09:31:19 <BobBall> #startmeeting XenAPI
09:31:20 <openstack> Meeting started Wed Mar 16 09:31:19 2016 UTC and is due to finish in 60 minutes.  The chair is BobBall. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:31:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:31:23 <openstack> The meeting name has been set to 'xenapi'
09:31:31 <BobBall> Morning all - morning johnthetubaguy
09:32:00 <huanxie> morning all, morning johnthetubaguy
09:32:25 <jianghuaw> Hello, all.
09:32:29 <BobBall> #topic Blueprints
09:32:34 <johnthetubaguy> BobBall: hey, so I have questions on that test, I think you are missing one of the file changes, but lets get onto that in a sec
09:32:42 <BobBall> OK - I'll go first :)
09:33:09 <BobBall> I've got a Newton spec up for separating compute from XenServer
09:33:27 <BobBall> #link https://review.openstack.org/#/c/277452/
09:33:32 <BobBall> this is the one we talked about johnthetubaguy
09:33:41 <BobBall> when you were in Cambridge
09:33:57 <BobBall> So you've already "reviewed" it once ;)
09:34:25 <BobBall> I've added you to the review list - if you could give it another once over that'd be great.
09:34:25 <johnthetubaguy> heh
09:34:35 <johnthetubaguy> something for post RC1
09:34:48 <BobBall> Yeah - that's in the next few days isn't it?
09:35:11 <johnthetubaguy> should have been yesterday
09:35:26 <johnthetubaguy> still running around trying to make that happen
09:35:37 <BobBall> Fair enough.
09:35:57 <BobBall> We've actually got quite a few blueprints we'd like to talk about - when's the best time to go through them?
09:36:33 <johnthetubaguy> we can run through the highlights now
09:36:53 <johnthetubaguy> good to have the links in the notes from the meeting, so we can run through them later, etc
09:36:53 <BobBall> OK - so that's my main BP for Newton.
09:36:59 <johnthetubaguy> sounds good
09:37:06 <BobBall> jianghuaw: what do you have open?
09:37:15 <jianghuaw> #link: https://review.openstack.org/#/c/280099/
09:37:27 <jianghuaw> BP for " XenAPI: support VGPU via passthrough PCI"
09:37:51 <jianghuaw> #link: https://review.openstack.org/#/c/274045/
09:37:55 <jianghuaw> for " Xenapi: a new VDI store via streaming"
09:38:14 <johnthetubaguy> there is a dependency thier I guess
09:38:22 <johnthetubaguy> for the VDI streaming an moving the compute
09:38:37 <BobBall> So the first one is because in Mitaka we had a BP that was shelved because we were trying to add a new vGPU type to filter on.  As libvirt will likely present any vGPU device (and some devices such as AMD Tonga will be SR-IOV) as a fake PCI device, we've adopted that approach there.
09:38:56 <BobBall> VDI streaming is indeed a pre-requisite for moving the compute yes.  That's listed in the compute spec.
09:39:10 <jianghuaw> yes.
09:39:10 <johnthetubaguy> sounds good, just checking my memory really
09:39:22 <BobBall> huanxie? Anything from you?
09:39:28 <huanxie> yes, Bob
09:39:39 <huanxie> #link https://blueprints.launchpad.net/nova/+spec/xenserver-give-support-on-neutron-security-group
09:39:46 <huanxie> this is for neturon security group
09:40:26 <BobBall> Ah yes.  Hopefully that's a spec-less blueprint - but that's up to you johnthetubaguy.  Effectively we need to add an interim Linux bridge in the same way that OVS does
09:40:31 <BobBall> -OVS+libvirt
09:41:15 <BobBall> We'll also have a Neutron BP I hope to add support for XenAPI using a daemon mode
09:41:43 <BobBall> And anything from huazhihao?
09:42:35 <huazhihao> I have spec-less BP https://blueprints.launchpad.net/nova/+spec/bare-vhd-image
09:43:28 <BobBall> Ah - awesome.  Yes, I hope that one's specless as well.
09:43:34 <BobBall> I think that covers all of the BPs?
09:43:45 <johnthetubaguy> that feels specless
09:43:57 <johnthetubaguy> although I think it needs more details in the blueprint
09:44:11 <johnthetubaguy> what would the glance options be, why is it useful, etc
09:44:23 <huazhihao> Understood, thank you
09:44:34 <BobBall> Yes - agreed.  huazhihao can add some more details there instead of having a spec.
09:44:40 <johnthetubaguy> needs submitted for approval once we have the process sorted for Newton, will work with the next PTL to get that sorted quickly
09:44:41 <BobBall> OK - now the fun one
09:44:45 <huazhihao> Will do
09:44:50 <BobBall> #topic Bugs & QA
09:45:05 <BobBall> So... There is a big bug in XenAPI live migrate that was discovered yesterday
09:45:41 <BobBall> There has been a whole bunch of change in the area and, while some of the patches were tested and shown to work, there were further patches which were assumed not to have an impact but which seem to have broken all live migration.
09:46:03 <BobBall> #link https://bugs.launchpad.net/nova/+bug/1557585
09:46:04 <openstack> Launchpad bug 1557585 in OpenStack Compute (nova) "Xenapi live-migration does not work at all now" [High,In progress] - Assigned to John Garbutt (johngarbutt)
09:46:24 <BobBall> We don't have a fix yet
09:46:30 <BobBall> but hopefully will have one today
09:46:59 <BobBall> johnthetubaguy is mainly working on it and I'm testing patches
09:47:17 <johnthetubaguy> so my patch changed the driver.py
09:47:23 <johnthetubaguy> it feels like you are missing that change
09:47:44 <BobBall> https://github.com/BobBall/nova/commit/7acd7acafbf03103f2db53f50b3f5619af363c58 I think I have it
09:47:54 <BobBall> https://github.com/BobBall/nova/commits/lm-fix is the branch I'm testing
09:48:00 <johnthetubaguy> mostly because I just found a typo in the code that should have been populating sr_uuid_map that isn't running, for some reason, which is odd
09:48:28 <BobBall> Since there are multiple changes needed, I'm putting them in a github branch and just pulling that branch into my pair of devstacks and restarting all of the nova services
09:48:29 <johnthetubaguy> ah, so maybe it just swollowed the exception during pre-live-migration
09:48:46 <BobBall> Could be.
09:48:46 <johnthetubaguy> I need to start getting unit tests on them, its just a broken code sketch at this point, to get my head straight
09:49:06 <johnthetubaguy> so the latest patch should be better, I think...
09:49:22 <BobBall> OK - but the next plan of action is for me to push a gerrit review for the aggregate bug change (https://github.com/BobBall/nova/commit/3589212cef33736962376aa2fb82ad38ea57ac28) and rebase your and Pawel's changes on that
09:49:25 <johnthetubaguy> do you have a full log trace from the tests you ran on there, well just for the one live-migrate?
09:49:51 <johnthetubaguy> can we put that on the top?
09:49:51 <BobBall> Yes; we're logging everything
09:50:10 <johnthetubaguy> I mean, could I give them a quick read please, for the pre-live-migrate stuff?
09:50:24 <BobBall> huh? I'm confused
09:50:42 <johnthetubaguy> so the logs for the run where you hit the sr_uuid_map error
09:50:49 <BobBall> You want the full logs? sure.
09:51:10 <johnthetubaguy> ideally, yeah
09:51:18 <johnthetubaguy> it feels like it was an error caused by an earlier failure
09:51:27 <BobBall> both sides?
09:51:39 <BobBall> and just nova logs?
09:51:41 <johnthetubaguy> its easier than me working out which side I think it happened
09:51:47 <johnthetubaguy> yeah, just nova-compute
09:52:20 * johnthetubaguy note to self, get test cluster on my laptop running again, once no longer PTL
09:52:21 <BobBall> I wonder if it'd be easier for me to just connect two of the devstacks in RAX cloud
09:52:33 <johnthetubaguy> yeah, quite possibly
09:52:40 <BobBall> I'll do that.  Then we can both connect live.
09:53:01 <johnthetubaguy> cool, I keep meaning on setting that up for the team here actually
09:53:09 <BobBall> setting what up?
09:53:16 <johnthetubaguy> creating xenservers in our cloud
09:54:23 <BobBall> Got two nodes reserved; "just" need to run devstack on them with the right magic
09:54:41 <BobBall> I'll work on that in the next 30 minutes and if I can't get it working I'll just dump the logs from the two nodes I have locally.
09:55:02 <BobBall> Anything else needed v.short term?
09:55:46 <BobBall> OK - then we'll finish that topic
09:55:48 <BobBall> #topic AOB
09:55:51 <BobBall> Anything else :)
09:56:30 <johnthetubaguy> nothing from me
09:56:35 <johnthetubaguy> thanks all
09:57:10 <BobBall> huanxie huazhihao jianghuaw?
09:57:20 <jianghuaw> nothing from me. thanks.
09:57:33 <huanxie> #link https://review.openstack.org/#/c/213112/
09:57:33 <huazhihao> That is it.
09:57:47 <huazhihao> Thank you Bob and John
09:57:53 <huanxie> I just noticed I need johnthetubaguy help to review this again
09:58:20 <BobBall> huanxie: Good point.  It's missed Mitaka, so we'll need to focus on Newton.  I guess it's one for review post-RC1 when johnthetubaguy has more time.
09:58:40 <johnthetubaguy> yeah, looks that way, I am afraid
09:58:49 <huanxie> Thanks a lot:)
09:58:52 <johnthetubaguy> digging on this live-migrate issue
09:59:02 <BobBall> Hoepfully it'll be a quick +2 johnthetubaguy since you've already +2'ed it - but let's talk again on that one soon.
09:59:10 <BobBall> We'll close there then.  Good meeting - thanks all.
09:59:13 <BobBall> #endmeeting