Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-02-08
16:43:14 johnthetubaguy ah, got you
16:43:33 hrw johnthetubaguy: like I told earlier: feature-matrix-gp.ini/operation.create-delete-server lists two tests: 9a438d88-10c6-4bcd-8b5b-5b6e25e1346f (test_list_servers) and 585e934c-448e-43c4-acbf-d06a9b899997 (test_list_servers_with_detail) and I lack both in any tempest run logs
16:44:01 hrw johnthetubaguy: so if I go right then create-delete functionality is missing, right??
16:46:03 johnthetubaguy hrw: correct, if its not tested in the gate, you would mark it as missing
16:46:58 johnthetubaguy by gate I mean 3rdparty I guess
16:47:33 hrw johnthetubaguy: http://logs.openstack.org/28/541728/8/check/build-openstack-sphinx-docs/b9ac718/html/user/feature-classification.html
16:48:49 hrw johnthetubaguy: table is bogus for aarch64: is marked as missing or unknown despite working. just because tests used in ini files test for other stuff than names say
16:49:04 hrw johnthetubaguy: or listed tests are not used anymore
16:49:32 johnthetubaguy hrw: yeah seems like its bit rotted
16:49:35 hrw hm. no volumes in my setup
16:49:37 hrw (openstack) volume create --image f91ae3a7-e067-466c-b1b3-e592ea58d266 --size 8 --availability-zone nova my-new-volume
16:49:40 hrw internal endpoint for volumev2 service in RegionOne region not found
16:50:03 johnthetubaguy probably worth deleting at some point...
16:50:17 hrw johnthetubaguy: so instead of filling table with 'sorry, but nothing works here' I prefer to fill support-matrix one with 'this works, this does not and that is not tested'
16:50:42 johnthetubaguy hrw: yes makes sense
16:52:09 johnthetubaguy separate thing to work out how this matches up with interop tests and the gate, sounds worrying
16:52:39 mriedem ameeda: i don't know what the "test" part is for in bindep, probably need to ask about that in #openstack-infra
16:53:25 mriedem hrw: http://git.openstack.org/cgit/openstack/tempest/tree/tempest/api/compute/servers/test_create_server.py#n87 does test create/delete
16:53:40 mriedem hrw: the test class setup creates some servers, and the teardown deletes those servers
16:53:46 mriedem http://git.openstack.org/cgit/openstack/tempest/tree/tempest/api/compute/servers/test_create_server.py#n54
16:54:22 mriedem johnthetubaguy: hrw: interface attach/detach is specifically neutron only
16:54:48 mriedem https://developer.openstack.org/api-ref/compute/#create-interface
16:54:52 mriedem https://developer.openstack.org/api-ref/compute/#detach-interface
16:54:54 hrw mriedem: then why feature-matrix-gp lists list_servers instead of server_create etc?
16:55:10 mriedem hrw: likely because it's also a defcore interop test
16:55:26 mriedem any tempest compute api test that has a server in it creates a server
16:55:29 mriedem so they probably just picked one
16:55:43 hrw https://refstack.openstack.org/#/results/952ebce2-af82-4ed6-81ba-860e80d38af4 is one of our latest runs. 75% passed.
16:56:24 hrw anyway... let me check what other entries I am missing in support matrix
16:56:47 mriedem my guess is if tempest is blowing up on your CI run, things aren't getting cleaned up properly and you're seeing cascading failures
16:56:58 mriedem if you just run tempest.api.compute.servers.test_create_server.ServersTestJSON.test_list_servers it should be fine
16:57:00 hrw other than volumes which for some reason does not work on my setup
16:57:04 hrw mriedem: possible
16:58:25 hrw found. no cinder ;D
17:01:33 johnthetubaguy mriedem: yeah I think we picked interop ones over any other
17:02:14 johnthetubaguy hrw: the basic one doesn't use volumes I thought
17:02:30 mriedem it doesn't
17:02:40 mriedem he's just saying that's why he's "unknown" about anything wrt volumes
17:03:02 hrw anyway redeploying with cinder now
17:03:20 johnthetubaguy ah
17:03:42 mriedem yeah you kind of need cinder for volumes
17:03:53 mriedem unless you're still running essex somewhere
17:04:24 hrw essex for me is some part of uk or sth like that ;D
17:04:55 openstackgerrit Eric Fried proposed openstack/nova master: Ensure resource classes correctly https://review.openstack.org/539738
17:04:55 openstackgerrit Eric Fried proposed openstack/nova master: New-style _set_inventory_for_provider https://review.openstack.org/537648
17:04:56 openstackgerrit Eric Fried proposed openstack/nova master: SchedulerReportClient.update_from_provider_tree https://review.openstack.org/533821
17:04:57 openstackgerrit Eric Fried proposed openstack/nova master: Use update_provider_tree from resource tracker https://review.openstack.org/520246
17:04:57 openstackgerrit Eric Fried proposed openstack/nova master: Fix nits in update_provider_tree series https://review.openstack.org/531260
17:04:57 openstackgerrit Eric Fried proposed openstack/nova master: Move refresh time from report client to prov tree https://review.openstack.org/535517
17:04:58 openstackgerrit Eric Fried proposed openstack/nova master: Make generation optional in ProviderTree https://review.openstack.org/539324
17:04:58 openstackgerrit Eric Fried proposed openstack/nova master: WIP: Add nested resources to server moving tests https://review.openstack.org/527728
17:06:07 stvnoyes mriedem: fyi - I just updated that bug on multiattach + multi-create. The next issue is stickier than the first one. https://bugs.launchpad.net/nova/+bug/1747985
17:06:08 openstack Launchpad bug 1747985 in OpenStack Compute (nova) "Create Multiple Servers fails when a volume is specified even with a multiattach volume" [Low,Confirmed]
17:06:26 bauzas mriedem: diablo, not essex
17:06:34 bauzas mriedem: cinder was a thing by essex
17:06:45 mriedem nova-volume still existed in essex
17:06:54 smcginnis Split out in folsom.
17:07:05 mriedem check and mate
17:07:09 efried jaypipes: gory manual rebase on ==> Ensure resource classes correctly https://review.openstack.org/539738 No functional change, ready for re-review (ditto remainder of series)
17:07:13 bauzas wait
17:07:25 bauzas nova-volume was the thing in Essex ?
17:07:34 bauzas I thought it was diablo
17:07:37 smcginnis Trivial Persuit - OpenStack Edition :)
17:07:47 bauzas the only change I made when upgrading to Folsom was to deploy Quantum
17:08:01 bauzas I don't recall needing to spin cinder
17:08:07 mriedem https://github.com/openstack/nova/blob/essex-eol/bin/nova-api-os-volume
17:08:41 bauzas mmm, I should check the governance files directly and see if that's just an incubated/integrated question
17:08:46 mriedem https://github.com/openstack/cinder/tree/essex-eol
17:08:57 hrw the first version of openstack I touched code was Liberty. Mitaka/Newton were ones I packaged. Ocata+ were ones I contributed
17:09:34 openstackgerrit Merged openstack/osc-placement master: Update doc link in README.rst https://review.openstack.org/542002
17:09:50 mriedem stvnoyes: re comment 2, yes that's what i was saying when i thought this would fail, but with a multiattach volume we should be able to create >1 attachment to that same volume
17:10:09 mriedem stvnoyes: or is the issue that we create the first attachment, but we can't create the 2nd attachment until the 1st attachment is 'completed'?
17:10:44 stvnoyes currently cinder won't allow you to attach to a reserved volume. has to be in-use or available
17:11:08 stvnoyes so yes, the attach has to be completed
17:11:19 bauzas snap, I dislike to be wrong https://wiki.openstack.org/wiki/ReleaseNotes/Folsom#OpenStack_Block_Storage_.28Cinder.29
17:11:48 mriedem stvnoyes: we could easily recreate by adding a min_count=2 to the server create call in this tempest test https://github.com/openstack/tempest/blob/70ebe69cc2e4fa8a06071d48d19f61d2f01a33f6/tempest/api/compute/volumes/test_attach_volume.py#L393
17:12:02 mriedem i can push a patch quick
17:12:16 bauzas probably because migrating from nova-volumes to cinder was seamless - compared to deploying Quantum *cough* *cough*
17:12:18 stvnoyes yeah I have a temptest test already to do this. I can put it up for rv if you like
17:12:53 smcginnis bauzas: Hah
17:13:17 ameeda mriedem: Thank you, please check the channel , they answered me
17:14:28 stephenfin dansmith: Alright, so I think I've figured this all out, if you're curious
17:14:41 stephenfin this being the RPC versioning for those functions
17:15:03 stvnoyes mriedem: let me know if you want me to post a review for the multi-create + multi-attach tempest test I have. (Once you go > 1 vms you'll need to set validated=False on server_create.)
17:15:34 dansmith stephenfin: shoot
17:15:55 mriedem stvnoyes: easier than this? https://review.openstack.org/542322
17:16:34 stephenfin dansmith: So the issue isn't with the HostState object any more - it's with the HostState.numa_topology attribute
17:16:34 ameeda mriedem: FYI: ameeda: in that case "test" is just an arbitrary profile name we've settled on to indicate that it's not a runtime dependency just a test dependency
17:16:51 stvnoyes mriedem: that's pretty good.
17:17:34 mriedem ameeda: that's what i figured, so yeah just use that, we only need this dep for building docs
17:17:48 stephenfin dansmith: Anything that plays with HostState objects needs to call functions like this https://github.com/openstack/nova/blob/master/nova/scheduler/filters/numa_topology_filter.py#L32-L33
17:18:08 stvnoyes Though I was using the volume as a data disk, not a boot disk. just to avoid the corruption on a m-a boot disk.
17:18:21 stephenfin That's because we set HostState.numa_topology to a value that we're retrieving from ComputeNode objects (ComputeNode.numa_topology)
17:18:21 ameeda mriedem: cool :)
17:18:39 stephenfin and, for legacy reasons, the ComputeNode.numa_topology attribute is a JSON blob
17:19:22 stephenfin dansmith: I've traced the flow here http://paste.openstack.org/show/666746/
17:20:39 stephenfin dansmith: So, basically, if we want to _really_ get rid of all that crap, we need to bump the version of ComputeNode to 2.0 and store objects, or add some wrapper that will auto-convert the JSON to an object
17:20:55 dansmith stephenfin: that's going to be a lot of work
17:21:00 stephenfin Yup :(
17:21:11 dansmith definitely not happening as part of this effort before GA

Earlier   Later