09:32:12 <BobBall> #startmeeting XenAPI
09:32:13 <openstack> Meeting started Wed Nov  2 09:32:12 2016 UTC and is due to finish in 60 minutes.  The chair is BobBall. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:32:14 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:32:16 <openstack> The meeting name has been set to 'xenapi'
09:32:29 <BobBall> Good morning/afternoon everyone
09:32:45 <BobBall> johnthetubaguy pingity ping?
09:32:56 <jianghuaw> Good morning BobBall.
09:33:28 <BobBall> #topic Blueprints / specs
09:33:38 <huanxie> Hello all
09:33:46 <BobBall> So - jianghuaw - can we have an update for your two specs from the discussions at the summit?
09:34:37 <jianghuaw> yes. For "generalizing PCI devices", the cores agreed that's a long term direction but don't think it should be in Ocata.
09:35:47 <jianghuaw> as the PCI refactoring as resource object is still in progress.
09:35:54 <BobBall> Is there any pre-work needed that's happening in ocata?
09:36:03 <BobBall> IS that expected to finish in Ocata or not?
09:36:35 <jianghuaw> Per Jay's input, PCI refactorying should be done in Ocata.
09:37:01 <jianghuaw> after that, it should be suitable for "generalizing PCI" going to P.
09:37:11 <BobBall> I see.
09:37:34 <BobBall> And you mentioned yet another different proposal for vgpu?
09:38:03 <jianghuaw> For seeking support VGPU, Jay suggested to use the custom resource class for VGPU.
09:38:50 <BobBall> That sounds full circle back to what we proposed in Liberty which is what Dan was not in favour of?
09:38:58 <jianghuaw> define single VGPU types in the section of [placement] in nova's configure file.
09:40:51 <jianghuaw> in liberty, we proposed to create a top resource for VGPU. and also rejected by Jay IIRC.
09:42:14 <BobBall> ok perhaps I don't understand the difference between those two?
09:42:19 <jianghuaw> But it seems it's determined to support custom resource classes.
09:42:44 <jianghuaw> it seems there is plan to support FPGA with custom resource type also.
09:42:58 <BobBall> I see
09:43:55 <jianghuaw> In liberty, the refactoring for resource pool is at the early phase.
09:44:33 <jianghuaw> and there is no concept of the custom resource class. All resources are defined as top resources.
09:45:03 <BobBall> Understood
09:45:14 <BobBall> so for vgpu we need to rewrite the spec to work on custom resource types
09:45:23 <BobBall> vGPU attempt #4
09:45:49 <jianghuaw> Unfortunately it seems yes.
09:46:25 <BobBall> OK - any other updates from the summit?
09:47:15 <jianghuaw> the nova-network will be removed from ocata, so we should think to remove our nova network CI accordinly?
09:47:59 <BobBall> Deprecated or removed?
09:48:21 <huanxie> deprecated iirc
09:48:24 <BobBall> Typically we remove after a cycle of deprecation
09:48:32 <BobBall> so nova network may continue to work for Ocata I think?
09:48:40 <jianghuaw> or yes.
09:48:50 <BobBall> We should probably continue testing until Infra remove their nova network CI
09:48:58 <BobBall> Once that's gone, then yes, let's disable ours.
09:49:05 <jianghuaw> fair enough.
09:49:33 <BobBall> OK - any more for any more?
09:50:25 <huanxie> https://etherpad.openstack.org/p/ocata-nova-neutron-session
09:50:35 <BobBall> For some reason my compute keeps locking up - I may disappear unexpected.
09:50:36 <huanxie> this link gaves plan on nova-network
09:51:15 * johnthetubaguy wonders in very late and hides at the back
09:51:19 <huanxie> "we can't drop nova-network until cellsv2 is available for upgrading from cellsv1"
09:51:45 <BobBall> Ah - ok
09:51:52 <johnthetubaguy> yeah, we have just put a patch to stop nova-network starting, unless you are using cells v1
09:51:54 <BobBall> so we should keep our nova-network CI for the forseable future :D
09:52:06 <BobBall> oh right maybe not
09:52:22 <BobBall> because our CI isn't using cells (unless it's devstack default?)
09:52:40 <BobBall> Got a link to the patch johnthetubaguy ?  Just worth us looking to see if it does stop our CI working
09:55:14 <johnthetubaguy> just digging it up
09:55:32 <johnthetubaguy> https://review.openstack.org/#/c/392236/
09:55:53 <jianghuaw> If it stops nova-network starting, it surely will break our CI once that patch got merged.
09:56:02 <BobBall> awesome - yes - it does break our nova-network CI
09:56:06 <johnthetubaguy> yeah, we have a -ve vote from lots of CI on there
09:56:13 <johnthetubaguy> which is expected
09:56:34 <BobBall> So we should just double check that the failure is due to nova network being disabled
09:56:46 <BobBall> and if it is, let's stop nova-network from voting
09:56:52 <BobBall> then we can remove it entirely once this patch has merged.
09:57:05 <johnthetubaguy> http://dd6b71949550285df7dc-dda4e480e005aaa13ec303551d2d8155.r49.cf1.rackcdn.com/36/392236/2/check/dsvm-tempest-nova-network/11cfd32/logs/screen-n-net.txt.gz
09:57:08 <johnthetubaguy> awesome
09:57:38 <BobBall> That'll free up some CI resources for the fun new test that huazhihao is working on :)
09:57:51 <huazhihao> Indeed
09:58:15 <jianghuaw> awesome.
09:58:23 <BobBall> OK
09:58:26 <BobBall> Rapidly running out of time
09:58:29 <BobBall> #topic AOB
09:58:32 <BobBall> Let's skip to AOB
09:58:37 <BobBall> Anything else we should cover today?
09:58:39 <jianghuaw> For this spec, generally I think we've got agreement. But we need some cores help to review and approve it before the spec freezing date.
09:58:40 <jianghuaw> Link: https://review.openstack.org/#/c/379279/
09:59:09 <BobBall> Ah yes - your review on the os-xenapi spec would be appreciated johnthetubaguy :)
09:59:24 <johnthetubaguy> yeah, there was that wording to iron out, looks like thats done now :)
09:59:36 <huanxie> Hope you can help review my vif hotplug patch, maybe we can discuss it next time https://review.openstack.org/#/c/365505/, thanks:)
09:59:40 <BobBall> We have the library created and are committing code to it
09:59:56 <johnthetubaguy> oh, whats the formatting on line 19 about?
09:59:58 <BobBall> so first release of the library shouldn't be too far away at which point we can do the Nova changes.
10:00:08 <johnthetubaguy> https://review.openstack.org/#/c/379279/3/specs/ocata/approved/add-os-xenapi-library.rst@19
10:00:28 <johnthetubaguy> looks like a merge or cut and paste error I guess?
10:00:34 <BobBall> Looks like a newline was added in the wrong place
10:00:37 <BobBall> look at the changes from 2-3
10:00:53 <BobBall> added a newline rather than deleting part of the formatting
10:00:54 <jianghuaw> indeed, will revise it. thanks.
10:01:14 <johnthetubaguy> cool
10:01:28 <johnthetubaguy> ping me when thats ready in the main channel
10:01:46 <BobBall> +1
10:01:48 <BobBall> Awesome
10:02:06 <BobBall> So johnthetubaguy - did you spot the query about reviewing the vif hotplug patch?
10:02:30 <johnthetubaguy> not yet
10:02:36 <BobBall> 09:59 < huanxie> Hope you can help review my vif hotplug patch, maybe we can discuss it next time https://review.openstack.org/#/c/365505/, thanks:)
10:02:41 <BobBall> I didn't mean did you review it :)
10:04:12 <johnthetubaguy> huanxie: I made you the assignee on the blueprint, so that should help you edit it, if needed.
10:04:34 <huanxie> johnthetubaguy, Thanks
10:04:43 <johnthetubaguy> will try take a peak
10:04:48 <BobBall> Perfect.
10:04:49 <johnthetubaguy> would be good to get that added
10:04:51 <BobBall> Any more for any more?
10:04:59 <johnthetubaguy> especially now neutron is... more important
10:05:02 <BobBall> :D
10:05:11 <jianghuaw> new patch upload
10:05:12 <BobBall> Let's close there then.
10:05:14 <jianghuaw> uploaded.
10:05:14 <BobBall> Thanks everyone.
10:05:18 <BobBall> #closemeeting
10:05:18 <jianghuaw> thanks.
10:05:20 <huanxie> Thanks all:)
10:05:21 <BobBall> #endmeeting