16:02:31 <primeministerp> #startmeeting hyper-v 16:02:32 <openstack> Meeting started Tue May 7 16:02:31 2013 UTC. The chair is primeministerp. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:02:33 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:02:35 <openstack> The meeting name has been set to 'hyper_v' 16:02:51 <primeministerp> so let's begin 16:03:27 <primeministerp> I guess a good place to start is next steps post summit 16:03:37 <primeministerp> where we currently are in that process 16:03:53 <primeministerp> what is priority after the design session 16:04:33 <primeministerp> alexpilotti: would you like to start? 16:04:47 <alexpilotti> sure 16:04:51 <primeministerp> kpavel_: hi there 16:05:14 <primeministerp> alexpilotti: or is there something else we should address prior? 16:05:28 <alexpilotti> quite a lot coming for Havana 16:05:50 <primeministerp> let's start at the top priority 16:05:55 <primeministerp> wmiv2 16:05:56 <alexpilotti> we got a lot of requests about how to handle clustering to begin with 16:06:32 <alexpilotti> most Windows workloads are based on the assumption that the HA is handled by the OS 16:06:33 <primeministerp> from a "hyper-v cluster" as a compute node perspective 16:06:58 <alexpilotti> we need to support a Hyper-V windows cluster as a compute node 16:07:20 <alexpilotti> we can use an approach similar to what vsphere took 16:07:39 <primeministerp> but use the hyper-v/cluster api's as the access point 16:07:47 <primeministerp> to that functionality 16:07:56 <alexpilotti> yep 16:08:03 <primeministerp> just to confirm 16:08:20 <alexpilotti> teh idea is that the cluster is seen as a single compute node 16:08:45 <alexpilotti> but Nova will be able to get data about each individual host 16:08:48 <primeministerp> and the best part is hyper-v clustering features ship w/ the free hyper-v server 16:08:53 <alexpilotti> yep 16:09:08 <alexpilotti> the only thing is that we'll need a DC 16:09:25 <primeministerp> we already do need one for live migration 16:09:36 <alexpilotti> in order to have a 100% free solution 16:09:47 <primeministerp> then samba4 16:09:52 <alexpilotti> we can try to see how Samba4 works 16:09:54 <alexpilotti> yep :-) 16:10:08 <alexpilotti> I tried w a beta a few months ago 16:10:17 <primeministerp> I have some people to reach out to if we have issues 16:10:20 <alexpilotti> and I had issues with Kerberos delegation 16:10:23 <primeministerp> from the samba side of the things 16:10:37 <alexpilotti> that'd be great 16:10:51 <alexpilotti> on the API perspective 16:10:57 <primeministerp> well on the samba perspective 16:11:04 <primeministerp> and on the api perspective 16:11:07 <alexpilotti> Windows has some higher level Powershell API 16:11:10 <primeministerp> and i can dig internally to see 16:11:26 <alexpilotti> sorry, I was starting a new sentence :-) 16:11:26 <primeministerp> what is available for the cluster api interfaces 16:11:43 <alexpilotti> so we have Powershell and COM low level interfaces 16:11:44 <primeministerp> and documentation 16:11:59 <alexpilotti> MS is suggesting Powershell, but it's a useless overhead for us 16:12:08 <alexpilotti> so I'd go with COM 16:12:18 <alexpilotti> depending on the level of PINTA, to say so :-) 16:12:29 * primeministerp silently agrees 16:12:55 <primeministerp> we don't need additional overhead 16:12:55 <alexpilotti> We still have to target this one, but Havana-2 is a good target IMO 16:13:14 <alexpilotti> BTW Havana-1 is behind the corner (ca 1 month) 16:13:21 <primeministerp> wtf 16:13:34 <alexpilotti> well, next month is June :-) 16:13:54 <primeministerp> yes it is 16:13:57 <alexpilotti> let me see if I find the deadline scheduling 16:14:08 <alexpilotti> so I expect Havana-2 in July 16:14:15 <alexpilotti> and Havana-3 in August 16:14:34 <alexpilotti> lst year we started working end July / begin August 16:14:42 <alexpilotti> in time for Folsom-3 16:14:54 <alexpilotti> so it makes sense (unfortunately)! :-) 16:15:25 <primeministerp> I'm not sure but i thought I heard something about releases shifting dates slightly 16:15:30 <primeministerp> but I coudld be wrong 16:15:35 <primeministerp> anyway 16:15:41 <primeministerp> so 16:15:57 <alexpilotti> #link https://wiki.openstack.org/wiki/Havana_Release_Schedule 16:15:59 <alexpilotti> here it is 16:16:01 <primeministerp> thx 16:16:15 <alexpilotti> even worse, May 30th 16:16:24 <primeministerp> heh 16:16:36 <alexpilotti> but the good news is that the final one is after the summer ;-) 16:16:59 <alexpilotti> which means that the Gods decided not to **** up our summer this time :-D 16:17:23 <primeministerp> alexpilotti: and we're not trying to get code resurrected ;) 16:17:35 <primeministerp> so moving on 16:17:35 <alexpilotti> lol ;-) 16:17:39 <alexpilotti> ok 16:17:40 <primeministerp> past cluster 16:17:41 <alexpilotti> V2 16:17:45 <alexpilotti> WMI 16:17:47 <primeministerp> yes 16:17:51 <primeministerp> that's the big one 16:18:13 <alexpilotti> the Grizzly refactoring gives us a great start 16:18:23 <alexpilotti> all the code is well layered 16:18:47 <alexpilotti> meaning that we have to add the V2 "utils" classes 16:19:05 <alexpilotti> and a factory that will allocate the right one based on the OS 16:19:17 <primeministerp> alexpilotti: to deal w/ backward compatibality 16:19:18 <alexpilotti> V1 on 2008R2 (legacy, soon unsupported) 16:19:27 <alexpilotti> and V2 for 2012 and above 16:20:05 <alexpilotti> the testing will be way easier, as all the rest of the code will be untouched 16:20:15 <primeministerp> *nod* 16:20:19 <alexpilotti> anyway it'll still be a long work 16:20:33 <alexpilotti> that blocks the next step (VHDX stupport) 16:20:41 <primeministerp> alexpilotti: we need to followup on win-next 16:20:51 <alexpilotti> cool 16:20:59 <primeministerp> alexpilotti: we'll i 16:21:11 <alexpilotti> I'd schedule WMIV2 for Havana-1 16:21:21 <alexpilotti> in order to unlock the rest 16:21:31 <primeministerp> alexpilotti: sounds good to me 16:21:39 <alexpilotti> Questions or should we go on? 16:21:55 <primeministerp> sounds good 16:21:59 <primeministerp> move on 16:22:08 <pnavarro> +1 16:22:33 <primeministerp> alexpilotti: quantum 16:22:58 <pnavarro> NVGRE ! NVGRE ! 16:23:16 <alexpilotti> There's something more than NVGRE coming ;-) 16:23:30 <alexpilotti> but we have to wait a few days to get a final confirmation 16:23:32 <primeministerp> let's just say we're finalizing details 16:23:59 <alexpilotti> NVGRE is still in the list, but priority would be shifted quite a bit down to say so 16:24:10 <primeministerp> yes 16:24:33 <alexpilotti> on QUantum we need anyway the Agent support in our plugin 16:24:49 <primeministerp> so let's hold till next week on that discussion so we can confirm/deny 16:24:57 <alexpilotti> I mean, the new Agent API. I was trying to sneak them in for G3 16:25:07 <alexpilotti> but Quantum was frozen already 16:25:09 <primeministerp> alexpilotti: H3 16:25:14 <alexpilotti> G3 16:25:17 <primeministerp> o 16:25:19 <alexpilotti> in march 16:25:20 <primeministerp> gotcha 16:25:24 <alexpilotti> so the work is mostly done 16:25:37 <alexpilotti> I just need to retarget the patchset 16:25:54 <alexpilotti> there are a gazillion BP on Quantum 16:26:09 <alexpilotti> one of them is related to merge all the agents in one 16:26:12 <primeministerp> i would imagine 16:26:17 <alexpilotti> *the plugins, sorry 16:26:40 <alexpilotti> most plugins are very similar: OVS, LinuxBridge, Hyper-V, etc 16:26:51 <alexpilotti> so a refactoring makes definitely sense 16:27:45 <primeministerp> and from release perspective? 16:27:59 <primeministerp> on refactor work 16:28:40 <alexpilotti> I suggest to postpone the Quantum discussion to the next meeting, hopefully we'll have more details to disclose 16:28:40 <alexpilotti> what do you think primeministerp? 16:28:51 <primeministerp> alexpilotti: yes 16:29:32 <primeministerp> alexpilotti: quantum/networking discussion until we get appropriate confirmation 16:29:41 <primeministerp> er postpone 16:30:09 <primeministerp> are you including the VHDX work in the WMIV2 migration? 16:30:41 <alexpilotti> sure 16:30:44 <primeministerp> ok 16:30:45 <alexpilotti> ops, no :-) 16:30:52 <primeministerp> I didn't think it was 16:30:55 <primeministerp> wmi v2 first 16:31:00 <alexpilotti> WMIV2 "unlocks" the VHDX badge :-) 16:31:01 <primeministerp> vhdx after that is complete 16:31:02 <primeministerp> yes 16:31:09 <primeministerp> +200 mana 16:31:13 <primeministerp> for hyper-v 16:31:14 <primeministerp> ;) 16:31:33 <alexpilotti> so right after we are done with V2 VHDX support is fairly fast to achieve 16:31:57 <alexpilotti> so H2 is definitely a good target 16:32:01 <primeministerp> good 16:32:16 <primeministerp> fibre channel support? 16:32:19 <primeministerp> next? 16:32:22 <alexpilotti> sure 16:32:31 <primeministerp> if i can have san and hba's 16:32:40 <alexpilotti> FC dependes largely on HP's lab stuff 16:32:45 <alexpilotti> cool 16:32:55 <primeministerp> so I have both in my lab already 16:32:57 <alexpilotti> as soon as we have a lab set up we can start developing 16:32:58 <primeministerp> I just need to connect 16:33:03 <primeministerp> it together 16:33:18 <primeministerp> I can see if i can cobble it so we have something 16:33:22 <primeministerp> to start 16:33:25 <primeministerp> if interested 16:33:25 <alexpilotti> it's gonna be a bigger PINTA compared with iSCSI 16:33:32 <primeministerp> i would imagine 16:33:36 <alexpilotti> pnavarro: anything to add on FC? 16:34:28 <primeministerp> well 16:34:47 <primeministerp> priority over that work would possibly be AD/Keystone work 16:35:54 <primeministerp> ? 16:35:59 <primeministerp> sorry that was a question 16:37:39 <primeministerp> alexpilotti: we'll continue then 16:37:49 <primeministerp> alexpilotti: think people are nodding off 16:37:51 <primeministerp> ;) 16:38:05 <alexpilotti> oki, sorry about that 16:38:08 <alexpilotti> :-) 16:38:09 <primeministerp> np 16:38:22 <primeministerp> so take aways 16:38:22 <alexpilotti> AD / keystone sure 16:38:31 <primeministerp> let's focus on the H1 16:38:33 <primeministerp> milestones 16:38:33 <alexpilotti> I'd like to hear Cern opinion 16:38:40 <primeministerp> luis_fdez: ping 16:39:38 <primeministerp> it's late over there 16:39:52 <primeministerp> let's finish up, then 16:39:55 <primeministerp> alexpilotti: ^ 16:39:55 <alexpilotti> oki, let's move on 16:40:07 <primeministerp> alexpilotti: so that's a good start 16:40:16 <alexpilotti> We're going to set up the BPs 16:40:21 <primeministerp> yes 16:40:25 <alexpilotti> there's quite a lot of work as usual 16:40:41 <primeministerp> as usual 16:40:41 <alexpilotti> ah, there's that annoying bug on cinder-volume to fix 16:40:50 <primeministerp> pnavarro: ^ 16:40:54 <alexpilotti> the eventlet stuff 16:41:03 <alexpilotti> I got a guy writing me about that 16:41:11 <primeministerp> can you cc me on the thread 16:41:20 <primeministerp> if possible 16:41:37 <alexpilotti> pnavarro: would you like to sync about it and also talk about the Havana stuff that you'd like to work on? 16:44:22 <alexpilotti> ok, no ping from pnavarro :-) 16:44:29 <pnavarro> sorry al 16:44:31 <pnavarro> alex 16:44:35 <pnavarro> I was on phone 16:44:36 <alexpilotti> np :-) 16:44:53 <alexpilotti> pnavarro: ^^ 16:45:42 <pnavarro> so, about cinder, yes, there are some fixes to do 16:46:03 <pnavarro> and the tests should be refactored to follow the other projects 16:46:15 <alexpilotti> pnavarro: +1 16:46:57 <alexpilotti> is there any specific area you'd like to work on for Havana? 16:47:00 <pnavarro> I'd like to add the missing features copy image to volume, volume to image 16:47:06 <alexpilotti> cool 16:47:38 <alexpilotti> ociuhandu told me that image to volume works already 16:47:50 <alexpilotti> as far as I remember 16:47:59 <pnavarro> mmm 16:48:05 <alexpilotti> but I have to recheck this 16:48:12 <pnavarro> in the driver there are NotImplement 16:48:35 <pnavarro> I can help in nova gaps, for example, ephemeral 16:48:53 <alexpilotti> cool, that's great 16:49:16 <ociuhandu> alexpilotti: have to check that, was a bit ago and have to go through the things to refresh the short memory buffer 16:49:22 <alexpilotti> the list of new features / BPs we started here is still uncomplete 16:49:54 <alexpilotti> ociuhandu: could be that it was simply related to the Linux driver on devstack? 16:49:58 <pnavarro> Fibre channel would be cool too 16:50:28 <alexpilotti> ociuhandu: and not to the WIndows one and my braincells just mixed them up? 16:50:40 <alexpilotti> pnavarro: cool! 16:51:05 <alexpilotti> since we are running late my 2c are on continuing the list of BPs next week 16:51:31 <alexpilotti> as we still have quite a lot of stuff in the list from the design summit 16:52:01 <alexpilotti> and from what we discussed today there's definitely enough to work for a while :-) 16:52:15 <primeministerp> sounds good 16:52:15 <alexpilotti> but if you guys prefer we can go on now 16:52:28 <primeministerp> alexpilotti: we have a meeting in a few mintues 16:52:30 <primeministerp> er minutes 16:52:36 <primeministerp> that we should just get over 16:52:38 <alexpilotti> yep 16:52:41 <primeministerp> perfect 16:52:48 <primeministerp> then let's continue this discussion next week 16:52:52 <alexpilotti> cool 16:53:53 <primeministerp> ok 16:53:56 <primeministerp> closing meeting 16:54:00 <primeministerp> thanks everyone 16:54:04 <primeministerp> #endmeeting