Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-09-01
14:31:45 mriedem something fails, cleanup the dest allocation
14:31:46 bauwser yeah I can see
14:31:47 cdent mriedem, dansmith ^ that post allocations wip is very wippy, but wanted to get some early commentary
14:32:36 sean-k-mooney o/ i dont see jay pipes online. is he traveling today. i think he was in europe this week?
14:32:48 openstackgerrit Stephen Finucane proposed openstack/nova master: conf: Rename two VNC options https://review.openstack.org/498387
14:32:50 mriedem sean-k-mooney: he's on vacation today
14:32:54 mriedem so out until tuesday
14:32:57 mriedem since monday is a US holiday
14:33:27 sean-k-mooney ah ok noting urgent.
14:34:37 openstackgerrit Stephen Finucane proposed openstack/nova master: conf: Deprecate 'keymap' options https://review.openstack.org/483994
14:38:28 rabel i think https://review.openstack.org/#/c/494169/ is ready for a final review (again). it even passed VMware NSX CI this time. ;)
14:40:53 mriedem rabel: question inline
14:44:33 openstackgerrit Matt Riedemann proposed openstack/nova master: doc: link to placement api-ref and history docs from main index https://review.openstack.org/498977
14:44:34 openstackgerrit Matt Riedemann proposed openstack/nova master: doc: link to versioned notification samples from main index https://review.openstack.org/500081
14:44:37 mriedem gibi: ^
14:44:38 mriedem see how that looks
14:45:39 openstackgerrit Elod Illes proposed openstack/nova master: Set error state after failed evacuation https://review.openstack.org/499237
14:46:34 gibi mriedem: looking
14:48:58 elod mriedem: bug-fix rebased
14:49:11 mriedem yup, just looked it over
14:49:14 mriedem waiting on tests but lgtm
14:51:40 elod ok, thx
14:52:33 gibi mriedem: in the index the first sentence of the section explicity talks about REST API which now conflicts with the added notification bullet point
14:52:42 gibi mriedem: "All end user (and some administrative) features of nova are exposed via a REST"
14:54:07 gibi mriedem: what if we don't add the notifications directly to the bullet list but add an extra sentence talking about the notifications after the bullet list?
14:55:09 mriedem ok i'll take a look
15:02:03 mriedem lpetrut: do you have some time to discuss https://bugs.launchpad.net/nova/+bug/1714247 ?
15:02:04 openstack Launchpad bug 1714247 in OpenStack Compute (nova) "Cleaning up deleted instances leaks resources" [Undecided,New]
15:06:58 alex_xu sdague: a patch related 'microversion or not' https://review.openstack.org/#/c/490722, it will be good if you can take a look at
15:10:14 gibi mriedem: I'm looking at: https://review.openstack.org/#/c/499399/2/nova/conductor/manager.py@760 . Does ironic custom resource provided by the compute resource provider or a separate resource provider per compute?
15:12:29 figleaf alex_xu: if you're still around, do you want to give your opinion on whether this needs a microversion or not? https://review.openstack.org/#/c/499826/
15:12:39 stephenfin sean-k-mooney, ralonsoh: Resolved your comments in https://review.openstack.org/#/c/483398/ finally
15:13:16 ralonsoh stephenfin: I'll put this patch in my todo list for today or monday morning
15:13:17 ralonsoh thanks!
15:14:48 stephenfin no problemo :)
15:18:52 cdent not clear if I’m going to have time for a rp update today. If I do it will be later in the day
15:21:14 openstackgerrit Matt Riedemann proposed openstack/nova stable/pike: doc: fix show-hide sample in notification devref https://review.openstack.org/500100
15:35:19 openstackgerrit Merged openstack/nova master: doc: fix show-hide sample in notification devref https://review.openstack.org/499963
15:41:07 openstackgerrit Matt Riedemann proposed openstack/nova master: add online_data_migrations to nova docs https://review.openstack.org/493442
15:41:37 stephenfin mriedem, sean-k-mooney: Where did we come down on https://bugs.launchpad.net/nova/+bug/1698010? Do we actually want to use a FQDN for hostname?
15:41:38 openstack Launchpad bug 1698010 in OpenStack Compute (nova) "neutron-based instances should not use the nova-network 'dhcp_domain' option" [High,In progress] - Assigned to Stephen Finucane (stephenfinucane)
15:43:04 openstackgerrit Stephen Finucane proposed openstack/nova master: add online_data_migrations to nova docs https://review.openstack.org/493442
15:47:13 mriedem dansmith: can you clarify a question i have in here? https://review.openstack.org/#/c/493442/6
15:47:15 openstackgerrit Rodolfo Alonso Hernandez proposed openstack/os-vif master: Set HostPluginInfo plugin_name as an Enum https://review.openstack.org/500111
15:49:03 dansmith mriedem: yeah, so really it should be only the controllers, but we had one incident where it wouldn't complete until all the computes have been upgraded
15:49:05 dansmith mriedem: the pci one
15:49:17 dansmith I was talking to someone recently about this, not sure if it was this person
15:49:41 dansmith I dunno what to say here, because in some cases it'd be one or the other, but almost always just controllers
15:51:01 mriedem in the case of the pci issue,
15:51:20 mriedem did running online_data_migrations break anything? or just not migrate things as expected?
15:51:35 dansmith nope it would just warn you that it couldn't finish because it needed the computes to be upgraded first
15:51:41 mriedem ah, ok
15:52:00 mriedem so, then i think we leave the ideal case in the man page, and if we have a known weird wrinkle, then we release note that
15:52:16 dansmith sounds good
15:59:57 mriedem evacuating local git
16:00:06 mriedem force=True
16:00:07 openstackgerrit Matt Riedemann proposed openstack/nova master: Remove dest node allocation if evacuate MoveClaim fails https://review.openstack.org/499878
16:00:08 openstackgerrit Matt Riedemann proposed openstack/nova master: Add a test to make sure failed evacuate cleans up dest allocation https://review.openstack.org/499877
16:00:08 openstackgerrit Matt Riedemann proposed openstack/nova master: Add recreate test for forced host evacuate not setting dest allocations https://review.openstack.org/499678
16:00:09 openstackgerrit Matt Riedemann proposed openstack/nova master: Create allocations against forced dest host during evacuate https://review.openstack.org/499399
16:00:09 openstackgerrit Matt Riedemann proposed openstack/nova master: Add recreate test for evacuate claim failure https://review.openstack.org/499874
16:00:10 openstackgerrit Matt Riedemann proposed openstack/nova master: Modernize set_vm_state_and_notify https://review.openstack.org/499799
16:00:10 openstackgerrit Matt Riedemann proposed openstack/nova master: Refactor out claim_resources_on_destination into a utility https://review.openstack.org/499718
16:00:18 stephenfin sdague: If you get a chance, would appreciate a sanity check on https://bugs.launchpad.net/nova/+bug/1704458
16:00:19 openstack Launchpad bug 1704458 in OpenStack Compute (nova) "The use_ipv6 flag not only influences nova networking" [High,Invalid] - Assigned to Stephen Finucane (stephenfinucane)
16:07:32 mriedem stephenfin: regarding that dhcp_domain thing for dns, i think it makes sense that if you're using neutron, you pull it from the neutron network resource as it appears your series is doing
16:07:48 mriedem stephenfin: i just have to get through all of your unrelated changes to get to the meat of the fix first... :
16:07:49 mriedem :)
16:09:12 stephenfin mriedem: Right, but the question is whether we should actually be using FQDN at all, rather than from neutron instead. sean-k-mooney seemed to suggest it was a bad idea, but my network/sysadmin foo is not good enough to say why not :)
16:09:41 stephenfin Also, I like refactoring. Sorry :D I can pull those out
16:10:11 mriedem run swift young stephen
16:10:18 mriedem don't look back....
16:18:37 openstackgerrit Matt Riedemann proposed openstack/nova master: doc: fix online_data_migrations option in upgrades doc https://review.openstack.org/500124
16:34:37 openstackgerrit Matt Riedemann proposed openstack/nova master: Set error state after failed evacuation https://review.openstack.org/499237
16:35:50 mriedem we're going to want to backport this fix ^ back through newton
16:35:58 mriedem so it'd be nice to start that, it's pretty trivial
16:36:32 mriedem prevents the source compute from deleting local instance if the evacuation fails in conductor to find a dest host
16:39:00 dansmith mriedem: so just thinking about this quickly,
16:39:31 dansmith mriedem: are you thinking that if we fail to schedule a new place and the source comes back up that there's value in having it still there?
16:40:03 dansmith because the way this was designed, once an evacuation starts we're *going* to delete it from the source compute to avoid a race between rebuilding it elsewhere and the compute coming back up and deciding to keep it or not
16:40:42 mriedem melwitt: commented on https://review.openstack.org/#/c/407346/ - cinder would also need changes if we do something like this in nova
16:42:25 mriedem dansmith: hmm
16:42:28 mriedem that's a good point
16:42:51 dansmith knowing the folks that use this a lot from a script,
16:42:58 melwitt mriedem: cool thanks. looks like it's not as straightforward was I initially thought
16:43:00 dansmith the last thing they want is vague behavior from this
16:43:14 mriedem i wish that would have been a comment in the _delete_evacuated_instances code because i spent a good chunk of time yesterday trying to figure out why we include 'accepted' in the filter
16:43:42 dansmith mriedem: was it in my robustify spec? I think the logic was sussed out there
16:43:44 dansmith anyway,
16:43:52 mriedem i didn't read the spec
16:44:16 dansmith I would think that what we want is to make sure you can call rebuild (or at least reset-state) on the instance and have it rebuilt somewhere else
16:44:55 dansmith because I think we've nulled out the instance.host here at this point right? meaning it won't be a simple restart on the source host
16:45:05 openstackgerrit Merged openstack/nova master: Functional test for regression bug #1713783 https://review.openstack.org/500057
16:45:07 openstack bug 1713783 in OpenStack Compute (nova) "After failed evacuation the recovered source compute tries to delete the instance" [High,In progress] https://launchpad.net/bugs/1713783 - Assigned to Matt Riedemann (mriedem)
16:45:26 mriedem dansmith: i don't think the instance.host is nulled out
16:45:30 mriedem which is part of the problem
16:45:39 mriedem reading the docstring for _destroy_evacuated_instances, it says,
16:45:48 openstackgerrit Merged openstack/nova master: add online_data_migrations to nova docs https://review.openstack.org/493442
16:45:48 mriedem "Check that the instances reported by the driver are still associated with this host. If they are not, destroy them"
16:45:57 mriedem but it doesn't actually compare the instances that it finds against self.host

Earlier   Later