14:03:21 <edmondsw> #startmeeting PowerVM Driver Meeting
14:03:22 <openstack> Meeting started Tue Mar 27 14:03:21 2018 UTC and is due to finish in 60 minutes.  The chair is edmondsw. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:03:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:03:26 <openstack> The meeting name has been set to 'powervm_driver_meeting'
14:03:30 <edmondsw> https://etherpad.openstack.org/p/powervm_driver_meeting_agenda
14:03:44 <edmondsw> #topic In-Tree Driver
14:03:53 <edmondsw> https://etherpad.openstack.org/p/powervm-in-tree-todos
14:04:29 <edmondsw> esberglu you want to give us a quick update?
14:04:38 <esberglu> Yep
14:04:51 <efried> รถ/  (new haircut)
14:05:11 <edmondsw> :)
14:05:23 <esberglu> Need to respond to a few things on network hotplug for mriedem
14:05:34 <esberglu> But looks like he's ready to +2 after that
14:05:42 <edmondsw> well, after we get the CI testing it
14:05:48 <edmondsw> how's that going?
14:06:15 <esberglu> edmondsw: Sounded like he was okay with putting it through anyways, he asked about that in -nova yesterday
14:06:32 <esberglu> edmondsw: Still haven't gotten active RMC was gonna talk to mdrabe today
14:06:35 <edmondsw> yeah, I saw, but the way I read it he was only ok with +2 once the CI is testing it
14:07:06 <esberglu> Could be, either way we need to get that working
14:07:09 <edmondsw> yep
14:07:38 <esberglu> vSCSI, Snapshot, DiskAdapter and Localdisk are all ready for review
14:07:42 <edmondsw> let's focus on getting that working in CI
14:07:54 <esberglu> Yep
14:08:20 <esberglu> Next up is resize but I'm gonna sit on that until we start moving on the rest
14:08:55 <edmondsw> +1 on vscsi
14:09:25 <esberglu> edmondsw: Do the pypowervm 1.1.12 in a follow on?
14:09:48 <edmondsw> I guess. that's what efried suggested, and I'll yield to him on that
14:10:18 <esberglu> Adding a todo to the etherpad
14:10:41 <efried> did I say that?
14:10:50 <edmondsw> yep, in the review
14:11:00 <efried> Oh, because we were going to wait until 1.1.13 to bump the minimum.
14:11:16 <edmondsw> 1.1.13 should be cut today, and as soon as that's in u-c we can bump the minimum
14:11:37 <edmondsw> if that happens before we see anyone looking at vscsi, we can make the change there, else we can try to do a followon
14:11:43 <edmondsw> sound like a plan?
14:11:59 <esberglu> Sounds good
14:12:15 <esberglu> That's all I had for IT
14:12:42 <edmondsw> I'm still trying to get to reviewing snapshot :(
14:12:43 <efried> runways start tomorrow.
14:12:48 <edmondsw> cool
14:13:01 <efried> Have we queued up the ready reviews in there yet?
14:13:22 <edmondsw> efried I think vscsi is the only one really ready
14:13:26 <edmondsw> I will add it
14:13:36 <esberglu> Nope was waiting until I had +1s from both of you on vSCSI
14:13:47 <esberglu> And CI for network hotplug
14:13:50 <efried> am I not +1 on that yet?
14:13:57 <edmondsw> you are... I wasn't until just now
14:14:01 <efried> okay.
14:14:15 <efried> Wanna wait a week, I can be +2 :)
14:14:24 <edmondsw> esberglu actually you go ahead and add vscsi to the runway
14:14:30 <esberglu> Oh yeah congrats efried!
14:14:31 <edmondsw> efried oh? :)
14:14:36 <edmondsw> congrats then!
14:14:42 <efried> thanks, provisionally.  The nom is out on the ML.
14:14:52 <edmondsw> hadn't noticed yet... that's awesome
14:15:13 <efried> I'm right chuffed, as they used to say on the east side (of London)
14:15:21 <edmondsw> as you should be
14:16:33 <edmondsw> one thing about localdisk
14:17:02 <edmondsw> did we work out what we're doing so that conf settings don't conflict between IT and OOT?
14:17:32 <esberglu> Oh right. No we didn't
14:18:08 <edmondsw> are you going to try my suggestion of an OOT change to check and see if the setting is already defined and if so not redefine it?
14:18:36 <esberglu> Sure
14:18:45 <edmondsw> ok, let us know how that goes
14:18:51 <edmondsw> anything else for IT?
14:18:52 <efried> That doesn't sound like it'll work.  I don't think you can predict the load order.
14:19:06 <efried> Perhaps ask around in -oslo
14:19:08 <edmondsw> that was my fear
14:19:17 <efried> Or on the dev ML.
14:19:23 <efried> I bet dhellmann has this answer off the top.
14:19:26 <edmondsw> I thought maybe we could assume nova was loaded before a nova virt driver, though
14:19:38 <esberglu> We could name the IT group something slightly different
14:19:43 <edmondsw> yeah, he'd be the person I'd go to
14:19:57 <edmondsw> esberglu I really don't want to name things differently if we can avoid it
14:20:05 <efried> Namespacing would be okay, except we really want the in
14:20:12 <efried> in-tree one to be called [powervm]
14:20:15 <edmondsw> a) I want consistency
14:20:25 <efried> oh, now I remember.
14:20:27 <edmondsw> b) I like the existing names better than anything else I've thought of
14:20:47 <efried> (edmondsw We wouldn't have to change the conf option names, just the group name.  But even so)
14:21:04 <efried> I remember how we need to tackle this.
14:21:05 <edmondsw> efried right... but like you said, we want the group name to stay the same
14:21:14 <efried> Because we already lockstep nova and nova-powervm.
14:21:28 <efried> We introduce the conf option into nova and then *remove* same from nova-powervm.
14:21:31 <esberglu> Can we just add it to nova and remove from nova-powervm
14:21:33 <esberglu> Yeah
14:21:52 <efried> Those patches land in quick succession and we're good.
14:22:11 <edmondsw> except for the time between the first merge and the second
14:22:32 <esberglu> efried: There might still be a problem with that, the group would get registered twice I think
14:22:33 <efried> which should be negligible.  Put Depends-On from the nova-powervm patch to the nova patch.
14:22:43 <edmondsw> and CI will be failing OOT for the IT change
14:23:06 <edmondsw> which I guess we can ignore
14:23:14 <efried> yeah
14:23:34 <efried> esberglu: Not sure about the group redef thing.  Another one for dhellmann.
14:23:36 <edmondsw> do we have Depends-On working?
14:23:53 <esberglu> no
14:23:58 <efried> oh, right.  No.  But that's okay.  We'll just land 'em quickly.
14:24:00 <edmondsw> yeah, I didn't think so
14:24:07 <edmondsw> k
14:24:35 <edmondsw> so esberglu will check with dhellmann on the group redef thing
14:24:40 <edmondsw> and we'll take it from there
14:25:09 <edmondsw> #topic Out-of-tree Driver
14:25:21 <edmondsw> https://etherpad.openstack.org/p/powervm-oot-todos
14:25:48 <efried> or just try it
14:26:22 <edmondsw> we've got a couple things waiting for the pypowervm bump
14:26:46 <edmondsw> already said that should release today, so the bump should happen later this week
14:27:34 <edmondsw> I think a couple of us have dropped more comments on the volume refactor, and it will need to be rebased again
14:27:54 <efried> Are we really going forward with that?
14:28:09 <efried> I mean, I'm all for the idea, but does tjakobs really have the bandwidth to go test all the drivers live?
14:28:21 <edmondsw> he hasn't said otherwise
14:28:29 <edmondsw> he was willing
14:28:40 <esberglu> I thought there was something in slack the other day suggesting otherwise
14:28:42 <efried> okay, wfm.  *I* ain't gonna do it.
14:28:54 <edmondsw> we'll see how that goes, but I would really like to see that merge
14:29:04 <edmondsw> (once thoroughly tested)
14:29:25 <efried> yeah, cause I'm going to be pretty hardline about not merging it unless it's tested.
14:29:31 <edmondsw> same
14:29:42 <edmondsw> I think everyone's in agreement there
14:29:52 <edmondsw> besides tjakobs, I want PowerVC to test it
14:30:03 <edmondsw> so we know the things they use are still working
14:30:14 <edmondsw> that should give us another layer of confidence
14:30:32 <edmondsw> anything else OOT?
14:31:01 <edmondsw> oh chhagarw is working on some stuff for iscsi that I should bring up
14:31:54 <edmondsw> we've got an issue there where I don't think it will be possible to setup the same volume through both NovaLink as a VIOS and through traditional VIOS
14:32:01 <edmondsw> so we're trying to work out how to handle that
14:32:08 <efried> at the same time?
14:32:29 <efried> uh, sorry, maybe that didn't make any sense at all.
14:32:30 <efried> Never mind.
14:32:55 <edmondsw> k
14:33:12 <edmondsw> at any rate, there will be nova-powervm changes coming for iscsi in one form or another
14:33:30 <edmondsw> anything else OOT?
14:33:35 <esberglu> nope
14:33:51 <edmondsw> #topic Device Passthrough
14:34:06 <edmondsw> I haven't been able to do much here the last week
14:34:09 <edmondsw> efried you have anything?
14:34:43 <efried> Today is spec review day.  Tomorrow runways start.  Other than continuing to shepherd outstanding patches and review related stuff, no update from last time.
14:35:03 <efried> update_provider_tree is queued up for the first runway, I think.
14:35:09 <edmondsw> yep, saw that
14:35:15 <edmondsw> very good
14:35:17 <efried> So I expect that to be merged within the next week or two.
14:35:27 <efried> But progress on NRP-in-a_c is still very slow.
14:35:38 <efried> And that's going to be a prereq for... everything else.
14:35:44 <edmondsw> a_c ?
14:35:44 <efried> Including granular
14:35:50 <efried> GET /allocation_candidates
14:35:54 <edmondsw> ah
14:36:01 <efried> The placement API that does scheduling
14:36:04 <efried> basically
14:36:49 <efried> meanwhile pretty much every spec on the slate talks about using granular as if it's already a thing
14:36:51 <efried> So no pressure.
14:37:23 <edmondsw> :)
14:37:33 <edmondsw> alright, moving on to CI
14:37:38 <edmondsw> #topic PowerVM CI
14:37:39 <edmondsw> https://etherpad.openstack.org/p/powervm_ci_todos
14:37:41 <edmondsw> esberglu?
14:37:55 <esberglu> Top priority is the network hotplug CI
14:37:59 <edmondsw> +1
14:38:04 <esberglu> Then I can hop back into the nodepool update
14:38:47 <esberglu> Everything is running smoothly again for the most part
14:39:05 <esberglu> IT and OOT are actually seeing the vOPT naming conflict thing at about the same rate
14:39:11 <esberglu> Meaning https://review.openstack.org/#/c/540453/ isn't really doing anything
14:39:22 <esberglu> And that it might not be the naming conflict we thought it was
14:39:52 <edmondsw> hmm
14:40:57 <esberglu> Need to hop back into that and debug
14:41:39 <edmondsw> should we prioritize multinode testing over the nodepool update?
14:42:24 <esberglu> edmondsw: I really wanted to finish the mgmt service upgrade and queens upgrade first
14:42:31 <esberglu> But keep getting pulled off elsewhere
14:42:56 <edmondsw> think we're getting close there?
14:42:58 <esberglu> Probably time to start weighing that choice though
14:43:27 <esberglu> Idk it depends on if the management service update fixes the issues that the queens undercloud is seeing
14:43:40 <edmondsw> you've already got a commit up for resize, and it will need multinode CI before it has a chance of merging
14:44:41 <esberglu> edmondsw: Yeah I will start looking at it more in depth, I'll try to have a plan moving forward for next weeks meeting
14:44:47 <edmondsw> sounds good
14:45:04 <esberglu> Nothing else from me
14:46:05 <edmondsw> #topic Open Discussion
14:46:34 <edmondsw> efried and I just got approval to book travel for the summit
14:46:44 <efried> Oh, cool, hadn't seen that.
14:46:54 <edmondsw> should be in your inbox
14:47:02 <efried> 222 unread as of right now.
14:47:23 <edmondsw> I plan to go ahead and do that today, see how booked the hotels are (probably full)
14:47:37 <edmondsw> will let you know what I find
14:47:56 <edmondsw> anything else to discuss?
14:47:57 <efried> I may ping you later for info about which hotels etc.  I don't remember seeing anything on the website last time I looked.
14:48:17 <edmondsw> hmm... I registered, but haven't looked for flights/hotels yet
14:48:26 * efried headless chicken / one-armed wallpaper hanger / one-legged man in ass-kicking contest today
14:48:41 <edmondsw> yeah, today is gonna be rough for me too
14:49:13 <edmondsw> have a dentist appointment, Ben is visiting so we have extra meetings with that, etc.
14:49:25 <efried> so, let's close this down if nothing else to discuss (that can't wait)
14:49:33 <edmondsw> yep
14:49:36 <edmondsw> #endmeeting