Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-02-16
16:43:48 BobBall Does you set vif_driver to the quark plugin?
16:43:50 johnthetubaguy uses ovs
16:43:54 BobBall or can we deprecate the vif_driver option in nova?
16:44:13 openstackgerrit Sujitha proposed openstack/nova master: Adding tags field to InstancePayload https://review.openstack.org/407228
16:44:58 johnthetubaguy BobBall: seems fine to me
16:45:07 mriedem heh the config option says this:
16:45:14 mriedem "NOTE: The XenAPIBridgeDriver should be used for running OVS or Bridge in XenServer."
16:45:22 mriedem i guess that's totally wrong?
16:45:47 BobBall heh... Indeed.
16:47:01 BobBall frickler: There was a new tempest test added which made use of floating IPs. I've excluded it now. Please re-check.
16:50:09 BobBall mriedem: sorry for the sadness in the XenServer CI - as mentioned to frickler above, we have identified the probable cause for the ssh timeouts (we had two neutron agents with the same hostname - which meant sometimes the wrong agent was picked for something or other...). A fix was made today, but further investigation is being done to prove it doesn't break secure groups (there was some suggestion yesterday it might, but today we believe it won't).
16:50:35 mriedem BobBall: that's fine, i know it's not failing on this patch
16:51:56 BobBall Indeed. There are many false negatives today - but rechecks should fix them now. Hopefully we'll have the proper fix in the CI early next week and we will re-enable these tests.
16:53:44 openstackgerrit Sean Dague proposed openstack/nova master: WIP: verify project_id when quotas are checked https://review.openstack.org/435010
16:54:05 sdague mriedem: that kind of works, no tests, I'll update the spec with the differences
16:57:50 mriedem BobBall: changing the default vif driver causes all sorts of unit test failure of course
16:59:03 BobBall mriedem: would it be enough to add help text in the vif driver, a release note and a log message when loading the deprecated one stating that it is deprecated? Then I can change the default, fix the UT and remove the old one in Pike?
16:59:42 mriedem yeah i think i'm going to leave the default and mark the option deprecated for pike - we could also just hack the tests to use the bridge driver to pass for now
16:59:49 mriedem i'm trying to find a spot to do that
16:59:59 openstackgerrit Sean Dague proposed openstack/nova-specs master: Updates for validation spec based on implementation https://review.openstack.org/435016
17:00:04 sdague mriedem: ^^^
17:03:27 cfriesen for nova microversion 2.42 is there going to be any work required in novaclient?
17:04:38 mriedem cfriesen: it's a bug fix, andreykurilin had a patch
17:04:56 mriedem https://review.openstack.org/#/c/424745/
17:06:06 andreykurilin mriedem: also there is a patch for 2.42 - https://review.openstack.org/#/c/429512/ . I suppose we need to merge 424745 change with 429512
17:06:08 mdbooth bauzas: https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L1049-L1053
17:06:33 mdbooth bauzas: That's the info that eventually ends up in the scheduler, right?
17:06:44 mriedem BobBall: got a workaround for the tests, pushing the change up shortly
17:07:01 BobBall Ah awesome - I was just looking into that.
17:07:12 mdbooth If so, I think the bug is that it should be looking at bdms, otherwise it doesn't know the actual sizes (or existence) of the root and ephemeral disks
17:07:18 mdbooth Also, it takes no account of swap
17:07:27 mdbooth Or kernel/ramdisk
17:07:44 mdbooth bauzas: Can you sanity check^^^?
17:09:09 cfriesen mriedem: andreykurilin: thanks, was hoping someone had done it since I'm proposing a change for 2.43. :)
17:10:21 andreykurilin cfriesen: unfortunately, I'll not have an ability to finish my change till March :(
17:11:45 cfriesen andreykurilin: okay. I see your change handles the fact that tags don't work for certain versions...what about handling the fact that they should work again as of 2.42?
17:12:12 andreykurilin cfriesen: it should be handled here - https://review.openstack.org/#/c/429512/
17:15:20 cfriesen andreykurilin: right, so now you accept 2.42. but doesn't the change at https://review.openstack.org/#/c/424745 mean that we won't use it?
17:15:36 cfriesen won't use tags, I mean, even though they're supported on the server side
17:16:39 andreykurilin cfriesen: change 424745 should clarify situation for microversions <2.42 . The change 429512 should return support of tags with 2.42
17:18:08 cfriesen 429512 would work standalone because it's assuming "there are no version-wrapped shell method changes for this". However, change 424745 introduces a version-wrapped change.
17:18:53 cfriesen so if one goes in first the other one has to be reworked to account for it otherwise tags won't work for version 2.42. (I think, anyway. I could be wrong.)
17:24:12 openstackgerrit Matt Riedemann proposed openstack/nova master: Deprecate xenserver.vif_driver config option and change default https://review.openstack.org/435037
17:24:19 mriedem BobBall: johnthetubaguy: ^
17:24:29 mriedem dansmith: sdague: ^ let me know if you think that's wacky
17:24:35 mriedem but i kind of want to get that into ocata to set the timer
17:27:09 sdague mriedem: I'd be fine with it. I think that given the lower level of xenserver usage, things like that can move a bit more nimbly
17:27:20 dansmith mriedem: why is it whacky?
17:27:22 cfriesen andreykurilin: mriedem: would it make sense to push https://review.openstack.org/#/c/429512 for Ocata so that the client/server match up on their max version support? Then we can go back and fix up the intermediate versions where tags were broken.
17:27:26 dansmith er, wacky.. heh
17:27:33 mriedem dansmith: just b/c i want to get it into ocata
17:27:37 mriedem not really wacky
17:28:01 mriedem cfriesen: huh?
17:28:04 dansmith seems decidedly unwacky to me
17:28:12 mriedem cfriesen: novaclient is frozen for ocata
17:28:16 mriedem except for backporting fixes
17:28:47 sdague BobBall: mriedem: why is this failing xenserver ci? - https://review.openstack.org/#/c/424428/12
17:29:11 cfriesen mriedem: so we have nova with max version 2.42 but novaclient with 2.41?
17:29:38 mriedem sdague: everything is
17:29:39 cfriesen mriedem: and no way to make use of the fact that 2.42 fixes tags?
17:29:40 mriedem known issue
17:29:53 mriedem cfriesen: novaclient will work with 2.42
17:30:05 mriedem oh i guess we didn't set the max api thing in novaclient
17:30:08 cfriesen API_MAX_VERSION = api_versions.APIVersion("2.41")
17:30:22 cfriesen that's what https://review.openstack.org/#/c/429512 does
17:30:26 mriedem yeah, well, we can't release novaclient for ocata right now
17:30:37 mriedem cfriesen: i think we'll just need to backport that as a bug fix later
17:30:39 mriedem to stable/ocata
17:31:10 sdague mriedem: ok, so are we waiting for it to work?
17:31:18 sdague or is it safe to push it in
17:31:52 mriedem sdague: it's safe
17:32:07 mriedem the xenserver ci uses the ovs vif driver already
17:32:11 mriedem i checked that earlier
17:32:20 mriedem b/c the xenserver ci on master is neutron, and has to use the ovs driver
17:34:06 sdague mriedem: eta on xenserver ci coming back to life?
17:34:32 mriedem sdague: (10:50:14 AM) BobBall: mriedem: sorry for the sadness in the XenServer CI - as mentioned to frickler above, we have identified the probable cause for the ssh timeouts (we had two neutron agents with the same hostname - which meant sometimes the wrong agent was picked for something or other...). A fix was made today, but further investigation is being done to prove it doesn't break secure groups (there was some suggestion yesterd
17:34:39 mriedem (10:52:02 AM) BobBall: Indeed. There are many false negatives today - but rechecks should fix them now. Hopefully we'll have the proper fix in the CI early next week and we will re-enable these tests.
17:35:07 openstackgerrit Roman Podoliaka proposed openstack/nova-specs master: Strict isolation of group of hosts for image and flavor https://review.openstack.org/381912
17:36:12 mriedem sdague: so maybe results within the hour or two?
17:36:25 mriedem sdague: keeping in mind i need to backport this to ocata and get it merged there before we can cut rc2 today
17:37:53 mriedem sdague: oh you got the bottom change too, thanks
17:37:59 mriedem fwiw john was +2 on that yesterday
17:38:07 mriedem i just found some corner cases that needed updating, so the bottom change is safe
17:38:13 sdague mriedem: yeh
17:51:28 mriedem backports are up for those xen patches
17:51:30 mriedem going to lunch
17:51:51 openstackgerrit Ben Nemec proposed openstack/nova-specs master: Configurable PXE Boot for Instances https://review.openstack.org/435052
18:12:36 openstackgerrit Sujitha proposed openstack/nova master: Change tags to default field in Instance object. https://review.openstack.org/415298
18:12:48 openstackgerrit MarkMielke proposed openstack/nova master: DiskFilter should only apply when storage is requested https://review.openstack.org/430091
18:18:27 openstackgerrit Jay Pipes proposed openstack/nova master: placement: rename can_host to shared https://review.openstack.org/433977
18:18:27 openstackgerrit Jay Pipes proposed openstack/nova master: placement: get inventory shared with provider https://review.openstack.org/433978
18:25:22 openstack Launchpad bug 1581230 in OpenStack Compute (nova) "Nova should not inspect cinder volume state during attach" [Medium,In progress] - Assigned to Ildiko Vancsa (ildiko-vancsa)
18:25:22 jaypipes mriedem1: on https://bugs.launchpad.net/nova/+bug/1581230, walt asks "Nova shouldn't care about the internal state management of Cinder Volumes. What happens if Cinder adds a new state that Nova doesn't know about?"
18:25:45 jaypipes mriedem1: I guess I'm wondering why internal state is being exposed out of Cinder's API, then?
18:26:24 mriedem jaypipes: well it's true that nova doesn't need or want to check the attached state of a volume if we're just going to reserve it, because the reserve action will do that atomically for us on the cinder side
18:26:31 mriedem the resulting error from the nova side is the same either way
18:26:46 jaypipes mriedem: k.
18:26:47 mriedem unreserve on the cinder side is idempotent
18:27:03 mriedem i think all of that old attach status checking is probably from nova-volume
18:27:06 jaypipes mriedem: cool (as it should be)

Earlier   Later