14:00:01 <esberglu> #startmeeting powervm_driver_meeting
14:00:02 <openstack> Meeting started Tue Jan 30 14:00:01 2018 UTC and is due to finish in 60 minutes.  The chair is esberglu. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:06 <openstack> The meeting name has been set to 'powervm_driver_meeting'
14:00:57 <efried> @/
14:01:06 <edmondsw> o/
14:01:31 <esberglu> #topic In-tree Driver
14:02:05 <esberglu> Nothing here from me
14:02:11 <edmondsw> so we got config drive, OVS, and SEA merged for Queens
14:02:11 <esberglu> Unless we want to start looking at plans moving into Rocky
14:02:13 <edmondsw> but not vSCSI
14:02:16 <esberglu> Yep
14:02:30 <efried> We can propose the Rocky bp any time now.
14:02:31 <edmondsw> I think our goals for Rocky will be...
14:02:41 <edmondsw> a) vSCSI
14:02:51 <edmondsw> b) migrate
14:02:57 <edmondsw> c) resize
14:02:59 <edmondsw> d) snapshot
14:03:12 <edmondsw> any comments on that?
14:03:18 <efried> c)? Did we decide we can't already do that?
14:03:33 <efried> oh, sorry, resize meaning migrate-to-same-host :(
14:03:48 <edmondsw> as in, change the flavor of an instance
14:03:59 <edmondsw> yes, would stay on the same host
14:04:34 <esberglu> Sounds good to me
14:04:37 <efried> Do we have localdisk driver in tree?
14:04:42 <efried> And... do we care about that?
14:04:46 <esberglu> efried: We do not
14:04:54 <edmondsw> I'm also looking at volume multiattach... it may be almost a no-op that we could squeeze in with the vSCSI support
14:05:15 <esberglu> There is a WIP patch for localdisk https://review.openstack.org/#/c/409401/
14:05:21 <edmondsw> efried you mean local volumes or local boot disk?
14:05:28 <efried> boot
14:05:34 <edmondsw> I thought that was already inm
14:06:24 <edmondsw> hmm... will need to take a look at that and see how much more there is to do there
14:06:30 <edmondsw> maybe we can squeeze that in as well
14:06:36 <efried> The other thing we ought to have on the slate is porting from get_available_resource to update_provider_tree during the Rocky time frame.  Both in-tree and out.  The framework stuff (that we didn't quite finish in Queens) ought to go in relatively early, and I want us to be right on top of it.
14:07:03 <efried> I would prioritize that over some of those other things.
14:07:49 <edmondsw> efried sure, and I'm hoping you can run with that
14:08:00 <edmondsw> I'm thinking that's a separate effort from the port-in bp
14:08:23 <edmondsw> i.e. if needs a bp, would be a separate bp
14:08:32 <edmondsw> thoughts?
14:09:01 <efried> Agree it's a separate effort, yes.  And yes, probably bp-worthy.
14:09:02 <esberglu> edmondsw: It makes sense to be a separate bp
14:09:13 <edmondsw> so we're all on the same page there
14:10:01 <esberglu> I'll add some notes to the etherpad with our current rocky plans
14:10:11 <efried> cool
14:10:22 <edmondsw> sounds good
14:10:27 <esberglu> Anything else to discuss IT?
14:10:38 <edmondsw> let's hold off on writing the bp for a few days
14:11:03 <edmondsw> that's it
14:11:16 <esberglu> #topic Out-of-tree Driver
14:11:35 <esberglu> The queens-3 tags are now released for all 3 projects so they
14:11:41 <esberglu> are also in feature freeze now
14:12:38 <esberglu> There's a new changeset up for the volume refactor, haven't had a chance to look at that yet
14:13:01 <edmondsw> esberglu yep, I saw that. hoping to look at it today
14:13:08 <edmondsw> but we couldn't merge until Rocky?
14:13:56 <esberglu> edmondsw: Technically we run the project so we have the power to push stuff during ff
14:14:00 <esberglu> But it's frowned upon
14:14:04 <efried> Not "couldn't", but shouldn't, yeah.
14:14:20 <edmondsw> yeah, I feel better holding off on that to Rocky
14:14:27 <edmondsw> not that we can't go ahead and review
14:14:36 <edmondsw> I kinda expect another iteration or two
14:15:49 <esberglu> yep
14:16:33 <esberglu> I think that's about it going on OOT atm
14:16:46 <edmondsw> yep
14:17:20 <esberglu> #topic Device Passthrough
14:18:00 <efried> We're going to have a meeting later.
14:18:20 <efried> We got a substantial amount done in Q, but the bits that tie it all together are still pending.
14:18:57 <edmondsw> I hope to come out of that mtg with a much better understanding of what we need to do for the bp mentioned above
14:18:59 <efried> One such piece is NRP in GET /allocation_candidates.  Which is a big deal for NRP - but not really such a big deal for Power.  I'm not sure we're going to need to model with nested, really.
14:19:27 <efried> Which would be good, because that's the part that's actually the furthest behind.  (Jay owns it)
14:19:52 <efried> The part that I think we can really use is the update_provider_tree business.
14:20:22 <efried> There are six approved patches thrashing in the gate still.  But the cherry on top is the next few patches, which I just finished up yesterday, and should go in early in Q
14:20:23 <efried> but also
14:20:37 <efried> we could easily carry those in pvcos if we wanted to make use of it right away.
14:20:52 <edmondsw> s/Q/R/
14:20:58 <efried> yeah
14:21:25 <edmondsw> glad to hear you say that. I was hoping we could carry things in pvcos without too much trouble if need be
14:21:33 <edmondsw> we'll talk more in a few hours
14:21:46 <efried> It's relatively little code, couple hundred lines.  And it works.
14:22:01 <edmondsw> working is good...
14:22:03 <efried> yeah
14:23:52 <efried> If you want to see it working, check out this functional test suite: https://review.openstack.org/#/c/520246/42/nova/tests/functional/test_servers.py@1594
14:24:45 <edmondsw> next topic?
14:25:02 <esberglu> #topic PowerVM CI
14:25:18 <esberglu> I've been upgrading staging to pike
14:25:34 <esberglu> I have a mostly working stack
14:25:46 <esberglu> But can't get snapshots to work yet
14:26:02 <esberglu> They just get stuck in queued status and never move anywhere
14:27:25 <esberglu> The management services work fine with the new undercloud afaik, but once the image template builds there's no snapshot so it can't move past that
14:27:38 <esberglu> Hoping that once I get snapshot working the rest will be fine
14:27:48 <edmondsw> good luck
14:28:27 <esberglu> Other than that I still haven't figured out what's going on with that update server name non-ascii test
14:28:40 <esberglu> Haven't been able to reproduce it on a test instance
14:30:07 <esberglu> The last item is the multiple unplug calls on instance deletion
14:31:02 <esberglu> I might bug thorst about that, I think he probably knows more about _push_vif_event
14:31:20 <edmondsw> sure. Not a huge priority, but would be nice to clean up a bit
14:32:04 <esberglu> Yep. And there's one other error I've been seeing in a few runs trying to delete vopts
14:32:06 <esberglu> "Unable to perform operation while the virtual DVD is loaded"
14:32:16 <esberglu> Haven't started debugging that one yet
14:32:41 <edmondsw> k
14:32:59 <esberglu> #topic Open Discussion
14:33:17 <esberglu> Nothing else from me
14:33:53 <edmondsw> nor here
14:33:56 <efried> I can't think of anything
14:34:36 <esberglu> #endmeeting