13:01:14 #startmeeting hyper-v 13:01:15 Meeting started Wed Sep 7 13:01:14 2016 UTC and is due to finish in 60 minutes. The chair is claudiub|2. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:17 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:01:18 hi all 13:01:19 The meeting name has been set to 'hyper_v' 13:01:22 helloo 13:01:58 hi 13:02:16 Hi 13:02:26 helloo. :) 13:02:37 bit late today 13:02:46 have we started 13:03:02 yeah, the meeting started. was waiting for more people to join 13:03:14 any idea if anyone else is joining? 13:03:51 sonu is joining 13:04:04 he mentioned that today 13:04:07 we can start 13:04:27 cool 13:04:36 #topic monasca status 13:05:04 soo, tomorrow I have a skype meeting set up with Roland regading it 13:05:23 nice 13:05:25 so, yeah, waiting for tomorrow 13:05:31 what time is it ? 13:05:58 it is sometime in the evening 13:06:17 your evening ? 13:06:30 yeah. 10 pm my local time. 13:06:47 ok.. its too late for me 13:06:57 i may not be able to join 13:06:59 so, they've been setting up a hyper-v node for testing 13:07:05 its past mid niht for me 13:07:11 ok 13:07:16 they want to see with their own eyes that it works 13:07:22 ok 13:07:31 i saw your mail with the instructions 13:07:35 yeah. 13:07:50 did not see them reply back that they are tryin the install 13:07:50 other than that, just minor comments and reviews on the patches. 13:08:07 #topic Hyper-V OVS vif plug driver 13:08:15 just a minor update 13:08:20 the os-vif patch merged. 13:08:38 the nova patch ... correct ? 13:08:54 #link os-vif patch: Adds Windows support for OvsPlugin: https://review.openstack.org/#/c/344433/ 13:09:04 just the os-vif patch. 13:09:20 the nova patch was frozen before the os-vif patch merged. 13:09:29 ok 13:09:44 #topic Windows Containers status 13:09:51 atuvenie: helloo. :) 13:10:24 hello :) 13:10:46 so, the folks over at aprenda have a working branch, they plan to make a pull request 13:11:10 I have to look over that and see how to integrate their work with our plans for using OVS 13:11:52 that's about it 13:12:27 yeay, thanks for the news. :) 13:13:50 are containers stable in windows ? 13:13:59 never tried that .... yet 13:14:40 atuvenie: ^ 13:15:16 sagar_nikam: well, for my purposes sure, I have not seen major changes between TP4 and TP5 13:15:42 atuvenie: thanks 13:16:04 sagar_nikam: I do know there are plans to change the powershell cmdlets for working with containers to align with docker 13:16:24 ok 13:17:02 moving on 13:17:03 once you have it all working... let us know... we will try and let you know how it all works 13:17:35 #topic Newton release status 13:17:48 soo, still testing. looks fine so far. 13:18:12 ok 13:18:47 we've added a checkbox in our Openstack Hyper-V Compute driver installer for enabling / disabling neutron security groups. 13:19:26 why ? dont we want SGs by default ? 13:20:04 yes, by default, but it can be turned off, if someone doesn't want security groups anyways 13:20:22 ok 13:20:38 we're also doing some performance tests and improvements. for the moment, there are ~10% smaller spawn times. 13:20:52 still in progress 13:21:40 ok 13:21:43 #topic open discussion 13:22:05 i was speaking to Sonu today. His team is planning for scale tests on Mitaka sometime 13:22:05 that's about it from my part. 13:22:26 sagar_nikam: any estimates on when that's going to happen? 13:22:33 we have done scale tests on liberty and found 2x scale improvement 13:22:44 as compared to Kilo 13:23:01 cool. :) 13:23:01 now we want to try Mitaka and compare with liberty 13:23:12 in Kilo we wen tto 1000 VMs 13:23:21 in liberty we went to 2000 VMs 13:23:21 yeah, mitaka should be even better, from what we've tested in the last cycle. 13:23:41 if we had more hardware, i am pretty sure .. we would have gone further 13:23:42 i forgot what the improvement rate was from liberty to mitaka 13:24:18 claudiub|2: not sure.. depends on the slot which we get for scale hardware 13:24:42 liberty to mitaka... not yet done 13:24:49 that is the next plan 13:24:58 as and wehn we get a slot in the scale lab 13:25:10 but is planned 13:25:15 sagar_nikam: i see. well, let us know how it goes. :) 13:25:25 yes... we will let you know 13:25:34 and see if we can improve further 13:25:43 this time i am trying to get more hardware 13:25:53 so as to reach 4000 to 5000 VMs 13:26:05 not sure when it will happen 13:26:06 ooh, nice. :D 13:26:28 all depends on the scale lab agreeing to provide more hardware for hyperv 13:26:47 let us see how it goes... will keep you informed 13:27:06 nice. :) any other news from your side? 13:27:14 i had a question 13:27:23 sure 13:27:42 i remember you or somebody from cloudbase had submitted a BP and patches for 13:27:49 GPU support 13:27:53 ah yes 13:27:55 to instances 13:27:56 RemoteFX 13:28:02 what happened to that 13:28:03 yep, that merged in Newton. :) 13:28:06 fully 13:28:08 yes RemoteFX 13:28:17 wow !!! cool 13:28:30 and it has been available in compute-hyperv since Kilo, i think 13:28:34 did you actually test with hardware which had GPU ? 13:28:47 yep, that's how we tested it 13:28:51 nvidia 13:28:54 ok 13:29:06 yeah, not sure the exact model of the card 13:29:08 but worked fine 13:29:14 I've tested it for 2016 as well 13:29:15 i had a similar BP for VMWare some 3 releases back 13:29:34 and I even got to install steam + video games in a VM. :D 13:29:35 submitted multiple patches... never got merged as my code was not getting reviewed 13:29:54 I got to have fun with that blueprint. :D 13:30:03 i see 13:30:26 claudiub|2: you have some write up or some docs on how to use nova hyperv driver for RemoteFX ? 13:30:31 #link Adds RemoteFX support to the Hyper-V Driver: https://review.openstack.org/#/c/42529/ 13:30:46 I have a draft article about it 13:30:54 it should be published some time soon. 13:31:22 ok 13:31:23 at the very least, you can read the nova release notes regarding RemoteFX: https://review.openstack.org/#/c/42529/26/releasenotes/notes/bp-hyper-v-remotefx-1474ef1a082ad1b0.yaml 13:31:45 you'll have to have enable_remotefx = True in the nova.conf file though 13:32:01 you can read about it the release notes 13:32:01 ok 13:32:09 this is in newton ? 13:32:12 yes 13:32:13 or mitaka 13:32:15 ok 13:32:27 i will have to wait till newton to use it 13:32:27 but then again, the feature exists in compute-hyperv since kilo. :) 13:32:37 anyways, I've only tested the feature with windows 10 VMs 13:33:01 ok 13:33:09 i guess it should work with windows 7 as well 13:33:22 and I noticed that the guest OS must have support for RemoteFX, otherwise, its not useful to add a RemoteFX vGPU to the instance 13:33:52 ok 13:34:07 can you share the doc when it is ready 13:34:20 we will try it as soon we move to newton 13:34:21 for example, windows 10 Pro didn't support RemoteFX, windows 10 Enterprise supports RemoteFX 13:34:41 also are the patches easily backportable to mitaka 13:35:04 i will try backport in our downstream repo 13:35:39 hm, I'd say for the most part, yes. 13:36:05 but there might be a few merge conflicts, especially in the tests 13:36:18 ok 13:37:07 anything else? 13:37:24 not anything as of now 13:37:42 we will soon start moving fully to mitaka 13:37:51 and start the tests all over again 13:37:57 will keep you posted 13:38:04 sounds good. :) 13:38:16 i am done 13:38:25 also, let us know if you're going to test remote fx as well 13:38:42 remote fx ... let me try sometime 13:38:46 not so soon 13:38:52 ok 13:39:02 first we will try to check all other functionality in mitaka 13:39:06 and the scale tests 13:39:17 later priority will be remotefx 13:39:31 i see 13:39:48 well then, we can end this meeting early. :) 13:39:57 yes... 13:39:58 hopefully by next week we will have some good news about monasca. :) 13:40:07 yes i hope so 13:40:11 thanks all for joining, see you next week! 13:40:17 keep me posted on how the reviews go 13:40:20 on skype 13:40:28 will do. :) 13:40:33 i will try to get any help that you need 13:40:49 even i want monasca hyperv to be merged at the earliest 13:40:49 thanks. :) 13:41:15 #endmeeting