%3A%5C%22Setting%20instance%20to%20ERROR%20state.%3A%20CantStartEngineError%5C%22
18:51:43 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: Deprecate file injection https://review.openstack.org/522027
19:04:09 _ix Good afternoon folks. Is there a way to force a sync with the database such that only the nodes that the hypervisors are running are populated in the controller db?
19:04:47 _ix Every now and again, it looks like nova successfully deletes a node, but there's a subsequent error for powering the machine down because the task state of 'deleting' won't allow it.
19:04:57 _ix So, we have a few VMs that are lingering.

Earlier   Later