16:14:53 <scottda> #startmeeting cinder_haaa
16:14:53 <openstack> Meeting started Tue Aug 16 16:14:53 2016 UTC and is due to finish in 60 minutes.  The chair is scottda. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:14:55 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:14:57 <openstack> The meeting name has been set to 'cinder_haaa'
16:15:09 <scottda> ping DuncanT, aspiers, geguileo, scottda, dulek, ddeja
16:15:17 <geguileo> scottda: o/
16:15:23 <scottda> ping patrickeast
16:16:21 <openstackgerrit> Nate Potter proposed openstack/python-cinderclient: Add "start_version" and "end_version" support to  argparse  https://review.openstack.org/340129
16:16:43 <scottda> So, it looks like the cinder multi-node config stuff merged for devstack and (I think) the cinder experimental job
16:17:10 <scottda> But I ran geguileo 's manual tests on the single node and that worked great as well.
16:17:38 <geguileo> scottda: I just pushed a new series today and updated the post and file to reflect it
16:17:44 <geguileo> scottda: Now all operations should be cluster aware
16:18:10 <scottda> Nice job on that geguileo . I was thinking of recording my next round of manual tests with those new patches ^^^ and putting out a webcast on youtube..
16:18:18 <scottda> geguileo: would that be OK with you?
16:18:40 <scottda> I did this for microversions, and some people thought it helped to understand the feature.
16:18:51 <geguileo> scottda: I haven't tested this last batch much
16:19:05 <geguileo> scottda: So I would hate to see the video of a failure
16:19:06 <geguileo> XD XD XD
16:19:06 <scottda> Although it's better for people to actually run the tests. You've done a great job of documenting how to do that...
16:19:18 <scottda> haha. IF it fails, I won't publish it.
16:19:24 <geguileo> scottda: I'm Ok if you want to do a video
16:19:34 <geguileo> scottda: But I agree that people should test it
16:19:34 <scottda> OK. I'll give you all the credit :)
16:20:01 <scottda> yeah. I think when people test this it removes a lot of fear of this feature...
16:20:21 <scottda> The cleanup, mainly. It really helped me to piece together how it was all working.
16:20:49 <geguileo> scottda: The credit of the video would be all yours  :-)
16:21:04 <scottda> geguileo: What were you going to ask smcginnis about job distribution spec? I'm just curious.....
16:21:13 <geguileo> scottda: Yeah, I'm planning on creating a good devref
16:21:34 <geguileo> scottda: To remove part of that fear and to make sure that new code doesn't break things
16:21:47 <geguileo> scottda: Oh, what I was going to ask him is that we have a spec merged in Newton
16:21:52 <geguileo> scottda: But it's not up to date
16:22:04 <geguileo> scottda: So I pushed some time ago an update for review
16:22:13 <geguileo> scottda: And now it seems that it should be moved to Ocata
16:22:28 <geguileo> scottda: But I think that would be a mistake, because the spec in Newton does not match the code in Newton
16:22:53 <geguileo> scottda: And then we would have the same spec in both Newton and Ocata, and the Newton one would be just wrong
16:23:01 <scottda> #link https://review.openstack.org/#/c/232595/
16:23:06 <geguileo> So I wanted to make sure that's what we wanted to do
16:23:07 <patrickeast> ew, yea lets just update the N one
16:23:21 <scottda> yeah, that's confusing. I'll let smcginnis  figure that one out :)
16:23:39 <scottda> but +1 to update N
16:23:40 <geguileo> #link https://review.openstack.org/#/c/327283
16:23:55 <geguileo> ^ That's the update
16:24:15 <scottda> ahh, that one.
16:25:17 <scottda> patrickeast: Anything to report on your CI efforts running HA stuff?
16:25:57 <patrickeast> scottda: seems to be fine, its in a quiet mode right now since I'm fiddling around with trying to get it to say UNSTABLE when its just devstack or somethings fault
16:26:17 <patrickeast> got a bunch of FC HBA's added in so I can start doing multi-node FC junk too
16:26:28 <geguileo> Nice!
16:26:37 <scottda> kk. Great that you are doing this testing for HA with your CI
16:27:53 <scottda> geguileo: Do you think we should still be looking at multi-node for some automated testing, or does your single-node scenario seem adequate for a start?
16:28:27 <geguileo> scottda: I think multinode is important to test the DLM part as well
16:28:51 <geguileo> scottda: Since my tests were meant to tests my patches in isolation we were fine with single node
16:29:06 <geguileo> scottda: But the next step should really be to test this in multi-node with a DLM
16:29:22 <geguileo> Because we have to get the DLM part properly tested as well
16:29:29 <scottda> yup
16:29:43 <patrickeast> one thing i've got on my backlog eventually is to test multi-node and see what kind of damage is done when killing a node mid-data-transfer for backup, generic migration, etc
16:30:04 <patrickeast> if anything just to have instructions for customers who get stuck and need help cleaning up the system
16:30:37 <geguileo> patrickeast: That is really something that important
16:30:57 <geguileo> patrickeast: Because uncertainty is not nice  ;-)
16:31:10 <scottda> That'd be cool. I can envision that we might evolve geguileo 's cleanup code to cover more use cases, like the backup, migration, etc...
16:31:37 <geguileo> scottda: We certainly should, since some basic stuff is not even covered
16:31:54 <geguileo> scottda: I think some other people will want to contribute on that
16:32:25 <scottda> Yeah, I guess that's really part of an overall cinder HA story. Something to think about in the future...
16:32:41 <geguileo> But I'll have to create a proper devref as hemna suggested so people know exactly what things need to change for adding a new cleanup status
16:33:14 <geguileo> Because it's not trivial if you are not really familiar with the code and the mechanism
16:33:20 <geguileo> But I'll get there
16:33:22 <geguileo> XD
16:33:35 <scottda> Your efforts are beyond reproach :)
16:34:20 <scottda> Anything else today?
16:35:38 <geguileo> Not from me
16:35:51 <geguileo> I'll try to have the API cleanup enpoint by the next meeting
16:36:13 <scottda> great! We'll call it a meeting then...
16:36:15 <scottda> #endmeeting