Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-06-12
14:35:23 stephenfin tl;dr: I can't decide if either the Python patch or pycodestyle is broken, but it's the edgiest of edge cases that we'd only really see in tests, so we can work around it
14:37:51 mriedem rafaelweingartne: no,
14:37:58 rafaelweingartne hmm
14:37:59 mriedem the instance mappings are cleaned up when you archive
14:38:07 mriedem nova-manage db archive_deleted_rows
14:38:35 mriedem rafaelweingartne: 2.21 is about instance actions / events
14:39:07 mriedem rafaelweingartne: https://docs.openstack.org/python-openstackclient/latest/cli/command-objects/server-event.html
14:39:08 gibi stephenfin: what a detective work!
14:42:46 dansmith mriedem: can you quickly weigh in on this? I thought you might care about the use of a sentinel value in nova-manage. Otherwise I'm good with it: https://review.opendev.org/#/c/662383/5
14:45:40 mriedem ew
14:47:11 dansmith before you shame him, note that it was my idea :)
14:48:53 mriedem i'm just trying to think of alternatives,
14:49:19 mriedem i saw melwitt pointed out https://review.opendev.org/#/c/603998/ but i'm not sure what that change is doing,
14:49:27 mriedem with takashi's change,
14:50:10 mriedem if you specify one but not the other it's an error,
14:50:38 mriedem i was thinking, what if we change the logic such that if you specify one but not the other, we just don't read from the config for the other (don't set/update the other)
14:50:48 mriedem if you don't specify either, we use config,
14:50:54 mriedem if you specify both, we set/update both,
14:50:56 dansmith that just seems confusing to me
14:50:59 mriedem if you only provide one, we only use one
14:51:14 mriedem i think that's more in line with how osc set commands work
14:51:27 dansmith I was actually expecting to have a --transport_url=(keep|config|[url]) sort of thing, but..
14:52:02 dansmith the confusing bit is that we do the magic config thing if we don't specify either.. if we specify one, then we do a totally different set of things
14:52:32 mriedem i'm assumine when this was written, we expected people were either using all of the command options or none to let config handle it
14:52:33 mriedem not a mix
14:52:39 dansmith this has tripped up more than one person who was surprised by one or the other behaviors
14:52:54 dansmith well,
14:53:06 dansmith I think it actually came from create, where that makes more sense
14:53:14 dansmith and we imported the same behavior for update where it does not
14:54:39 mriedem yeah ok i can see that,
14:54:57 mriedem so in that case i think it makes more sense to only update the field provided via the option if only one option is provided, and ignore the other (don't use config)
14:55:46 dansmith le sigh
14:57:58 mriedem hey, you asked me
14:58:03 mriedem anyway, i commented
14:58:16 mriedem if we need a tie breaker i vote that we ask dean to weigh in
14:58:37 mriedem mean dean okerlund
14:58:40 mriedem from WWF fame
15:07:42 openstackgerrit Stephen Finucane proposed openstack/nova master: Fix double word hacking test https://review.opendev.org/664940
15:08:43 bauzas sean-k-mooney: hey
15:08:54 bauzas sean-k-mooney: yeah I tested it and it worked
15:09:05 bauzas ie. stopping the instance and then restarting it
15:09:08 stephenfin efried: I need to rebase this series now, right? https://review.opendev.org/#/c/651311/
15:09:19 stephenfin to pick up the fix to test_hacking.py
15:09:37 sean-k-mooney bauzas: ok looking at the code im not sure where we figureout the mdevs in that case but i also didnt look
15:09:41 sean-k-mooney to far
15:09:49 bauzas sean-k-mooney: I provided a change for that
15:09:52 bauzas sec
15:10:09 sean-k-mooney my concern was we had to hit the placmenet api to look up the allcoation again
15:10:17 sean-k-mooney which i dont think we shoudl have to do.
15:10:57 efried stephenfin: I wouldn't think you need to rebase; I think zuul automatically rebases you against tip of master before starting tests
15:11:04 bauzas sean-k-mooney: https://review.opendev.org/#/c/564257/
15:11:29 bauzas and https://review.opendev.org/#/c/533642/
15:11:41 sean-k-mooney bauzas: that wont fix it
15:12:00 sean-k-mooney the second on might
15:12:09 stephenfin efried: Yeah, I'm not sure. That said, I think I'm going to have to recheck every one of them again anyway so maybe a rebase would be the surer thing
15:12:30 stephenfin Yeah, they're all -2
15:12:49 efried stephenfin: Yeah, that would be an easy way to get them all back in the queue. Feel free to re+W the ones that were already +A.
15:12:58 sean-k-mooney bauzas: quickly looking i dont think either of those will fix the edgecase i am thinking of
15:13:09 bauzas sean-k-mooney: mmm N
15:13:10 bauzas ?
15:13:39 sean-k-mooney bauzas: if the instance is not found on the host in libvirt you retrun {]
15:13:39 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove cells v1 parameter from 'ComputeTaskAPI.resize_instance' https://review.opendev.org/651311
15:13:39 openstackgerrit Stephen Finucane proposed openstack/nova master: Stop passing 'kwargs' to 'rebuild_instance' https://review.opendev.org/651312
15:13:40 openstackgerrit Stephen Finucane proposed openstack/nova master: Stop passing 'delete_type' to 'terminate_instance' https://review.opendev.org/651313
15:13:40 openstackgerrit Stephen Finucane proposed openstack/nova master: filters: Stop handling cells v1 https://review.opendev.org/651314
15:13:41 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove nova.compute.*API() shims https://review.opendev.org/660527
15:13:41 openstackgerrit Stephen Finucane proposed openstack/nova master: Add reno for removed cells v1 policies https://review.opendev.org/662031
15:13:42 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove 'InstanceUnknownCell' exception https://review.opendev.org/662411
15:13:42 openstackgerrit Stephen Finucane proposed openstack/nova master: Ensure controllers all call super https://review.opendev.org/660950
15:13:47 sean-k-mooney bauzas: we can talk after standup
15:18:08 sean-k-mooney bauzas: basically if we stop the instance then we undefine the domain so if start it when we call _get_all_assigned_mediated_devices
15:18:18 sean-k-mooney bauzas: we retrun {}
15:18:41 bauzas I'm pretty sure I tested it
15:18:44 stephenfin mriedem: Are you okay for me to rebase https://review.opendev.org/#/c/621061/ onto the cells v1 removal series (all but one +W'd now). It conflicts with that series and I'd like to use it as a new base for https://review.opendev.org/#/c/651315/
15:22:02 mriedem sure
15:22:19 mriedem maybe you want to re-apply your +2 when doing so
15:39:57 Sundar sean-k-mooney: Re. tempest CI with a fake driver/device, I have some questions
15:41:21 Sundar sean-k-mooney: When there is no real device, the VM bringup is going to fail. Not only that, there would be no VM in 'virsh list', so we can't do 'virsh dumpxml ...' to see if the right device(s) got attached.
15:41:37 Sundar So, how would would one do any testing?
15:42:45 sean-k-mooney Sundar: there are a few ways. one we can use dummy devices
15:43:10 sean-k-mooney but the intent is to test teh end to end workflow not assert a spcific device exits
15:43:35 Sundar What would be the success metric for the test?
15:46:43 mriedem Sundar: tempest isn't going to be doing virsh list anyway
15:46:57 mriedem tempest tests should be hypervisor agnostic, like the compute api should be hypervisor agnostic
15:47:27 mriedem Sundar: the basic success metric for tempest tests are going to be that the API behaves as expected,
15:47:49 mriedem i.e. if i create a server with a flavor configured for an ARQ device profile, that once the server is ACTIVE there is an attached ARQ in cyborg
15:48:02 mriedem and when the server is deleted, the corresponding ARQ resource is also deleted
15:49:08 mriedem Sundar: whitebox integration testing is something that would likely be done in a tempest plugin / tests that are configured to only run when there is real hardware in a 3rd party CI env
15:50:32 Sundar mriedem: "once the server is ACTIVE" -- with a fake device, it never becomes active. It will go to error state.
15:51:20 mriedem Sundar: because of the event not being sent from cyborg, or because libvirt will fail to start the domain?
15:51:26 Sundar We need to ensure the device gets attached to the VM
15:51:28 sean-k-mooney Sundar: not if we add a bus type for dummy devices and handel that in nova
15:51:54 Sundar mridem: latter - libvirt will fail it
15:52:13 mriedem then i agree with sean-k-mooney that we'd need to stub something out if we know we're using a fake device,
15:52:20 sean-k-mooney Sundar: we dont we jsut need to pass info to nova to let it know its a fake device and libivirt can just not try to atach it
15:52:22 mriedem which is gross, but i'd rather have that than no api integration testing
15:52:33 Sundar sean-k-mooney: 'bus type for dumy devices' -- is this a new thing? It is not there today, right?
15:52:54 mriedem wouldn't there be some metadata on the ARQ resource that can tell us (nova) that it's a fake device?
15:52:55 sean-k-mooney sure but its useful for testing
15:53:03 mriedem similar to vif type on a port

Earlier   Later