14:02:15 <edmondsw> #startmeeting PowerVM Driver Meeting
14:02:16 <openstack> Meeting started Tue Jun 26 14:02:15 2018 UTC and is due to finish in 60 minutes.  The chair is edmondsw. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:20 <openstack> The meeting name has been set to 'powervm_driver_meeting'
14:02:32 <edmondsw> link: https://etherpad.openstack.org/p/powervm_driver_meeting_agenda
14:02:45 <edmondsw> ping gman-tx mdrabe mujahidali chhagarw
14:03:08 <edmondsw> #topic In-Tree Driver
14:03:10 <edmondsw> link: https://etherpad.openstack.org/p/powervm-in-tree-todos
14:04:05 <efried> edmondsw: It would be neat to include TechM in these meetings once they're set up.
14:04:10 <edmondsw> efried agreed
14:04:26 <edmondsw> I don't know that we have a lot of IT work going on right now
14:04:52 <edmondsw> efried anything you want to talk about here?
14:05:04 <efried> Will defer to dev passthru topic.
14:05:10 <edmondsw> k
14:05:21 <edmondsw> #topic Out-of-Tree Driver
14:05:39 <edmondsw> #link https://etherpad.openstack.org/p/powervm-oot-todos
14:06:10 <edmondsw> mdrabe any update on the MSP work?
14:06:19 <mdrabe> Haven't started it yet
14:06:53 <edmondsw> k
14:07:10 <edmondsw> I saw a new patch set from chhagarw but haven't looked yet
14:07:43 <edmondsw> esberglu has proposed some things porting changes back from the IT driver
14:07:52 <edmondsw> I think I still owe him a couple reviews there
14:08:11 <esberglu> Yep. They should be pretty easy reviews
14:08:12 <efried> I think I'm up to date on those.
14:08:21 <edmondsw> yeah, I looked at a couple on Friday I think
14:08:42 <edmondsw> maybe all of them? I don't recall now
14:08:42 <esberglu> Everything we needed to backport is either covered there or by a TODO on the etherpads
14:08:58 <edmondsw> tx
14:09:02 <edmondsw> anything else for OOT?
14:09:20 <edmondsw> #topic Device Passthrough
14:09:24 <edmondsw> efried ^
14:09:53 <efried> Okay, so Jay is back from his PTO (btw he left Verizon and is now working for yahoo) and I'm hoping he'll merge the last nrp-in-alloc-cands patch today.
14:10:00 <efried> or at least this week.
14:10:26 <efried> Which will clear the way for us to implement our update_provider_tree with nested GPUs
14:10:44 <efried> Which means edmondsw and I need to talk about what that's going to look like.
14:10:49 <efried> And probably write it down.
14:11:29 <efried> Though procedurally we want to avoid having to propose a blueprint to nova, because we're past spec freeze.  We maybe should have done that at the beginning of the cycle.
14:12:00 <edmondsw> do it OOT first?
14:12:12 <edmondsw> and port IT for stein?
14:12:19 <efried> Yeah, that's a good point.
14:12:58 <efried> Anyway, the first todo is to have that design discussion.  edmondsw, when are you available to do that?
14:13:17 <edmondsw> good question :)
14:13:31 <edmondsw> I can probably find some time this afternoon if that works
14:13:36 <efried> yuh.
14:13:50 <efried> Does mdrabe need to be involved too?
14:14:27 <edmondsw> doesn't have to be, but he might want to join us
14:15:04 <edmondsw> invite him
14:15:45 <edmondsw> sounds like we're done there for now
14:15:47 <efried> Oh, I figured you would do the calendar thing since you've got all the restrictions.
14:15:54 <edmondsw> efried ok :)
14:15:59 <efried> Anyway, beyond that I've still got some fup to do integrating new placement-isms into the scheduler - the newer generation handling stuff is exposing races (as it should) and I need to fix 'em.
14:16:12 <efried> and I owe cyborg some spec review.
14:16:34 <efried> The more I look at cyborg the more involved I feel we're going to need to be to do our accelerator stuff.
14:16:46 <efried> Like, we're going to need to write os-acc plugins as soon as that's a thing.
14:16:48 <edmondsw> yes, I've wanted us involved from the start
14:17:06 <efried> https://review.openstack.org/#/c/577438/  <== review me
14:17:20 <edmondsw> I tried to push that when I first heard of it but nobody had bandwidth to get really involved there
14:17:45 <edmondsw> I'm glad you're getting more involved there
14:17:46 <efried> Well, we're only now getting to the point where the nova/cyborg interaction is being defined.
14:17:57 <efried> so it's not like we missed any boat or anything.
14:18:02 <edmondsw> right, I think we're fine
14:18:16 <efried> yup, just need to stay on top of it.
14:18:25 <edmondsw> ++
14:18:29 <edmondsw> anything else?
14:18:43 <efried> Not sure if there are others on the pvc team who want/need to become familiar in this space.
14:19:08 <efried> nope, that's it from me.
14:19:19 <edmondsw> at this point I think it would be me and mdrabe, maybe madhavi as well
14:19:44 <edmondsw> #topic PowerVM CI
14:19:57 <edmondsw> #link https://etherpad.openstack.org/p/powervm_ci_todos
14:20:30 <efried> as of yesterday, I think the CI was ill.  Is it back?  (I haven't looked yet today)
14:20:42 <edmondsw> PowerVM CI is failing for everything
14:20:48 <edmondsw> http://ci-watch.tintri.com/project?project=nova
14:20:53 <esberglu> I looked at ci-watch yesterday, looks like it has been failing since it got back online
14:21:05 <esberglu> Failing the test_show_update_rebuild_list_server test
14:21:11 <edmondsw> mujahidali have you looked at this?
14:21:18 <esberglu> Which was updated here https://github.com/openstack/tempest/commit/1fa4464404dd4400e1c0669dda29d696d3e5badb
14:21:29 <esberglu> And here's the gerrit review https://review.openstack.org/#/c/526485/
14:21:54 <esberglu> I haven't looked into why that test is failing, that should be enough to get mujahidali started
14:22:14 <edmondsw> esberglu tx
14:22:30 <esberglu> We can add it to the skip list temporarily if it isn't easy to debug
14:22:49 <edmondsw> I'll try to help take a look at that after the mtg
14:23:54 <edmondsw> esberglu there was an email question to you about moving the zuul merger instances to mujahidali
14:24:21 <edmondsw> please look for that and reply when you can, should be quick
14:24:35 <esberglu> edmondsw: I responded to the thread right before the meeting
14:24:37 <mujahidali> yeah, I looked into the failure, but didn't get much from the logs, thanks esberglu for the hlp
14:24:42 <edmondsw> oh, I see it now, sorry
14:25:24 <edmondsw> mujahidali when the CI is having issues like this, please ping me and/or efried so we know
14:25:42 <mujahidali> sure
14:25:43 <edmondsw> that you've seen it
14:25:51 <edmondsw> and are working. And maybe we can help
14:26:50 <edmondsw> the change that was blocking stable/pike vSCSI CI work merged (in neutron, if I remember correctly), so that should be unblocked when we can get back to it
14:27:49 <esberglu> Ocata?
14:28:02 <mujahidali> https://review.openstack.org/#/c/574726/
14:28:05 <edmondsw> esberglu oh, yeah... ocata
14:28:25 <edmondsw> in devstack, for neutron
14:29:05 <edmondsw> mujahidali in last week's meeting we agreed that you would split out the pike and queens stuff so that we could get that merged
14:29:09 <edmondsw> and we would do ocata later
14:29:22 <edmondsw> but if that has merged now, it may be easier to just do ocata here as well
14:29:27 <edmondsw> unless you hit more issues
14:29:34 <esberglu> +1
14:29:46 <edmondsw> doesn't look like anything has happened with 6596 since last week
14:30:19 <edmondsw> mujahidali please take a look at that once we get the CI working again
14:30:42 <mujahidali> sure
14:30:55 <mujahidali> will try to close it asap
14:31:37 <edmondsw> esberglu do we have to use x86 for the zuul merger nodes, or could they be ppc64le?
14:31:50 <mujahidali> I have started looking into the multinode-setup as well , currently, trying out the steps mentioned by :esberglu on Staging environment.
14:32:58 <edmondsw> ++
14:33:03 <esberglu> edmondsw: They could be ppc64le
14:33:40 <esberglu> I don't know of any reason they couldn't be I should say
14:33:55 <esberglu> Should be fine as long as its still Ubuntu 16.04 I would think
14:34:03 <edmondsw> so mujahidali you might want to try setting those up in os4pcloud, where I think we can keep them from expiring
14:35:00 <edmondsw> not sure it's worth the risk... but something to consider
14:35:22 <edmondsw> esberglu any other pros to having in jupiter?
14:35:37 <mujahidali> I will first deploy the ppc64 instances and will add them in host list and see if it's working then will revert to jupiter
14:35:53 <edmondsw> e.g. other things run there, so we're not susceptible to networking issues with os4pcloud today
14:35:59 <mujahidali> *if ppc not working
14:36:44 <esberglu> mujahidali: Test it out on staging first
14:36:52 <edmondsw> ++
14:36:57 <mujahidali> ++
14:37:10 <edmondsw> anything else for CI?
14:37:16 <chhagarw> are there any open issues with nova-powervm CI
14:37:31 <edmondsw> chhagarw yes, the CI is totally busted right now
14:37:38 <edmondsw> we're working on it
14:38:56 <edmondsw> #topic Open Discussion
14:39:14 <edmondsw> we got back the draft image of our PowerVMStackers mascot
14:39:20 <edmondsw> I emailed it around
14:39:41 <edmondsw> everyone ok with it? The feedback I've heard so far has been positive
14:39:44 <esberglu> edmondsw: I like it
14:39:55 <mujahidali> +1
14:40:35 <edmondsw> efried and gman-tx also gave thumbs up
14:41:00 <edmondsw> and svenkat
14:41:27 <edmondsw> unless I hear something negative shortly, I'll reply to Kendall and tell her it's good.
14:41:34 <edmondsw> that's all I had... anything else?
14:42:40 <edmondsw> chhavi__ did you want to give an update on iSCSI work?
14:43:08 <edmondsw> you missed the OOT discussion where that would normally go
14:44:06 <edmondsw> guess not
14:44:11 <edmondsw> alright, thanks folks
14:44:13 <edmondsw> #endmeeting