21:00:14 #startmeeting nova 21:00:14 Meeting started Thu Apr 27 21:00:14 2017 UTC and is due to finish in 60 minutes. The chair is mriedem. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:15 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:18 The meeting name has been set to 'nova' 21:00:39 howdy pardners 21:00:40 o/ 21:00:45 o/ 21:00:50 o. 21:00:52 o/ 21:00:54 \o 21:01:09 \o/ 21:01:09 * bauzas waves 21:01:57 ok we can get started 21:01:59 #link agenda https://wiki.openstack.org/wiki/Meetings/Nova#Agenda_for_next_meeting 21:02:07 #topic release news 21:02:14 #link Pike release schedule: https://wiki.openstack.org/wiki/Nova/Pike_Release_Schedule 21:02:37 next big thing in the scheduler is the summit in a little more than 1 week 21:02:43 *schedule 21:02:44 :) 21:02:56 next milestone is p-2 on June 8th 21:03:11 #info Blueprints: 70 targeted, 65 approved, 8 completed, 11 not started 21:03:23 #link pike-1 recap and pike-2 focus: http://lists.openstack.org/pipermail/openstack-dev/2017-April/115700.html 21:03:29 #help There are still some approved blueprints looking for owners: http://lists.openstack.org/pipermail/openstack-dev/2017-April/115899.html 21:03:46 questions about the release? 21:04:01 mriedem I *may* be able to take on that other keystoney one that overlaps with service-catalog stuff. Keep me in mind if no other takers. 21:04:11 the service token one? 21:04:14 yah 21:04:20 ok. i think some of that is already done 21:04:22 but ok 21:04:32 #topic bugs 21:04:50 no critical bugs 21:05:02 #link check queue gate status http://status.openstack.org/elastic-recheck/index.html 21:05:06 gate has been ok 21:05:14 #link 3rd party CI status http://ci-watch.tintri.com/project?project=nova&time=7+days 21:05:22 i don't have any news about 3rd party ci 21:05:39 anything for bugs? 21:05:57 #topic reminders 21:06:13 #link Pike Review Priorities etherpad: https://etherpad.openstack.org/p/pike-nova-priorities-tracking 21:06:29 continue updating ^ as your changes progress 21:06:35 and subteams review things 21:06:41 #link Forum planning: https://wiki.openstack.org/wiki/Forum/Boston2017 21:06:49 #info Start creating etherpads for your sessions. There is an optional template: http://lists.openstack.org/pipermail/openstack-dev/2017-April/115972.html 21:06:58 the etherpads are linked into that wiki ^ 21:07:10 #link https://etherpad.openstack.org/p/BOS-Nova-brainstorming Forum discussion planning for nova (add your name if you are going) 21:07:35 i saw in the ML today that there are going to be some reservable rooms, 21:07:47 but i don't know if we have space in the schedule for anything that requires that 21:07:56 #link https://ethercalc.openstack.org/Boston_Forum_Hacking_Rooms 21:08:00 or any topics that aren't already covered in forum sessions 21:08:21 yeah, 21:08:31 personally i've found i don't have much in the way of free time given the sessions i'm already going to 21:09:04 if anyone can think of a major nova issue we need to discuss that isn't already covered in a session let me know 21:09:08 and we can see if there is a slot we need 21:09:18 #info Forum sessions: https://www.openstack.org/summit/boston-2017/summit-schedule/global-search?t=forum 21:09:32 anything about the forum? 21:09:45 #topic Stable branch status 21:09:51 stable/ocata: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/ocata,n,z 21:09:55 stable/newton: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/newton,n,z 21:10:11 we're working some fixes through those and then i'll be looking to do a patch release next week 21:10:20 i also need to check that for novaclient on stable 21:10:28 i have requested a 1.4.1 os-vif release for ocata this week 21:10:29 not merged yet 21:10:42 #topic subteam highlights 21:10:57 i'm filling in for cells v2 21:11:06 major review focus is on two things 21:11:11 1. quotas https://review.openstack.org/#/c/446239/ 21:11:22 2. multicell https://review.openstack.org/#/c/458634/ 21:11:29 a good chunk of the latter has merged 21:11:52 that's about it 21:11:59 edleafe: scheduler? 21:12:04 Short meeting this week, as nothing contentious was under discussion. 21:12:04 We got an update on the status of various placement specs and patches, and things seem to be moving along well. 21:12:07 cdent wished for additional help harassing jaypipes during his upcoming talk in Boston. 21:12:10 That's it! 21:12:22 talk(s) 21:12:30 there are multiple opportunities to ruin his day 21:12:44 I'm counting you to help with that, mriedem 21:12:47 i don't imagine tdurakov is around for live migration updates 21:13:11 there is a live migratoin related bp that's looking for an owner 21:13:20 and another that's looking for someone to take over the patches 21:13:36 https://blueprints.launchpad.net/nova/+spec/live-migration-force-after-timeout 21:13:40 https://blueprints.launchpad.net/nova/+spec/live-migration-per-instance-timeout 21:13:55 api meeting highlights 21:14:23 i was there but can't really remember 21:14:25 :/ 21:14:39 we talked about something for an inordinate amount of time 21:14:57 on the upside, microversions 2.43, 2.44 and 2.45 are merged now 21:15:09 gibi: notifications meeting highlights? 21:15:23 It was a sort meeting this week 21:15:23 Focus is on two things in the subteam 21:15:23 1) https://review.openstack.org/#/q/status:open+topic:bp/additional-notification-fields-for-searchlight 21:15:27 2) instance.volume_attach, the patch series starts here: https://review.openstack.org/#/c/401992/ 21:15:30 that is all 21:15:56 gibi: köszönöm 21:16:02 yeah? 21:16:08 szivesen 21:16:08 :) 21:16:16 efried: powervm? 21:16:20 Great progress since last week. Three ready for final reviews (one waiting for verification to start behaving again): https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/powervm-nova-compute-driver+is:mergeable Once one or two more of those go through, will start in on the others in the series again. 21:16:37 That's it from me, unless any questions/concerns. 21:16:57 ok 21:16:58 cool 21:17:12 for nova/cinder this week, 21:17:14 Review focus is on the new detach flow: https://review.openstack.org/#/c/456896/ 21:17:28 i've got a series up through swap volume, needs some tweaking, 21:17:57 others have some patches up for other operations dealing with detach, like instance delete 21:18:23 and i think someone (ildikov?) is going to work on resurrecting the attach patch with the new flow that makes everything go 21:18:37 #topic stuck reviews 21:18:44 there was nothing on the agenda, 21:18:49 anyone have anything to bring up here? 21:19:08 #topic open discussion 21:19:15 Special scenario/filter tests: http://lists.openstack.org/pipermail/openstack-dev/2017-April/115800.html (hferenc) 21:19:32 gibi: coworker of yours? 21:19:37 hferenc_: ^? 21:19:38 mriedem: yes 21:19:43 yes 21:19:48 thx 21:19:57 so in short: I'd like to add some testcases that couldn't be merged to tempest because they require a special environment for testing. 21:20:04 For me testing the aggregates isolation scheduler filter would be the most important but if it's possible I'd like to add more tests later on. 21:20:13 Sorry if this is not the right time and place to discuss this. I just wanted to shed some light on the topic. I'd appreciate any review. 21:20:26 hferenc_: so why can't we test this with in-tree functional tests? 21:20:31 using fixtures 21:20:36 wouldn't it be possible to use functional tests in-tree ? 21:20:40 meh 21:20:42 jinx 21:20:58 bauzas: great minds think alike, right? :) 21:21:10 :p 21:21:16 https://review.openstack.org/#/c/315786/ is something that you need devstack for https://review.openstack.org/#/c/315786/ 21:21:19 oops 21:21:20 It would be ok these with functional tests - i'm also working on that 21:21:30 i think the scheduler filters could be tested in tree 21:21:46 https://review.openstack.org/#/c/315786/ couldn't be 21:21:48 sure thing, we already have some 21:21:50 but functional tests cannot be run against deployments 21:22:03 hferenc_: it depends on what you need to test 21:22:23 if you're ok using fixtures for running services and a fake virt driver, just to test placement logic with the filters, i think in-tree is fine 21:22:33 if you're testing actual hypervisor features, then yes you need an openstack deployment 21:22:57 we have talked in the past about having a dsvm-integration suite of tests in-tree that relies on a backing devstack installation 21:23:02 for testing things like libvirt 21:23:11 tempest is really about testing the apis 21:23:19 so https://review.openstack.org/#/c/315786/ doesn't really make sense in a tempest plugin imo 21:23:58 i can see a lot of value in a dsvm-integration set of tests in-tree so we could run tests in serial, like then we could test evacuate 21:24:51 i guess your watchdog test goes through the api b/c of the flavor setup 21:25:43 so this dsvm-integration suite would spin a devstack but it would run some functional test instead of tempest 21:25:45 so that could probably actually be a tempest plugin test 21:25:51 gibi: yes 21:26:01 gibi: i think of it like the functional tests in novaclient 21:26:01 is there any example for that in other modules we can copy? 21:26:05 those rely on a devstack 21:26:07 ohh 21:26:16 and they run novaclient CLIs against that devstack 21:26:19 so it's not tempest 21:26:23 but it's not a fake env either 21:26:30 sounds interesting 21:26:41 i think we can test scheduler filters using functional tests in tree with fixtures, 21:27:05 and things like this watchdog action test could be done via a tempest plugin, or in-tree integration tests (like novaclient), 21:27:21 for the placement test, we have some requirement to prove the tenant separation works against a deployment (like user acceptance tests) 21:27:22 like we'd probably have a dsvm-integration-libvirt, dsvm-integration-ironic, something like that 21:28:08 so we made those filter tests in tempest to be able to used against a deployment 21:28:16 and we would like to share it :) 21:28:32 gibi: it's a very narrow and specific deployment is the problem 21:28:48 you can get the same thing using the in-tree functional tests 21:28:54 which run actual services and populate the database 21:29:07 they just don't use real cinder/neutron/virt driver, but you don't need those for scheduler placement tests 21:29:48 for example https://github.com/openstack/nova/blob/master/nova/tests/functional/regressions/test_bug_1671648.py 21:29:49 sure for the functional validation the current functional test env is ok 21:30:00 but for user acceptance it isn't 21:30:04 ^ tests that if we fail to build on a compute, we properly retry on another compute 21:30:14 but I feel we wont solve the user acceptance test problem in nova :) 21:30:41 what i'm hearing is you guys have an internal business process dictating how you do your testing 21:30:50 which, whatever, that's fine 21:31:31 i personally feel it's a waste to setup a CI job with a very specific environment and set of filters just to run a single test 21:31:38 when you can do it in-tree to test the same thing 21:32:01 if your job could reconfigure itself on the fly so it can test multiple scenarios, that's a different story 21:32:09 that's what our live migration job does 21:32:27 https://github.com/openstack/nova/blob/master/nova/tests/live_migration/hooks/run_tests.sh 21:32:31 most probably we want to much as a first step :) 21:32:51 I think we can look into the dsvm-integration thing for the watchdog test 21:33:02 so maybe we take this to the mailing list, i can dump some comments there 21:33:11 mriedem: sure 21:33:11 i'm not sure if there is anything you really need from me here, 21:33:23 you guys have a requirement and you've delivered on it for your corporate overlords 21:33:25 thanks for the feedback 21:33:49 in general i'm always happy with more testing, so +++ there 21:33:58 ok, anything else for open discussion? 21:34:17 mriedem: thank you 21:34:22 yw 21:34:28 alright let's wrap it up, thanks everyone 21:34:30 #endmeeting