13:01:01 <esberglu> #startmeeting powervm_driver_meeting
13:01:02 <openstack> Meeting started Tue Jun 13 13:01:01 2017 UTC and is due to finish in 60 minutes.  The chair is esberglu. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:01:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:01:05 <openstack> The meeting name has been set to 'powervm_driver_meeting'
13:01:26 <edmondsw> o/
13:02:46 <edmondsw> efried thorst_afk you here?
13:02:53 <thorst_afk> kinda here
13:03:19 <esberglu> #topic In Tree Driver
13:03:28 <esberglu> #link https://etherpad.openstack.org/p/powervm-in-tree-todos
13:03:53 <edmondsw> I have been working on a couple patches
13:04:09 <edmondsw> https://review.openstack.org/#/c/471773/
13:04:14 <edmondsw> https://review.openstack.org/#/c/472794/
13:04:47 <edmondsw> second was at efried's request via comment in the first, but I think should probably go first
13:05:34 <edmondsw> he just gave his comments yesterday, which I plan to go through today and then ping a couple cores for their thoughts, make sure this is acceptable before I invest a lot of time in UTs
13:05:51 <esberglu> Yeah that order makes sense from a quick glance
13:06:32 <edmondsw> other than that, we got a +2 on esberglu's support matrix patch... got a link for that?
13:06:47 <esberglu> https://review.openstack.org/#/c/470999/
13:06:55 <edmondsw> that's the one
13:06:55 * thorst_afk is now afk
13:07:38 <esberglu> Anything we need discussion on in tree?
13:07:39 <edmondsw> I'll add mriedem and sdague on that and see if we can get it merged
13:07:50 <esberglu> edmondsw: Sweet
13:08:42 <edmondsw> done
13:09:16 <edmondsw> I'm sure efried would have more to talk about here... maybe we come back to it when he shows up
13:09:18 <edmondsw> next topic?
13:09:20 <esberglu> Yep
13:09:30 <esberglu> #topic Out of Tree Driver
13:10:34 <esberglu> Not sure how much activity has been going on here, I've been mostly on CI the last handful of days
13:10:35 <efried> Sorry, previous "meeting" ran over.
13:10:46 <edmondsw> I've got a change out for pypowervm, 5421, that could use some reviews
13:10:58 <esberglu> efried: np. You have anything you want to talk in tree quick before we move to out of tree?
13:11:09 <edmondsw> I guess that's as much IT as OOT, or not really either
13:11:44 <efried> What's the prognosis for hypervisors devref?
13:12:24 <esberglu> I was gonna start that in the background today while I try to get stacks working
13:12:31 <efried> Cool cool.
13:12:41 <efried> I'll be interested to see where that ends up, and what it looks like.
13:12:52 <efried> Do you have experience with RST?
13:13:02 <esberglu> Nope but there are plenty of examples
13:14:10 <efried> Yup; remind me when you get started and I'll show you a couple of tricks for rendering so you can test it out and make sure it's a) valid, and b) doing what you want.
13:14:27 <esberglu> efried: Ok sounds good
13:14:44 <edmondsw> I'll have to get you to show me those tricks as well... or better yet, someone should make sure they're documented somewhere and just link :)
13:15:01 <edmondsw> I know how to see things once they're in gerrit, but not before
13:15:28 <efried> Yeah, that's the most reliable one, since that's how it's actually going to be rendered.
13:15:34 <efried> The other one is to use rst2html.py
13:16:09 <edmondsw> esberglu once it's up in gerrit, just click the docs gate link
13:16:13 <efried> right
13:16:21 <esberglu> Easy enough
13:16:32 <efried> But come to think of it, I'm sure there's a way to do that docs build locally.  Probably tox -e docs in fact.
13:16:53 <edmondsw> hmm... yeah, probably
13:17:05 <efried> Anyway, moving on.
13:17:33 <efried> Otherwise, I don't have anything specifically PowerVM in tree.  Unless we want to start looking ahead to queens.
13:18:22 <edmondsw> are you going to tackle some of those minor cleanup things before or after starting on queens stuff like network, config_drive, etc.?
13:18:35 <efried> Before would be better.
13:18:45 <edmondsw> that's what I was thinking, since it'll be a while before we can merge for queens
13:19:00 <efried> Yeah, I don't think those have an owner yet.
13:19:07 <edmondsw> they don't
13:19:26 <edmondsw> might be good stuff for esberglu
13:19:35 <edmondsw> get his feet wetter :)
13:20:01 <efried> btw, is mdrabe invited to these?
13:20:08 <esberglu> edmondsw: Sure I can take a stab at it. I'll start looking through those patches again, been a while
13:20:10 <edmondsw> we can all pick those off as able
13:20:15 <esberglu> efried: Yeah
13:20:18 <edmondsw> esberglu +1
13:20:34 <edmondsw> I thought mdrabe was invited, but let me check
13:20:39 <esberglu> edmondsw: Yeah he is
13:20:55 <esberglu> Alright are we ready to move to OOT?
13:21:03 <edmondsw> esberglu hmm... I was going to say I don't see him on the invite I have, but you may have added him later
13:21:09 <edmondsw> and mdrabe is here!
13:21:11 <efried> Oh, esberglu on the devref thing, note that we have some docs OOT.  A lot of it won't be appropriate (like how to install and set up the OOT driver) but could be a good starting point.
13:21:13 <edmondsw> speak of the devil...
13:21:35 <mdrabe> Having some trouble with my client
13:22:02 <edmondsw> no worries
13:22:18 <efried> Yeah, so train of thought there was something like, in order for esberglu to get to those items, he needs some breathing room in the CI - and wasn't mdrabe going to be helping create said space - say, where's mdrabe, was he invited?
13:23:00 <efried> Anyway, I think we're done talking about in-tree now.
13:23:03 <esberglu> efried: Yeah I'm having him do his first deploy today
13:23:19 <efried> Schweet.  We can talk about it in the CI topic.
13:23:20 <edmondsw> :) as I understand it, mdrabe is starting to learn there on CI, and I'm also hoping esberglu will have more space from getting these patches in that allow quicker CI changes without rebuilds
13:23:45 <edmondsw> we're all over the place today... anything for OOT efried?
13:23:54 <esberglu> edmondsw: That's the idea
13:23:57 <edmondsw> since that's the topic we're supposedly in
13:24:01 <edmondsw> ;)
13:24:34 <efried> I think there's going to be a ripple OOT from the power-off changes.
13:24:49 <efried> First we're going to have to cut a new pypowervm release, for that and the iscsi stuff.
13:25:00 <edmondsw> yep
13:25:20 <efried> Then propose it in g-r and let it percolate.  Then chhavi's change will start to verify, and I can do the above.
13:25:28 <edmondsw> I'd like to see 5421 also go into that pypowervm release
13:25:44 <efried> right
13:26:00 <efried> mdrabe Can you please review 5421?
13:26:12 <mdrabe> yep
13:26:16 <edmondsw> ty sir
13:26:20 <efried> And thorst_afk if you're not too busy saving the world.
13:26:34 <edmondsw> I think he's actually afk right now
13:26:48 <edmondsw> he likes to keep us guessing, though ;)
13:26:53 <efried> Yeah, but he'll see this eventually.
13:26:57 <edmondsw> yep
13:28:14 <esberglu> Ready to move to CI?
13:28:28 <edmondsw> backports from IT to OOT on the TODO list is another thing that anyone can dip into as they have time and could be good learning material for some of us newer guys
13:28:33 <efried> So OOT will also be impacted eventually by edmondsw auto-disable ...
13:28:36 <efried> yeah, what he said.
13:28:39 <edmondsw> yep
13:29:07 <edmondsw> efried I'd like to get the common changes in, then do you think the auto-disable for pypowervm should go IT or OOT first?
13:29:18 <edmondsw> I think we said IT first?
13:29:41 <efried> That was the plan, though I'm not sure there's a solid reason why.
13:29:55 <edmondsw> we'll stick with it for now anyway
13:29:59 <edmondsw> done with OOT?
13:30:37 <edmondsw> #topic CI
13:30:52 <esberglu> #link https://etherpad.openstack.org/p/powervm_ci_todos
13:31:04 <esberglu> The external prep_devstack change is in
13:31:26 <esberglu> And the branch specific local.conf files change is ready, just need reviews
13:31:34 <esberglu> 5425
13:31:38 <edmondsw> esberglu you want to give us a quick status update, what's working / failing?
13:32:12 <edmondsw> I pulled up 5425 right before this meeting started, will get back to it when we're done
13:32:12 <esberglu> Ocata was working yesterday with the local.conf file in that latest patch
13:32:33 <esberglu> newton, master (IT and OOT) were both failing
13:32:50 <edmondsw> I thought we were getting master working first
13:33:10 <esberglu> edmondsw: I just had them running in the background to test the branch specific patches on staging
13:33:43 <esberglu> I was working through master devstack issues all day yesterday
13:33:55 <esberglu> It was a real rabbithole
13:34:08 <esberglu> That's what I'm planning on doing today too
13:34:48 <edmondsw> esberglu the idea to go ahead and merge 5425 since it doesn't make anything worse (pending review, of course) and then fix each branch individually?
13:34:57 <esberglu> edmondsw: Yeah
13:35:13 <esberglu> With priority being OOT master, IT master, OOT stable branches
13:35:24 <edmondsw> +1
13:35:35 <efried> Once you get stuff stacking, please send a note to Jay.
13:35:47 <efried> And the rest of us, for that matter, since we'll be impacted next time we try to stack too.
13:35:50 <esberglu> But with how long stacking takes I'm going to try to get them working simultaneously at least somewhat
13:36:01 <esberglu> efried: Yep
13:36:36 <esberglu> Other than that just tackling the list of TODOs as I am able. Hoping to get mdrabe more involved now that some of his other work has settled down (I think?)
13:38:06 <esberglu> #topic Driver Testing
13:38:16 <esberglu> I know jay is being blocked by the stacking issues
13:38:20 <efried> yup
13:38:43 <esberglu> Anything else going on there?
13:38:49 <efried> Actually, he also ran into a nvl dependencies problem.
13:38:58 <edmondsw> oh?
13:39:12 <edmondsw> anything to worry about?
13:39:14 <efried> Hopefully he asked the right people in slack and got that figured out.
13:39:19 <edmondsw> k
13:39:31 <efried> He might've caught the repo in a weird state or something.
13:40:45 <edmondsw> chhavi did you see my comments on https://review.openstack.org/#/c/467599/ ?
13:41:52 <edmondsw> efried that is now backward compatible with existing pypowervm versions, so it doesn't have to wait for the pypowervm release, right?
13:42:12 <efried> oh, right, we were gonna do that.  I'll have to look again.
13:42:24 <edmondsw> yep, Jenkins passed
13:42:37 <efried> But the new code path still won't work without an updated pvm-novalink
13:42:55 <efried> Need to think through that a bit...
13:43:09 <edmondsw> you mean when we enable more tests?
13:43:21 <efried> Well, that's what needs thought.
13:43:40 <edmondsw> that's probably in queens, right?
13:44:00 <edmondsw> s/probably//
13:44:24 <efried> No.  If we're relying on some new REST behavior via this code, we may get 500s when it runs.
13:44:26 <edmondsw> at least for IT
13:44:35 <edmondsw> yeah, OOT is another story
13:44:53 <efried> Yes, for IT, queens or later.  I'm talking about OOT+pypowervm+pvm-nvl interplay.
13:45:03 <edmondsw> +1
13:45:24 <edmondsw> too easy to get confused with this matrix of possibilities :)
13:45:35 <efried> Totally.  Anyway, I'll re-review that change with that in mind.
13:45:40 <edmondsw> tx
13:46:04 <edmondsw> move on to open discussion?
13:46:13 <esberglu> #topic Open Discussion
13:46:37 <efried> The service tokens changes have merged.  I think mriedem is closing the blueprint.
13:46:37 <esberglu> I was thinking I would create another etherpad with an agenda for this meeting
13:46:43 <efried> +1
13:46:58 <esberglu> With the idea being it will go faster and people can bring stuff up even if they can't make it
13:46:58 <edmondsw> efried awesome
13:47:08 <esberglu> efried: Nice
13:47:12 <edmondsw> esberglu +1
13:47:26 <efried> esberglu --strict also merged in ocata, in case you were looking to use that.
13:47:56 <esberglu> efried: We don't need to do the discover_hosts thing in ocata
13:48:08 <esberglu> Well in the undercloud we do
13:48:08 <efried> okay.
13:48:11 <edmondsw> efried I don't think I ever reviewed the final service tokens change... have a link?
13:48:29 <edmondsw> would like to give it a quick glance and see where it ende dup
13:48:41 <efried> https://review.openstack.org/412634 https://review.openstack.org/419662
13:48:51 <efried> a glance indeed.
13:49:00 <edmondsw> :)
13:49:31 <efried> As I mentioned yesterday, I really didn't do anything on these - just kept rebasing until they went green.
13:49:44 <efried> Biggest piece of work was chasing down the glanceclient bugs.
13:50:18 <edmondsw> I'm not actually on either of those reviews
13:50:29 <edmondsw> didn't you have one up at one point? What happened to that?
13:50:51 <efried> You're thinking of the service URLs bp
13:51:07 <efried> That one's still in limbo pending mordred's infra support for it.
13:51:15 <edmondsw> https://review.openstack.org/#/c/458257/
13:51:48 <edmondsw> yeah, I was getting confused
13:52:18 <efried> Yeah, sorry, I didn't think to add you to the service tokens reviews.
13:52:45 <efried> I know we talked about you being more involved, but then I didn't end up actually having to do much.
13:53:07 <edmondsw> s'ok
13:53:22 <edmondsw> very glad to have that in
13:53:44 <edmondsw> anything else guys?
13:53:51 <efried> power-off changes, I should mention.
13:54:07 <edmondsw> yeah, what's your plan for the separate changeset for my #3?
13:54:21 <edmondsw> able to get that up this week?
13:54:29 <efried> Should be able to, yeah.
13:54:40 <edmondsw> would be nice to get that in the new pypowervm release along with the other things
13:54:41 <efried> I'm going to introduce a new method power_off_progressive
13:54:50 <edmondsw> +1
13:55:05 <efried> The only bit I'm still not sure about is how to specify options.
13:55:26 <efried> Should be via a PowerOffOpts
13:55:40 <efried> But I should validate that guy.
13:55:47 <edmondsw> isn't the current progressive logic hardcoded, doesn't really use any options?
13:55:47 <efried> Cause it should NOT have an operation set.
13:56:11 <edmondsw> i.e. do you even need to let the caller specify anything?
13:56:20 <efried> At least restart.  But there's others as well I believe.
13:56:44 <edmondsw> yeah, probably restart... can't think what else
13:57:04 <edmondsw> would have to look again
13:57:06 <efried> Nope, you're right, just restart.
13:57:15 <efried> Soooo....
13:57:26 <efried> Maybe progressive doesn't take a PowerOffOpts at all.
13:57:36 <efried> Just takes a restart kwarg
13:57:55 <efried> Which feels painful because we jumped through hoops to get rid of the restart kwarg in the existing power_off
13:58:28 <efried> And it paints us into that same corner in case we somehow need to add more opts in the future.
13:58:40 <efried> So yeah, still need to think about this some more.
13:59:32 <edmondsw> k
13:59:33 <efried> Then there's the weird translation bug.
13:59:54 <efried> btang and I discussed it some yesterday, and got it figured out, I think.
14:00:25 <edmondsw> awesome
14:00:47 <efried> We believe the problem is that the i18n test case is setting a locale directory globally, which it shouldn't.  Depending what order the tests run, that can cause problems.
14:01:10 <efried> So I'm going to try to nail that down in the next day or two.
14:01:19 <efried> One more minor pypowervm thing I want to mention.
14:01:41 <efried> We have a couple of tests that run over code that sleep()s, but the tests aren't mocking sleep(), so the tests actually take a relatively long time.
14:01:55 <efried> Easy to see which - just look at the longest-running-tests report at the end of a tox run.
14:01:56 <edmondsw> ooo... good catch
14:02:19 <efried> So if someone wants to pick those off and fix 'em...
14:02:26 <efried> Let the team know so we don't overlap.
14:02:44 <edmondsw> sounds like we need another TODO etherpad for pypowervm
14:03:03 <efried> This is an unusual little glut, really.
14:03:19 <efried> We've normally got one thing at a time, directly driven by some other piece of work.
14:03:27 <edmondsw> k
14:03:50 <efried> But yeah, at the moment it wouldn't be bad to keep track somewhere other than my head.
14:03:57 <edmondsw> yeah...
14:04:01 <efried> But an openstack etherpad isn't really appropriate.
14:04:05 <edmondsw> no
14:04:06 <efried> Unless we bury it in the one we've got.
14:04:29 <efried> Can we get an etherpad server set up internally?
14:04:39 <efried> Cause I suuuuure do hate community wiki
14:04:48 <efried> or is it called connections?
14:05:07 <edmondsw> whatever it's called, I agree that etherpad is way better
14:05:15 <edmondsw> I'll see if i can figure out what our options are there
14:08:03 <edmondsw> alright, we're over, and I think done... tx guys
14:08:20 <efried> later y'all.
14:08:41 <edmondsw> #endmeeting