Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-07-16
15:07:39 sean-k-mooney it should be trivial to add a functional test to assert that it does not change
15:08:11 sean-k-mooney or rather that an attepmt to rebuild form dedicated to mixed will be rejected
15:08:29 gibi bauzas: I will ask the others on the meeting about a spec review day
15:08:35 gibi bauzas: next week is good for you?
15:08:51 bauzas yup
15:09:09 bauzas anyway, it's just a question, since we don't have a lot of open specs
15:09:09 gibi cool
15:09:27 bauzas https://review.opendev.org/#/q/project:openstack/nova-specs+is:open
15:09:29 gibi at most it won't be a super busy day :)
15:09:46 bauzas this said, I'll create a new spec for the reshape tool
15:09:49 sean-k-mooney its proably good to do before m2 in either case
15:09:58 gibi bauzas: and I will push one for SRIOV attach
15:10:12 bauzas ack, all good
15:10:15 gibi :)
15:10:54 bauzas gibi: I'd also write a spec for GPU NUMA affinity using the NUMA topology in placement, but it's more for discussing about it this cycle, not wanting to merge it
15:11:15 sean-k-mooney gibi: im proably going to try an add support for sriov PF detach as a bug fix this cycle too
15:11:27 sean-k-mooney im not sure if it would make sense to add it to your spec or not
15:12:01 huaqiang sean-k-mooney: I need time to digest your code.
15:12:18 sean-k-mooney gibi: when vnic_type=direct-physical was added detach was never updated
15:12:46 sean-k-mooney it tries to find the nic by its mac but we dont have a mac in this case
15:12:59 gibi sean-k-mooney: before you go into that check my detach code if that solves your case too ro not
15:13:02 gibi bauzas: ack
15:13:25 sean-k-mooney gibi: it might we just need to fine the device by its pci address not it mac
15:14:07 sean-k-mooney since we use a <hostdev> element instead of <interface type=hostdev> for pfs
15:15:08 sean-k-mooney sriov detach works for vnic_type=direct|macvtap|virtio-forwarder today just not direct-physical
15:15:42 gibi sean-k-mooney: works and removes the pci claim too?
15:16:22 sean-k-mooney gibi: ill admit i cant rememebr i think so but we dont currently block it because it complete succefully
15:16:43 sean-k-mooney so its updateing the neuton port and removing the port form the virtual interfaces table
15:17:00 gibi sean-k-mooney: I had to add https://review.opendev.org/#/c/740995/2/nova/compute/manager.py@7416 to make the pci device freed
15:17:05 sean-k-mooney i belive it free the device but it might be "working" but not doing the right thing
15:17:31 gibi but it can be that some periodic would free that
15:17:39 dansmith melwitt: the last dependency for this is about to merge in the gate.. wanna send it? https://review.opendev.org/#/c/734184/
15:17:53 sean-k-mooney gibi: ya i think the periodic is what we rely on
15:18:40 sean-k-mooney doing it premtivly is better but i think we get back into a consitent state
15:18:55 sean-k-mooney ill test it before i apply your patch and let you know
15:20:03 gibi sean-k-mooney: cool
15:25:50 sean-k-mooney gibi: by the way if i confim that we did not free the pci device do we want to backport a fix to block detach the same way we backported a patch to block attach for older releases?
15:26:08 sean-k-mooney i.e. if the periodic does not fix it
15:29:42 gibi sean-k-mooney: I'm not sure. If people using it then blocking it would be bad
15:30:12 sean-k-mooney it cloud be done vai a conig option i guess like numa live migration
15:30:25 sean-k-mooney we can cross that bridge when we come to it
15:30:29 gibi yeah
15:45:31 gibi nova meeting starts in 15 minutes on #openstack-meeting-3
16:30:55 artom Hey, if we set a bug as incomplete, can the reporter then switch it back once they come back with new logs?
16:31:35 artom My thinking is: to avoid other people looking at it needlessly, set it as incomplete, and let the reporter come back with the info. Is that how we do i?
16:31:38 artom it
16:31:40 gibi artom: they can switch it back
16:31:55 gibi just ask them to set it back to New
16:32:47 openstackgerrit Elod Illes proposed openstack/nova stable/rocky: Check cherry-pick hashes in pep8 tox target https://review.opendev.org/741492
16:44:24 artom Is (anti-)affinity in placement a thing we track?
16:44:33 sean-k-mooney artom: no
16:45:03 openstack Launchpad bug 1886160 in OpenStack Compute (nova) "Anti-affinity instance creation failed and the scheduling node was incorrect." [Undecided,New]
16:45:03 artom Context: https://bugs.launchpad.net/nova/+bug/1886160 looks like ye olde "the late anti-affinity check on the host is super racy" problem
16:45:11 artom So I'm looking for something to tell them other than "tough luck"
16:45:43 sean-k-mooney how about dont use multi create if you also use numa or sriov
16:46:31 sean-k-mooney the fact we now precompute a series of alternate host insead of rescudling on retry actully makes anti affinity worse
16:53:57 artom Right, we're < 30 untriaged, and I need lunch
17:04:28 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Add vTPM config support https://review.opendev.org/739209
17:04:28 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Simplify '_create_domain' function https://review.opendev.org/739208
17:04:29 openstackgerrit Stephen Finucane proposed openstack/nova master: crypto: Add support for creating, destroying vTPM secrets https://review.opendev.org/739211
17:04:29 openstackgerrit Stephen Finucane proposed openstack/nova master: scheduler: Request vTPM trait based on flavor or image https://review.opendev.org/739210
17:04:30 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Rename tests for '_create_guest_with_network' https://review.opendev.org/740464
17:04:30 openstackgerrit Stephen Finucane proposed openstack/nova master: manager: Prevent compute startup on invalid vTPM config https://review.opendev.org/739212
17:04:31 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Define constants in '_IntegratedTestBase' https://review.opendev.org/741281
17:04:31 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Move single use constants to their callers https://review.opendev.org/741280
17:04:32 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Add 'PlacementHelperMixin', 'PlacementInstanceHelperMixin' https://review.opendev.org/741283
17:04:32 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Remove 'test_servers.ServersTestBase' https://review.opendev.org/741282
17:04:33 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Add helpers for suspend, resume and reboot of server https://review.opendev.org/741285
17:04:33 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Make '_IntegratedTestBase' subclass 'PlacementInstanceHelperMixin' https://review.opendev.org/741284
17:04:34 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Add emulated TPM support to Nova https://review.opendev.org/631363
17:04:34 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Pass context, instance to '_create_domain' https://review.opendev.org/741286
17:04:35 openstackgerrit Stephen Finucane proposed openstack/nova master: docs: Add docs for vTPM support https://review.opendev.org/739213
17:04:35 openstackgerrit Stephen Finucane proposed openstack/nova master: WIP: Add support for resize and cold migration of emulated TPM files https://review.opendev.org/639934
17:04:40 openstackgerrit Stephen Finucane proposed openstack/nova master: api: Reject non-spawn operations for vTPM https://review.opendev.org/741500
17:38:55 openstackgerrit sean mooney proposed openstack/nova master: lookup nic feature by pci address https://review.opendev.org/739131
18:44:17 sean-k-mooney gibi: some comments in the patch. im not sure the interface you are using for the new fucntions is correct
18:44:39 sean-k-mooney overall the logic seam pretty sound but its not quite how i would do it
19:03:07 sean-k-mooney gibi: ya so the perodic task does not free the claim and they are not freed by the detach without your patch
19:03:24 sean-k-mooney gibi: deleting the instance does
19:04:04 sean-k-mooney so right now if you detatch an sriov interface it leaks the pci device until the vm is deleted
19:04:30 sean-k-mooney im going to check if we fix it with live migrations or cold migrations
19:07:06 sean-k-mooney yes live migration frees the claims
19:11:12 sean-k-mooney cold migration does not fix the issue an result in claiming a vf on the dest host that is unused
19:11:48 sean-k-mooney shevle and unshelve also claims an extra vf
19:13:28 sean-k-mooney gibi: so it look like anything that uses move claims does not correct it but sriov live migration which intentionally does not use move claims and caluatles the required pci device by looking at the attached interfaces does correct the issue
19:14:43 sean-k-mooney ok now that i have tested my patch ill test your and see if it fixes it.
19:33:00 openstackgerrit Tobias Urdin proposed openstack/nova master: [PoC] libvirt: Broadcast RARP post live migration https://review.opendev.org/741529
19:51:41 sean-k-mooney tobias-urdin: ^ that is not really a fix for the issue its just going to mask the race condition
19:55:38 sean-k-mooney i left a comment inline but you need to use the python bindings not virsh
20:51:30 mnaser hrm
20:51:56 melwitt mnaser: did you see my reply to you yesterday?
20:52:06 mnaser melwitt: oh yes, sorry, i mentally said "yes"
20:52:19 melwitt np, just wanted to make sure :)
20:52:33 mnaser my fun thing now is a vm that takes 725 seconds to provision
20:52:40 melwitt yikes
20:52:48 mnaser and the only strange thing is like
20:52:59 mnaser Creating image .... nothing from there until VM started
20:53:06 mnaser 12 minutes until it starts
20:53:34 mnaser and here's the thing.. the image it uses? it's 1 megabyte. and it's cached locally (verified by sha1sum the image uuid)
20:53:47 mnaser https://www.irccloud.com/pastebin/s9lKj7n7/

Earlier   Later