15:03:56 <johnthetubaguy> #startmeeting XenAPI
15:03:57 <openstack> Meeting started Wed Nov 19 15:03:56 2014 UTC and is due to finish in 60 minutes.  The chair is johnthetubaguy. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:03:58 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:04:00 <openstack> The meeting name has been set to 'xenapi'
15:04:09 <johnthetubaguy> BobBall: hows things going?
15:04:26 <BobBall> Not too bad
15:04:28 <BobBall> Not too bad at all
15:04:39 <johnthetubaguy> #topic XenServer CI
15:04:47 <BobBall> Oh yes - first subject of the day :)
15:04:49 <BobBall> CI is going well
15:04:55 <johnthetubaguy> cool
15:04:55 <BobBall> Good passes recently
15:05:00 <BobBall> https://review.openstack.org/#/c/133975/
15:05:06 <johnthetubaguy> I know there were some snapshot issues matel was digging on
15:05:15 <BobBall> that's the test change I've been rechecking
15:05:35 <BobBall> no snapshot issues in the CI
15:05:40 <BobBall> is that a seperate subject?
15:05:51 <BobBall> or is that Matel's neutron CI you're talking about?
15:06:27 <johnthetubaguy> ah, it could well be
15:06:55 <johnthetubaguy> I don't remember the details, just some 'bad' snapshots were created
15:07:20 <BobBall> Speaking of matel - here he is
15:07:31 <johnthetubaguy> BobBall: thats quite a lot of passes on that patch
15:07:36 <BobBall> What are the CI snapshot issues you've seen? :)
15:07:38 <johnthetubaguy> maybe its worth a whirl?
15:07:41 <BobBall> johnthetubaguy: not enough for my liking :)
15:07:46 <johnthetubaguy> lol, OK
15:08:13 <BobBall> So far only 8 passes.  If the fail rate was 10% then we still probably wouldn't have hit it
15:08:19 <BobBall> yet a 10% fail rate is huge
15:08:46 <matel> sorry guys, what's the topic?
15:08:59 <johnthetubaguy> topic is CI
15:09:07 <BobBall> 15:05 < johnthetubaguy> I know there were some snapshot issues matel was digging on
15:09:11 <johnthetubaguy> just curious about the snapshot issues you saw the other day?
15:09:14 <BobBall> I was trying to understand what they were
15:10:06 <matel> Ah, I re-generated the image later, and haven't looked at it since then.
15:10:16 <matel> Gimme a sec, let me see how bad is the situation
15:10:53 <BobBall> oh I see
15:11:02 <BobBall> it was snapshot issues with RAX in the image generatio
15:11:13 <BobBall> not snapshot issues seen in tempest failures?
15:11:15 <johnthetubaguy> yeah, upstream code bug
15:11:29 <johnthetubaguy> need to put the fix up, but it should be deployed everywhere now
15:11:37 <johnthetubaguy> I suspect it confused nodepool a bit
15:11:43 <BobBall> yeah
15:11:55 <johnthetubaguy> it will have tried to use a snapshot that wasn't actually a success
15:11:58 <BobBall> matel was working on us using the latest nodepool code
15:12:09 <BobBall> because there are a bunch of error handling improvements
15:12:15 <johnthetubaguy> nova would delete it, but not if you delete the instance first, :(
15:12:32 <matel> THe problem happened when I used nodepool to bake an image and launch instances of it.
15:13:05 <BobBall> I see
15:13:06 <matel> No stuck/building instances now.
15:13:06 <BobBall> fair enough
15:13:14 <BobBall> Joy.
15:13:21 <BobBall> So things seem to be OK ATM
15:13:26 <matel> yes
15:13:31 <BobBall> Even better.
15:13:33 <johnthetubaguy> cool
15:13:37 <johnthetubaguy> sounds like we fixed it
15:13:40 <johnthetubaguy> awesomeness
15:13:45 <johnthetubaguy> any more on the CI?
15:14:02 <BobBall> matel has been working on something
15:14:14 <matel> I am looking at using neutron
15:14:55 <matel> Using the next XS candidate seems to work - as it has a newer Openvswitch
15:15:09 <matel> Altough I still haven't passed all the smoke tests
15:15:20 <matel> Again, they are failing around volumes.
15:15:30 <matel> ..And securoty groups
15:15:46 <matel> So that's what I'm working on.
15:16:00 <matel> ... And I documented the osci deployment
15:16:02 <BobBall> Some changes were needed to devstack etc right?
15:16:07 <BobBall> all upstream now?
15:16:29 <johnthetubaguy> all good stuff
15:16:41 <matel> With the actual tips, it's not working.
15:17:08 <matel> BobBall: I don't quite get what you mean.
15:17:27 <BobBall> the changes needed to get neutron working
15:17:30 <BobBall> I thought there were some?
15:17:42 <matel> There were, they are merged AFAIK
15:17:50 <matel> But it's still not passing.
15:18:07 <matel> Let me look at the reviewboard
15:18:26 <matel> Yes, my devstack changes have been approves
15:18:37 <BobBall> Ah ok good
15:18:39 <BobBall> so nothing outstanding
15:18:46 <BobBall> just us needing to continue to investigate
15:18:49 <matel> SO it means, that if you deploy a neutron from upstream, you should be able to contribute to the investigation
15:18:50 <johnthetubaguy> cool, its a good problem to look into
15:18:54 <matel> Yes, exactly
15:19:06 <johnthetubaguy> looking like we might kill off nova-network in lima
15:19:15 <johnthetubaguy> #topic Open Discussion
15:19:25 <johnthetubaguy> so, any more things to talk about?
15:19:43 <matel> Not from my side
15:19:51 <BobBall> kill off?
15:19:59 <BobBall> surely the plan was to deprecate for at least 1 release
15:20:10 <BobBall> is the plan to formally deprecate for Kilo then and remvoe in Lima?
15:20:34 <johnthetubaguy> if all goes to plan, possibly, although it might be a release later than all that
15:21:16 <BobBall> yes
15:21:19 <BobBall> that might make more sense
15:21:27 <BobBall> actually getting rid of nova-network is going to be hard work :)
15:21:33 <BobBall> But fair enough
15:21:37 <BobBall> we'll just see where it goes
15:21:56 <BobBall> I'm aiming to have our CI running on Neutron before the end of Kilo
15:22:05 <johnthetubaguy> yeah, sounds like a good plan
15:22:10 <BobBall> so doesn't matter much
15:22:10 <BobBall> oh
15:22:24 <BobBall> matel - do we want to ask about quark?
15:22:25 <johnthetubaguy> anyways, any more for any more?
15:22:39 <matel> quark - not yet.
15:22:49 <matel> Let's try to give a go to bare neutron
15:22:56 <matel> ..or
15:22:59 <BobBall> Okies
15:23:08 <matel> johnthetubaguy: does quark support security groups?
15:23:21 <matel> .. maybe that's a question
15:23:29 <BobBall> Indeed.
15:23:43 <BobBall> Does RAX offer security groups as a product offering?
15:24:21 <johnthetubaguy> not right now
15:24:39 <johnthetubaguy> it kinda upsets the scalability of OVS before you start worrying about other things
15:24:49 <BobBall> heh.  Fun.
15:24:50 <johnthetubaguy> BobBall: but what problems are you seeing?
15:25:04 <matel> johnthetubaguy: all tests using secgroups are failing
15:25:14 <BobBall> it's more matel but I think it's that tempest tests security groups and they don't work with OVS plain
15:25:17 <matel> johnthetubaguy: I'll look at where the issue lies
15:25:51 <johnthetubaguy> hmm, OK
15:26:04 <johnthetubaguy> we don't use upstream ML2 OVS driver though
15:26:11 <johnthetubaguy> we use quark
15:26:19 <johnthetubaguy> which is quite a different beast
15:26:29 <BobBall> Indeed.  Sorry, brb.
15:26:52 <matel> johnthetubaguy: I would like to use upsteram components
15:27:10 <johnthetubaguy> matel: agreed
15:27:13 <matel> johnthetubaguy: If something is not possible with them, maybe we should use quark as well
15:27:20 <johnthetubaguy> quark is open source, but its sure not upstream right now
15:27:24 <matel> But give it a first go.
15:27:35 <matel> Is it owned by OS?
15:27:42 <matel> or is it a rackerlabs projects?
15:27:43 <johnthetubaguy> yeah, quark doesn't support security groups yet, so its not really the right answer for this either
15:27:57 <johnthetubaguy> I don't know for sure
15:28:16 <matel> tbh, I'm afraid that I'm gonna run into some serious issues there, but that's just my feeling.
15:28:33 <matel> I hope I'll be more clever next week.
15:28:57 <matel> Ah, johnthetubaguy.
15:29:15 <matel> Could you take a look at: https://github.com/citrix-openstack/install-nodepool/blob/2014-11/README.md
15:29:16 <BobBall> back
15:29:25 <johnthetubaguy> well, that stuff is confusing, getting something working without security groups will be a cool think anyways
15:29:36 <matel> Just a quick read, to make sure it makes sense.
15:29:57 <matel> OK, I'm on it.
15:30:03 <johnthetubaguy> matel: it looks very close to what I did when I tried to set the thing up
15:30:11 <BobBall> Yeah - we can have an exclusion list for now until we figure out how to do security groups with OVS
15:30:18 <johnthetubaguy> matel: I will try get someone to help out with that
15:30:43 <matel> johnthetubaguy: do you need help to read?
15:30:56 <johnthetubaguy> yeah
15:31:07 <matel> Oh, I didn't know that.
15:31:17 <johnthetubaguy> well, I need to find some time from somewhere
15:31:32 <matel> aye
15:32:00 <matel> ok, I'm done for today.
15:32:04 <BobBall> As am I
15:32:15 <johnthetubaguy> cools
15:32:16 <johnthetubaguy> thanks all
15:32:19 <johnthetubaguy> #endmeeting