12:59:49 <claudiub> #startmeeting hyper-v 12:59:50 <openstack> Meeting started Wed Aug 24 12:59:49 2016 UTC and is due to finish in 60 minutes. The chair is claudiub. Information about MeetBot at http://wiki.debian.org/MeetBot. 12:59:52 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 12:59:54 <openstack> The meeting name has been set to 'hyper_v' 13:00:10 <claudiub> helloo 13:00:15 <claudiub> waiting for people to join 13:02:14 <atuvenie> hello all 13:02:55 <sagar_nikam> Hi All 13:03:07 <claudiub> helloo. :) 13:03:16 <claudiub> sagar_nikam: anyone else joining us? 13:03:48 <sagar_nikam> claudiub: i think no.. we can start 13:03:53 <claudiub> yeay 13:04:17 <claudiub> #topic OpenvSwitch certification status 13:04:32 <claudiub> soo, about this, there are 2 good news, and one bad news. 13:04:44 <claudiub> one of the good news is that there is no bad news. :) 13:04:54 <claudiub> the other good news is that it is done. 13:05:31 <claudiub> #link Windows Server Catalog entry for OpenvSwitch: https://www.windowsservercatalog.com/item.aspx?idItem=18117a8c-c7bf-f20c-9185-3a53117b9875&bCatID=1638 13:05:55 <sagar_nikam> you mean microsoft certification for OVS is done ? 13:06:00 <claudiub> yep 13:06:05 <sagar_nikam> wow !!!!! 13:06:08 <sagar_nikam> thats great 13:06:13 <claudiub> yep. :) 13:06:19 <sagar_nikam> thank you all from cloudbase for getting it done 13:06:26 <claudiub> no problem. :) 13:06:59 <claudiub> ok, next topic. 13:07:30 <claudiub> #topic designate patch status 13:07:37 <claudiub> merged. 13:08:03 <claudiub> #link Microsoft DNS Server backend support in designate: https://review.openstack.org/#/c/332703/ 13:08:09 <sagar_nikam> looks like ... today it is good news day !!!! 13:08:25 <claudiub> :D 13:08:35 <claudiub> for now. :D 13:08:43 <lpetrut> hey guys 13:09:06 <claudiub> #topic Hyper-V OVS VIF plug driver 13:09:28 <claudiub> #link Windows support for OovsPlugin in os-vif: https://review.openstack.org/#/c/344433/ 13:09:31 <claudiub> this has a +2 13:09:34 <claudiub> yeay 13:09:55 <claudiub> the patch on nova though has a -2. naaay. 13:10:21 <claudiub> so, I'll have to reach an agreeable solution with danpb for this. 13:10:42 <claudiub> might have to make another patch to os-vif. 13:10:57 <sagar_nikam> ok 13:11:07 <claudiub> next topic 13:11:12 <sagar_nikam> can it make it to O ? 13:11:21 <claudiub> i hope it can still make it in N. 13:11:21 <sagar_nikam> os-vif changes 13:11:27 <sagar_nikam> i mean nova changes 13:11:34 <sagar_nikam> ok... thats good 13:11:48 <claudiub> we have until next week, i think to get it merged. 13:12:07 <claudiub> as next week it is going to be feature freeze 13:12:19 <sagar_nikam> ok 13:12:32 <claudiub> next topic 13:12:54 <claudiub> #topic monasca patches status 13:13:33 <claudiub> I've sent an email with logs, config files, metrics definitions and samples collected from my Windows Server 2012 R2 node. 13:13:55 <claudiub> might have to write a couple more unit tests. 13:14:09 <claudiub> but IMO, it is mergeable. 13:14:49 <claudiub> monasca doesn't have any tempest test that validates the agent 13:15:09 <claudiub> monasca doesn't have any tempest test that validates the agent's functionality or its collected metrics, so I guess that makes them hesitant when accepting code. 13:15:58 <claudiub> didn't get much review though.. 13:16:42 <claudiub> well, the monasca meeting is in 2 hours. going to sneak in there and push a little. 13:17:44 <claudiub> anyways, next topic 13:17:58 <claudiub> #topic Barcelona OpenStack Summit 13:18:34 <sagar_nikam> claudiub: let me try to check with monasca core team 13:18:36 <claudiub> soo, most of us are going to be there. There is also going to be a Winstackers work session as well. 13:18:39 <sagar_nikam> i will reach out to them 13:19:08 <claudiub> sagar_nikam: from your team, who is going to go to the Barcelona Summit? 13:19:29 <sagar_nikam> i think sonu: 13:19:34 <sagar_nikam> let me ask 13:19:41 <sagar_nikam> that is from india team 13:19:53 <sagar_nikam> i think paul murray will be there 13:21:08 <claudiub> cool. this is just an fyi, you folks welcome to join the work session. :) 13:22:04 <claudiub> #topic Hyper-V Cluster spec 13:22:27 <sagar_nikam> claudiub: i will let sonu know about it 13:22:29 <claudiub> well. it didn't get any reviews yet. 13:23:07 <sagar_nikam> :( 13:24:30 <claudiub> sagar_nikam: do you think you can try to get some help on this? 13:24:56 <claudiub> we could use some help in getting it approved. :) 13:25:13 <sagar_nikam> the BP ? 13:25:17 <claudiub> yeo 13:25:19 <claudiub> yep 13:25:30 <claudiub> otherwise the nova-core is just going to ignore it 13:25:31 <sagar_nikam> ok let me try 13:25:31 <sagar_nikam> link please ? 13:25:32 <sagar_nikam> ok let me try 13:25:33 <sagar_nikam> link please ? 13:25:49 <claudiub> #link Hyper-V Cluster spec: https://review.openstack.org/#/c/350559/ 13:26:04 <sagar_nikam> thanks. i will try 13:26:10 <claudiub> cool. thanks! 13:26:20 <claudiub> #topic open discussion 13:26:31 <claudiub> well, that all the news from me. 13:26:44 <claudiub> anything else that needs to be discussed? 13:29:27 <sagar_nikam> claudiub: i did a clone on monasca-agent 13:29:36 <sagar_nikam> newton 13:29:45 <claudiub> cool. did it work? 13:30:04 <sagar_nikam> initial thoughts ... it looks a bit complex to backport your patches to Mitaka 13:30:14 <sagar_nikam> i tried only today 13:30:19 <sagar_nikam> so will try again 13:30:23 <claudiub> yeah, that's what i said. 13:30:33 <claudiub> that's why i suggested trying the newton monasca-agent branch 13:30:45 <sagar_nikam> ok ... one more idea 13:30:53 <sagar_nikam> what if i take newton branch 13:30:59 <sagar_nikam> and then your patches 13:31:07 <sagar_nikam> and try to connect to my mitaka controller ? 13:31:10 <sagar_nikam> will it work 13:31:14 <claudiub> yeah, that's what i was suggesting. 13:31:20 <sagar_nikam> nova will work.. not sure monasca will work 13:31:30 <sagar_nikam> ah .... i missed that point 13:31:36 <sagar_nikam> ok ... will try tomorrow 13:31:45 <claudiub> not sure either, but it doesn't hurt to try. :) 13:31:58 <sagar_nikam> have you tested with devstack ? 13:32:16 <claudiub> yeah 13:32:27 <sagar_nikam> ok ... 13:32:34 <sagar_nikam> let me try first with my setup 13:32:37 <claudiub> that;s how i got it working, and how i tested it. :) 13:32:44 <sagar_nikam> else i will try with devstack 13:33:06 <sagar_nikam> if all the 14 patches merge... are we done with monasa for hyperv ? 13:33:17 <claudiub> you've seen my mail to the monasca folks, right? i've included a link with config files, logs, and so on. 13:33:38 <sagar_nikam> yes ... saw that 13:33:41 <sagar_nikam> nice... 13:33:59 <claudiub> sagar_nikam: yeah, if you use all the checkers i've mentioned in the email 13:34:08 <sagar_nikam> good 13:34:15 <claudiub> cpu, disk, host_alive, hyperv, iis, memory, network, process, wmi_check 13:34:23 <sagar_nikam> what is the plan on monasca-log ? 13:34:46 <claudiub> i haven't tried any other checkers, but roland said that those are good enough, as they are the most used 13:35:12 <claudiub> i didn't try monasca-log yet. 13:35:38 <sagar_nikam> are you planning ? 13:36:01 <sagar_nikam> is that required as part of monasca support for hyperv ? 13:36:22 <claudiub> i'll give it a shot, if time allows it 13:36:39 <claudiub> but it's optional, IMO. 13:36:45 <sagar_nikam> ok 13:37:12 <claudiub> if we can get the monasca-agent patches in, I'd be happy with the result for this cycle. :) 13:37:24 <sagar_nikam> i will try to get it reviewed 13:38:23 <sagar_nikam> now moving on... what's new coming in O 13:38:29 <sagar_nikam> not just nova 13:38:35 <sagar_nikam> other components as well 13:38:39 <sagar_nikam> for hyperv 13:39:02 <claudiub> we're thinking about it, and building a list. we'll have to propose blueprints and specs for most of them. 13:39:16 <sagar_nikam> ok 13:40:10 <claudiub> for now, merge the stuff that didn't get in this cycle: nova: vNUMA support, setting instance boot order, boot from ISO, set disk / volume QOS, UEFI secure boot, shielded VMs. i might be missing a few one. 13:40:17 <claudiub> the code is up for those already. 13:40:26 <sagar_nikam> ok 13:40:33 <sagar_nikam> that is a big list 13:40:40 <claudiub> as for other things: we're going to take a look at SRIOV, PCI device passthrough. 13:40:53 <sagar_nikam> ok 13:41:11 <claudiub> freezer support will be in O as well. 13:41:18 <sagar_nikam> ok 13:41:25 <claudiub> it isn't done yet. 13:41:28 <sagar_nikam> magnum ? 13:42:02 <claudiub> magnum works as long as kubernetes / docker swarm / mezos works on windows. 13:42:16 <sagar_nikam> ok 13:42:28 <claudiub> so, that's not tied to any openstack release cycle. 13:42:56 <sagar_nikam> i think manila support is already there 13:43:26 <claudiub> other than that, we're going to add objects to os-win, which will make it more user-friendly and easier to understand and work with. plus, it will improve the instance spawn times. 13:43:30 <claudiub> sagar_nikam: yep. 13:44:02 <sagar_nikam> claudiub: what change in os-win to speed instance spawn ? 13:44:08 <claudiub> currently, we're working on adding a Windows / Hyper-V CI for the oslo libraries, so we know that they won't break us in the future. :) 13:45:21 <claudiub> sagar_nikam: so, as you know, the time needed to spawn an instance on hyper-v was greatly reduced thanks to os-win and pymi. Last time we tested it, the spawn time was close to the libvirt times. We can reduce it even more. :) 13:45:39 <sagar_nikam> that is intresting 13:45:55 <sagar_nikam> i agree that os-win and pymi improved the perf drastically 13:46:07 <claudiub> the thing is, currently the hyper-v nova driver queries for the vm object for every operation: update_vm, attach_disk, attach_nic, power_on, and so on. 13:46:11 <sagar_nikam> just curious to know what more is planned 13:47:08 <sagar_nikam> ok got it 13:47:14 <claudiub> so, if we don't query the vm object every single time, the time improves. WMI queries can take a lot of time, and that's why we reduced and optimised them. 13:48:11 <claudiub> anyways. 13:48:34 <claudiub> python3 support is still a must. 13:48:53 <claudiub> nova still requires a lot of changes for that. 13:49:55 <claudiub> anyways, if there are any other features we should consider, let us know. :) 13:50:23 <sagar_nikam> livirt driver 13:50:32 <sagar_nikam> is adding lot of support for NFV usecases 13:50:50 <sagar_nikam> may be we can try some of those as BPs for hyperv 13:51:18 <claudiub> well, the vNUMA is used for the NFV usecases, as well as sriov and pci passthrough. 13:51:31 <sagar_nikam> agree 13:52:31 <claudiub> you guys have some nfv usecases / environment for it? once we're done, it would be cool to see how the performance on hyper-v compares to kvm. 13:52:58 <sagar_nikam> yes... i think 13:53:14 <sagar_nikam> let me check with scale team 13:53:30 <sagar_nikam> i will get back on it 13:53:39 <sagar_nikam> atleast the usecase tried for libvirt 13:53:45 <sagar_nikam> in scale tests 13:54:07 <sagar_nikam> not sure if we can get access to that setup... will try 13:54:14 <claudiub> cool, let me know when you find out. :) 13:54:55 <claudiub> anything else you'd like to discuss? 13:55:16 <sagar_nikam> i think if we can achieve the perf on par with libvirt 13:55:25 <sagar_nikam> for NFV usescases 13:55:31 <sagar_nikam> then it would be amazinf 13:55:36 <sagar_nikam> amazing 13:55:51 <sagar_nikam> especially with windows nano 13:56:02 <sagar_nikam> i think it will be a good fit for NFV 13:56:05 <claudiub> yep. :) that would be great. :D 13:56:08 <sagar_nikam> what is your opinion 13:56:18 <sagar_nikam> windows nano for NFV ? 13:56:53 <sagar_nikam> NFV is a big usecase ..especially for libvirt 13:56:57 <claudiub> hm, don't really know. I haven't tried nano yet. 13:57:06 <sagar_nikam> ok 13:57:12 <sagar_nikam> we are almost done with time 13:57:18 <sagar_nikam> thanks for the 2 good news 13:57:35 <sagar_nikam> hopefully next few weeks we will have the 3rd good news on monasca 13:57:35 <claudiub> no problem. :) 13:57:46 <claudiub> agree to that. :) 13:57:53 <sagar_nikam> i am done 13:58:02 <sagar_nikam> anything from your end ? 13:58:05 <claudiub> well then, we can end the meeting here. :) 13:58:14 <sagar_nikam> sure thanks !!! 13:58:29 <claudiub> bye folks, thanks for joining. see you next week! 13:58:32 <claudiub> #endmeeting