15:00:04 <johnthetubaguy> #startmeeting XenAPI
15:00:05 <openstack> Meeting started Wed Nov 13 15:00:04 2013 UTC and is due to finish in 60 minutes.  The chair is johnthetubaguy. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:06 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:08 <openstack> The meeting name has been set to 'xenapi'
15:00:17 <johnthetubaguy> hello everyone
15:00:25 <johnthetubaguy> who is around for today's meeting?
15:00:33 <zigo_> Hi!
15:01:34 <johnthetubaguy> cool, waiting for a few others to drop in
15:02:07 <johnthetubaguy> #topic Blueprints
15:02:31 <johnthetubaguy> #link https://etherpad.openstack.org/p/IcehouseXenAPIRoadmap
15:02:39 <johnthetubaguy> so we had a good design summit session
15:02:49 <johnthetubaguy> just wanted to make sure I got the summary of what we said correct:
15:03:02 <johnthetubaguy> https://etherpad.openstack.org/p/IcehouseNovaSummit
15:03:17 <johnthetubaguy> The big headlines are deprecating stuff
15:03:36 <johnthetubaguy> just wondering about who is going to come forward for some of the other features
15:03:48 <zigo_> I'm sorry I missed them... :(
15:03:52 <johnthetubaguy> and reminding people about getting blueprint in soon, and getting sponsors for those blueprints
15:04:32 <BobBallAway> Sorry
15:04:36 <BobBallAway> I'm here
15:04:41 <BobBallAway> looks like our clocks were wrong
15:04:41 <matel> Hi
15:04:48 <BobBallAway> Mate's clock says it's 2 minutes to
15:04:51 <matel> It's 14:58
15:04:52 <BobBallAway> but mine says 4 minutes past!
15:05:04 <johnthetubaguy> lol
15:05:13 <matel> I'm pretty sure someone has already solved this problem.
15:05:17 <johnthetubaguy> OK, so just wanted to go through my summary of the session
15:05:23 <johnthetubaguy> https://etherpad.openstack.org/p/IcehouseNovaSummit
15:05:39 <johnthetubaguy> mostly just the headlines about deprecating stuff
15:06:03 <johnthetubaguy> Then would be good to talk about blueprints for Icehouse-1 (and maybe Icehouse-2/3)
15:06:14 <BobBallAway> The deprecations should never be headlines... :)
15:06:32 <johnthetubaguy> ah, always headlines else no one reads it
15:07:20 <johnthetubaguy> So, I have some dodgy draft patchs for: VCPU masking and VIF hotplug for XenAPI, so I will try push those soon
15:08:01 <BobBallAway> perfect
15:08:02 <johnthetubaguy> Did them while waiting for the plane first thing in the morning, so it could be all rubbish, not looked at them again
15:08:06 <BobBallAway> Mate and I were just discussing VIF hotplug
15:08:36 <johnthetubaguy> Got a patch for the kernel args stuff, but not seen the KVM patch go in yet, but I may have missed it
15:09:01 <johnthetubaguy> Anyways, what do we think we want to do in Icehouse-1
15:09:02 <BobBallAway> Great
15:09:11 <BobBallAway> Is that something that RS wants to use?
15:09:38 <BobBallAway> The kernel args I mean
15:09:41 <johnthetubaguy> Not yet, as far as I know, I was just bored in the airport, the shops had shut
15:09:56 <BobBallAway> Wow - airport shops that were shut.  That's new!
15:09:57 <johnthetubaguy> just picked some easy stuff to hack on
15:10:11 <johnthetubaguy> Yeah, 2am flight out of HK
15:10:22 <BobBallAway> The real problem with Icehouse-1 is that it's only 4 weeks :(
15:10:30 <BobBallAway> I think of all things -1 should be longer than -2 and -3
15:10:37 <BobBallAway> #link https://wiki.openstack.org/wiki/Icehouse_Release_Schedule
15:10:43 <BobBallAway> is the release schedule for those who haven't seen it
15:10:49 <matel> Let's try to find some small items.
15:10:53 <BobBallAway> When I say 4 weeks, I mean 3 weeks.
15:11:00 <BobBallAway> 4 weeks total, 3 weeks left.
15:11:02 <johnthetubaguy> right, well if they are bigger then so be it right?
15:11:10 <johnthetubaguy> just start now and target Icehouse-2
15:11:15 <BobBallAway> Absolutely
15:11:20 <johnthetubaguy> so whats that list?
15:11:21 <BobBallAway> Targetting Icehouse-2 makes sense to me
15:11:40 <matel> Maybe it's only me, but do we have a proper list, which just lists the items?
15:11:53 <johnthetubaguy> I want to target icehouse-1, assuming I get them pushed up this week
15:12:05 <johnthetubaguy> for those little bits I mentioned, but we can ignore them
15:12:08 <johnthetubaguy> they are low priority
15:12:19 <BobBallAway> I don't think there is a proper list Mate, no
15:12:26 <BobBallAway> We've got the summit etherpad
15:12:31 <matel> Should we create one?
15:12:33 <johnthetubaguy> yeah, just the etherpad
15:12:37 <BobBallAway> but that's got multiple lists
15:12:40 <matel> Maybe a XenAPI wiki page?
15:12:42 <BobBallAway> yes - let's create one
15:12:45 <johnthetubaguy> why?
15:12:48 <matel> So that it's a good starting point?
15:13:08 <johnthetubaguy> just collaspse the stuff in the etherpad into a single list
15:13:11 <BobBallAway> Why on XenAPI wiki?
15:13:12 <johnthetubaguy> if thats what we want
15:13:15 <BobBallAway> yes agree with John
15:13:22 <matel> It would be useful for me to have a proper list.
15:13:31 <BobBallAway> https://etherpad.openstack.org/p/IcehouseXenAPIRoadmap_Condensed
15:13:31 <matel> Which would show the progress...
15:13:35 <johnthetubaguy> what is proper? what info are you looking at?
15:13:45 <johnthetubaguy> well thats the blueprints in launchpad right?
15:13:55 <matel> A list, who is assigned to it, and a link to the blueprint
15:13:59 <johnthetubaguy> we just need to decide what is worth going into a blueprint for Icehouse
15:14:10 <johnthetubaguy> sure
15:14:30 <matel> Okay, let me find the perfect launchpad query.
15:14:38 <johnthetubaguy> well, I think we are probably better doing this offline, and claiming things we want to work on in the etherpad?
15:14:46 <matel> My problem with etherpad is that it's a bit too dynamic...
15:15:05 <johnthetubaguy> its got versions and snapshots etc, just need to use them
15:15:15 <johnthetubaguy> its ours to use as we please right?
15:15:32 <johnthetubaguy> anyways, feel free to put it into a wiki if that works better for you
15:15:42 <matel> A launchpad query is fine.
15:15:45 <BobBallAway> See the condensed list I just made
15:15:50 <BobBallAway> https://etherpad.openstack.org/p/IcehouseXenAPIRoadmap_Condensed
15:15:50 <johnthetubaguy> just wondering if we can decide some big "Themes" of stuff we want to do
15:16:22 <johnthetubaguy> cool, so who ones to own the deprecation stuff?
15:16:30 <johnthetubaguy> wants to own^
15:16:43 <matel> deprecation stuff - what does that mean?
15:16:45 <BobBallAway> Absolutely not - it's fine to say it's deprecated, but I have no plans to actually remove it
15:16:58 <johnthetubaguy> yes, we need to add log messages and stuff
15:17:01 <johnthetubaguy> I can take that on
15:17:13 <matel> you guys overuse stuff
15:17:32 <BobBallAway> So you actually see that as a priority?
15:17:37 <johnthetubaguy> yup
15:17:59 <BobBallAway> heh :)
15:18:11 <BobBallAway> So you'll add a BP for that then
15:18:12 <johnthetubaguy> its extra stuff to maintain otherwise
15:18:31 <johnthetubaguy> I will do what we need to, not sure what that is yet
15:18:31 <matel> How can you specify a blueprint filter?
15:18:35 <BobBallAway> I guess you're managing those two too
15:19:14 <matel> https://blueprints.launchpad.net/nova/?searchtext=xenapi
15:19:20 <matel> Is this the list that we need?
15:19:34 <johnthetubaguy> yeah, that should do the trick
15:19:44 <johnthetubaguy> not sure if you can tag blueprints, maybe not
15:19:50 <BobBallAway> Thought you added the kernel command line support John?
15:19:56 <matel> Let's say this is the list.
15:20:11 <BobBallAway> That's the list of current blueprints
15:20:16 <BobBallAway> however we need to write new ones for Icehouse
15:20:21 <johnthetubaguy> indeed
15:20:30 <johnthetubaguy> so what you fancy for Icehouse
15:20:32 <matel> Or modify the existing ones.
15:20:57 <johnthetubaguy> matel: sure, in some cases, could just mark some obsolete as well
15:21:11 <BobBallAway> Personally I fancy us doing vGPU
15:21:21 <BobBallAway> but I haven't sized things to see if that fits
15:21:48 <johnthetubaguy> thats a good one, marketing wise
15:22:02 <johnthetubaguy> how about PCI passthrough as well, given thats going to be changing good to get that in there
15:22:19 <BobBallAway> It probably will, but I can't promise anything until next wednesday - we'll be having discussions internally on what we're prioritising
15:22:22 <BobBallAway> Maybe
15:22:30 <BobBallAway> PCI pass through is a poor-mans vGPU
15:22:36 <BobBallAway> and other people have expressed interest in helping
15:22:42 <BobBallAway> so I don't want Citrix to commit to it.
15:23:13 <johnthetubaguy> OK
15:23:19 <BobBallAway> I think we can probably say it's likely to make I-3 - even if we don't do it ourselves
15:23:44 <BobBallAway> Don't need to advertise
15:23:51 <johnthetubaguy> why?
15:23:54 <BobBallAway> I'm in discussions with someone about it ATM
15:24:06 <johnthetubaguy> OK, if they can be public it helps people
15:24:14 <BobBallAway> it will be entirely public
15:24:28 <BobBallAway> or did you mean now?  It's some guys from bull.net who are interested
15:24:35 <johnthetubaguy> what I mean is, we can same people are showing an intested, cool
15:25:40 <BobBallAway> yup
15:25:43 <johnthetubaguy> we will need full tempest before Icehouse-2
15:25:47 <johnthetubaguy> I mean 2
15:25:51 <johnthetubaguy> grrr
15:25:53 <johnthetubaguy> I mean 3
15:26:31 <BobBallAway> Yup
15:26:50 <BobBallAway> I'd like to see it in I-1 but realistically it might not make it till I-2
15:26:57 <BobBallAway> depends on what we have to do
15:27:07 <johnthetubaguy> Ok
15:27:18 <johnthetubaguy> I just saw your notes on neutron security groups, thats all
15:27:26 <johnthetubaguy> is that neutron full tempest?
15:27:29 <BobBallAway> yes
15:27:32 <johnthetubaguy> cool
15:27:38 <johnthetubaguy> using ML2?
15:27:41 <BobBallAway> nova network full tempest is "working"
15:27:58 <johnthetubaguy> right, in its broken way
15:28:23 <BobBallAway> Please expand?
15:28:27 <matel> Sorry, what's the question here?
15:28:35 <BobBallAway> Is neutron using ML2
15:28:41 <johnthetubaguy> well, I think you said the tests were not swesome at the moment
15:28:57 <BobBallAway> uhhh I did?
15:29:01 <johnthetubaguy> I was thinking about neutron + XenServer + ML2 and its status
15:29:14 <johnthetubaguy> anyways, we can come back to that
15:29:19 <matel> I don't know, if it's an ML2 config.
15:29:47 <matel> Afaik ML2 is just a framework, isn't it?
15:30:24 <matel> I think secgroups will be broken there as well.
15:30:44 <johnthetubaguy> duno about ML2
15:30:54 <johnthetubaguy> I got the idea it was the next generation plugin
15:31:01 <matel> But the fair answer is that I don't know what is the status of ML2 - XenServer
15:31:02 <johnthetubaguy> but anyways
15:31:13 <johnthetubaguy> I added a few more into the back list with my name
15:31:25 <johnthetubaguy> I notice there is nothing about getting XenServer onto supported APIs further
15:32:03 <BobBallAway> Correct
15:32:10 <BobBallAway> That will be a priority again soon
15:32:18 <BobBallAway> but we need changes in XenServre
15:32:21 <johnthetubaguy> well, that should go on the list I guess
15:32:25 <BobBallAway> and they won't fit in Augusta
15:32:27 <johnthetubaguy> ah
15:32:30 <BobBallAway> so XenServer changes will be post-Icehouse
15:32:31 <johnthetubaguy> maybe not on the list then
15:32:38 <johnthetubaguy> OK...
15:32:40 <BobBallAway> hence no point in making changes in Icehouse for them :)
15:32:46 <johnthetubaguy> so lets move away from blueprints
15:32:59 <johnthetubaguy> we clearly need to all go away and do our own planning
15:33:02 <johnthetubaguy> but we have a start
15:33:13 <BobBallAway> yes
15:33:30 <BobBallAway> That's our current thinking which is great
15:33:35 <johnthetubaguy> cool
15:33:49 <BobBallAway> but all caveated by we need to do internal planning
15:34:17 <BobBallAway> do you care about live migrate for resize if it's being deprecated?
15:34:32 <johnthetubaguy> resize down is deprecated
15:34:34 <johnthetubaguy> not resize up
15:34:52 <BobBallAway> Hang on - isn't resize down is to be removed rather than deprecated?
15:34:52 <johnthetubaguy> maybe that wasn't clear in the session, oops
15:35:00 <johnthetubaguy> we can't remove it
15:35:07 <BobBallAway> oh - why not?
15:35:13 <johnthetubaguy> we can deprecate it in Icehouse then remove in J
15:35:21 <johnthetubaguy> its our agreed contract with users
15:35:35 <matel> I guess we can't remove anything - we need to deprecate things first.
15:35:42 <matel> Which is good.
15:35:42 <BobBallAway> meh
15:35:44 <johnthetubaguy> and give them a chance to shout
15:35:54 <johnthetubaguy> to fix it up themselves, should they want it
15:35:54 <BobBallAway> we change the contract willy nilly in other places :P
15:35:56 <BobBallAway> but fair enough :P
15:36:07 <johnthetubaguy> yeah, we do define where the contract is firm
15:36:09 <johnthetubaguy> the API
15:36:17 <johnthetubaguy> almost everything else may change
15:36:26 <johnthetubaguy> RPC must be backwards compatible
15:36:35 <BobBallAway> ok
15:36:49 <johnthetubaguy> anyways
15:36:58 <johnthetubaguy> #topic Docs
15:37:06 <johnthetubaguy> any updates here?
15:37:11 <johnthetubaguy> any new plans?
15:38:11 <BobBallAway> 'fraid no updates or new plans
15:38:13 <BobBallAway> we need to do it
15:38:21 <johnthetubaguy> sure, we can agree that :)
15:38:34 <johnthetubaguy> #topic Bugs and QA
15:38:50 <johnthetubaguy> any updates on getting tempest running on all commits?
15:39:06 <BobBallAway> yes
15:39:14 <BobBallAway> There's lots of work to do and lots of options to look at
15:39:38 <BobBallAway> and we might need some hardware - so we'll probably be asking Rackspace if they can help out at all on that :)
15:39:47 <BobBallAway> Chances are we have to use TripleO to get gating
15:39:58 <johnthetubaguy> so how have you got running it in the cloud?
15:40:03 <BobBallAway> because even if we can get it working in RS cloud, we need redundancy and it can't run in HP cloud
15:40:11 <johnthetubaguy> we can have multiple regions
15:40:28 <BobBallAway> Monty was quite clear about RS vs HP cloud
15:40:30 <johnthetubaguy> I am worried here about tempest running, not us gating on things
15:40:46 <johnthetubaguy> I know we should have both to become a gate
15:40:48 <BobBallAway> he didn't mention regions - and I assume he knows about them
15:40:52 <BobBallAway> Ah, fair enough
15:41:06 <johnthetubaguy> I worry about tempest tests asap
15:41:11 <BobBallAway> well - our focus is gating.  Not sure if we'll be pursuing full tempest non gating
15:41:29 <johnthetubaguy> we can talk to him separately about different regions, we have a lot more regions now then when they last looked
15:41:34 <johnthetubaguy> Ah
15:41:42 <johnthetubaguy> can I look at helping with tempest
15:41:47 <johnthetubaguy> I don't want to drop out of tree
15:41:58 <BobBallAway> Agreed
15:42:03 <BobBallAway> What are you thinking we'd do though?
15:42:08 <johnthetubaguy> an non-voting zuul test using rax cloud seems good
15:42:09 <BobBallAway> use xenserver-core on a RS performance flavor?
15:42:15 <johnthetubaguy> yup
15:42:25 <BobBallAway> That's pretty close now I think/hope... ... ... :)
15:42:30 <BobBallAway> as we discussed
15:42:41 <johnthetubaguy> yes, but you said you were not going to bother with that any more
15:42:52 <johnthetubaguy> I see that as top priority myself
15:43:19 <BobBallAway> Let me refresh my memory on the current status then we can sync up outside of the meeting
15:43:24 <johnthetubaguy> OK
15:43:33 <johnthetubaguy> well I am happy to help join up some docs
15:43:36 <johnthetubaguy> dots^
15:43:51 <johnthetubaguy> if thats what we need to get tempest tests
15:44:03 <BobBallAway> As we talked about at the summit I think the missing piece is that we might need to have a Nova call to boot a VM in HVM mode
15:44:04 <johnthetubaguy> I can use the same thing to look at cloud-cafe tests, with cells, etc, etc
15:44:30 <johnthetubaguy> well, we have image sharing soon I hope, which is an alternative
15:44:45 <BobBallAway> Maybe
15:44:53 <BobBallAway> Monty wanted it to be set up on default images
15:44:58 <BobBallAway> rather than a custom made one
15:45:03 <BobBallAway> but perhaps he can be convinced on that
15:45:09 <johnthetubaguy> well, we can see what we can do
15:45:34 <johnthetubaguy> it might just mean building the cached template stuff etc could be a little harder
15:45:47 <johnthetubaguy> lets sync up about the options
15:45:50 <BobBallAway> yup
15:46:07 <johnthetubaguy> #topic Open Discussion
15:46:17 <johnthetubaguy> anything else people are worrying about?
15:46:21 <matel> A q
15:46:39 <johnthetubaguy> matel: fire away
15:46:42 <matel> Could we have this meeting in a video conference or something like that?
15:47:12 <matel> Or what is your idea about it?
15:47:20 <johnthetubaguy> not really, IRC is the default, because you get minuets
15:47:38 <johnthetubaguy> there is a phone bridge that openstack has setup
15:47:42 <BobBallAway> I think I'd rather see us change the things that mean that the IRC meetings are effective
15:48:10 <johnthetubaguy> sure, basically, what is causing you problems with participating?
15:48:16 <johnthetubaguy> ideas are welcome
15:48:27 <johnthetubaguy> lets try them, if its better, we keep them
15:48:28 <BobBallAway> they are not as useful as they should be ATM because the three of us are so close and talk during the week - so it's usually just a reminder of what we've already talked about
15:48:56 <johnthetubaguy> well, there is no point it just being a reminder, I usually find its new stuff for me
15:48:57 <BobBallAway> If we had more participants from RS or the guys from bull.net (which I'll suggest) then the meeting would make more sense :P
15:49:11 <johnthetubaguy> what time would work for bull.net?
15:49:29 <BobBallAway> I don't know - but I'm sure its' not a question of the meeting time.
15:49:38 <johnthetubaguy> there is little for other at Rax to contribute at the moment, we are not really doing loads of work on the xenapi layer, its mostly all above that
15:49:43 <BobBallAway> As I said - I'll suggest it
15:50:16 <johnthetubaguy> sure, well happy to get bull.net more involved in these meeting, would be good to understand what is blocking participation
15:50:43 <johnthetubaguy> we could try google hangout or whatever if this really doesn't work for everyone
15:50:51 <johnthetubaguy> but not sure how much that will buy us
15:51:10 <matel> Okay, then not do that.
15:51:29 <matel> I think it would be better if we had more people here.
15:51:51 <matel> having a hangout would be a more closed thing.
15:51:55 <BobBallAway> Even if they are only interested parties rather than people coding XenAPI
15:52:16 <johnthetubaguy> well, not sure who all these people would be, I have invited others working on XenAPI at Rax to drop in when they do related stuff
15:52:32 <johnthetubaguy> but thats just not very often, apart from chasing down the odd bug
15:53:08 <BobBallAway> It is a shame - cuz there are quite a few commits in the XenAPI tree from others at RS
15:53:58 <BobBallAway> e.g. one where it would have made a lot of sense was the ghosting work done a while ago, or the memory prediction stuff - they would have been useful to have IRC discussions about rather than discussions on the reviews
15:54:00 <matel> I suggested the hangout, because sometimes I get lost in the communication, but that might only be me.
15:54:17 <johnthetubaguy> well, we need to rework all that I think
15:54:23 <johnthetubaguy> so might be worth it
15:54:32 <johnthetubaguy> the main issue is this time doesn't work well for most of them
15:55:08 <BobBallAway> Well we can move an hour later or to a diff day if that works better
15:55:32 <matel> What time would work for them?
15:55:39 <johnthetubaguy> thing is the old time we had that worked fro them, didn't work for you guys
15:56:01 <matel> What's the time that was working for them?
15:56:03 <BobBallAway> But they weren't coming :)
15:56:11 <johnthetubaguy> sure
15:56:15 <BobBallAway> If they were attending the meeting then it would have been a good reason to find a comproimise or stick with that time
15:56:17 <johnthetubaguy> so lets just see what happens
15:56:23 <BobBallAway> but since it was only us it didn't make sense to do it so late
15:56:36 <johnthetubaguy> OK
15:56:57 <BobBallAway> But this time should be OK most of the time anyway
15:57:08 <matel> The main point here is that we just need to make it clear: these meetings are not useful enough, and we need to understand why is that.
15:57:13 <BobBallAway> I used to have my call with Ant at this time
15:57:14 <BobBallAway> :)
15:57:41 <BobBallAway> I agree with Mate - I know you said you get new stuff John - but from my perspective they could be - and should be - more useful than I find them ATM
15:57:47 <johnthetubaguy> yeah, he is in the texas office, all the nova devs are working from home
15:58:01 <BobBallAway> Ahh - didn't know that
15:58:09 <BobBallAway> so we have to work with west coast? :/
15:58:28 <johnthetubaguy> our lead dev is west coast, comstud
15:58:36 <BobBallAway> that's a pain...
15:58:48 <BobBallAway> hadn't realised that
15:59:43 <BobBallAway> Anyway
15:59:44 <BobBallAway> we're out of time
15:59:59 <BobBallAway> John - could you have a chat with comstud and see if it's the meeting time that stops him attending, or something else?
16:00:07 <BobBallAway> maybe others in the team you think would be useful
16:00:17 <BobBallAway> I'll talk with the bull.net guys to see if they can come sometimes
16:00:25 <johnthetubaguy> well I don't think he generally has anything to discuss about XenAPI, he is working on the objects stuff and cells right now
16:00:33 <johnthetubaguy> #endmeeting