13:01:31 #startmeeting hyper-v 13:01:32 Meeting started Wed May 4 13:01:31 2016 UTC and is due to finish in 60 minutes. The chair is lpetrut. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:01:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:01:36 The meeting name has been set to 'hyper_v' 13:02:54 #topic Openstack summit aftermath 13:03:34 so, we had some really productive meetings with the freezer, monasca, magnum and designate guys 13:03:54 lpetrut: that's nice 13:04:12 we held the winstackers session as well, but unfortunately none of your guys made it to the session 13:04:31 here's a link to the etherpad: https://etherpad.openstack.org/p/newton-winstackers-design-session 13:05:07 the topics mentioned here are roughly pointing out what we are planning to do for Newton 13:05:46 #topic Shielded VMs 13:06:23 lpetrut: had a question on summit meetings 13:06:30 sure 13:07:01 from the etherpad, you gave, we will have lot of nice support in Newton 13:07:28 what are the plans for monasca , freezer and designate 13:07:36 hi all 13:07:39 i see only magnum in the etherpad 13:07:41 Hi domi 13:07:49 #topic freezer 13:08:14 so, we talked with the guys, and first, we are planning to refactor the existing Windows related bits 13:08:34 ok, to use pyMI ? 13:08:34 this will go in os-win 13:08:43 ok 13:09:15 actually, they support older Windows versions such as Windows 7, Windows Server 2008, etc. and we'll have to use the old wmi lib in that case 13:09:43 this also means that we won't be able to drop the pywin32 lib in this case 13:09:47 *dependency 13:09:47 as part of newton, are we planning feature parity with linux in freezer or only important use cases ? 13:11:04 so, they were already leveraging VSS, we are planning to improve this, and use the Hyper-V VSS provider as well 13:11:41 we may also provide an MSI for freezer 13:11:53 ok 13:12:22 sounds cool 13:13:00 lpetrut: let me know if your team needs any info from freezer core team, i can try to reach them 13:13:09 or for reviews as well 13:13:18 sagar_nikam: sure 13:13:37 these changes are planned in newton ? 13:14:25 yep, I think we'll get everything in during Newton. At the moment, freezer is going through a refactoring, but we'll get the specs and os-win side code in ASAP 13:14:59 ok 13:15:18 at some point, we'll need to have a CI as well 13:15:39 ok 13:16:10 ok, next topic 13:16:20 monasca ? 13:16:26 #topic monasca 13:17:30 ensuring Windows support for Monasca seems pretty straight forward, leveraging stuff that we already had for Ceilometer 13:18:04 ok, thats good 13:18:15 there was one issue we had seen 13:18:27 cbelu sent a patch updating the existing Monasca windows agent to use os-win: https://review.openstack.org/#/c/311925/ 13:18:53 the existing agent does not start on windows 13:19:02 i mean monasca agent 13:20:17 the patch above is in os-win, can you post the monasca patch as well 13:20:26 not sure about the status of this agent, but I think that the patch Claudiu submitted may fix this issue 13:20:39 ok 13:22:27 #topic magnum 13:23:02 Claudiu Belu has spoke with dims, and it seems like we should make sure that kubernetes or docker swarm work on Windows 13:23:07 monasca support also planned in newton > 13:23:14 yep 13:23:18 ok 13:23:56 WIN 2016 has support for docker 13:24:06 atleast TP4 13:24:08 had it 13:24:12 yep 13:24:33 looks promisimg 13:24:35 is that sufficient or we also need kubernetes ? 13:24:40 for magnum support 13:26:22 we'll also need kubernetes 13:27:38 lpetrut: both docker-swarm and kubernetes needed for it to make to magnum ? 13:28:07 I think one of them is enough for now 13:28:38 ok 13:29:41 support for windows in magnum depends on WIN 2016 release date ? 13:30:11 if WIN 2016 does not get released by newton, what is our plan ? 13:30:38 I think we can have the code in regardless of the WS 2016 release date 13:30:50 ok cool 13:33:04 do we have any other topics? 13:33:08 #topic designate 13:34:00 so, adding Windows support over here seems to be pretty straight forward, as Designate will do zone transfers to manage records 13:34:45 i think designate already has some windows support, not sure how much though 13:36:47 not really sure about the current status, but I'm sure we'll focus on it during Newton 13:36:58 so, going back to shielded vms 13:37:03 #topic Shielded VMs 13:37:16 as TP5 is out, Iulia managed to push the updated code 13:37:28 https://review.openstack.org/#/c/283564/ - os-win 13:37:28 https://review.openstack.org/#/c/283567/ - compute-hyperv 13:37:44 ok 13:37:58 there will also be a tool, easing the way in which pdk files are generated 13:37:59 plan to upstream it ? in newton ? 13:38:22 yep, hopefully 13:38:35 ok 13:39:03 any other topics you guys would like to discuss? 13:39:52 yes, I tried to ceate a haproxy + freerdp setup 13:40:02 in newton, we will support cluster driver, FC, monasca, freezer, designate and magnum 13:40:09 lots of new features 13:40:15 nice progress 13:40:30 and freerdp dies because of haproxy's monitoring requests 13:40:53 the cluster driver is already supported in mitaka, but only by the compute-hyperv driver (not the in-tree one) 13:40:59 I think I commented on an issue dealing with a similar error 13:41:37 domi007: did you get in touch with our guy working on FreeRDP, Cosmin Muntean? 13:41:43 freerdp - i have observed one issue, sometimes, keystrokes dont work in the console, only clicks work 13:42:17 after restart of freerdp service, keystrokes work again 13:42:20 sagar_nikam: did you click in the grey bar before that? 13:42:50 lpetrut: let me get the github issue and see 13:43:14 domi007: no, i was able to type earlier some commands, and then i find that keystrokes fail 13:43:34 have seen in 2-3 times 13:44:17 by what I heard, there were some race conditions, not sure if those were fixed by now. Anyway, as soon as you find an issue, please file it on ask.cloudbase.it or the github repo and somebody will be on it ASAP 13:44:23 domi007: using ha-proxy, with freerdp- how is your experience ? 13:45:04 lpetrut: is his github name c64cosmin? 13:45:19 he commented on the issue 13:45:48 yep 13:46:01 sagar_nikam: I wanted to test it out but I hit the issue mentioned above 13:46:13 domi007: can you give me the link ? 13:46:16 for the issue 13:46:25 lpetrut: okay, he promised a new version 7 days ago 13:46:31 so I guess he is working on it 13:47:11 sagar_nikam: https://github.com/FreeRDP/FreeRDP-WebConnect/issues/127 13:47:31 got it, I'll try to sync with him. Since we had a lot of talk around FreeRDP during the recent meetings, maybe he'll be able to join as well 13:48:17 great. it seems like a good PoC software but lacks in some robustness 13:48:38 but it's great he is fixing it 13:48:43 lpetrut: how does freerdp scale ? 13:49:20 if we install it one machine with good RAM and CPU, how many RDP consoles can it support ? 13:49:34 i mean RDP consoles from horizon UI 13:50:34 not sure ATM, but I guess it would make sense to do some stress testing and see what can be improved 13:51:32 ok 13:51:41 that'd be great :) 13:51:46 domi007: are you planning stress testing of freerdp ? 13:53:04 sagar_nikam: no, we don't really have resources for that. The only stress test I did was trying to use haproxy with it but that killed it fast :) 13:53:17 ok 13:54:06 i am trying to use NLB cluster for FreeRDP 13:54:13 instead of ha-proxy 13:54:53 in other news we are testing ovs 2.5 with Liberty...sadly a Mitaka patch needed to cheery picked into Liberty, now we are testing if it works with that 13:55:01 providing the NLB cluster IP in nova.conf 13:55:17 sagar_nikam: that shouldn't make much difference, what are your results? 13:55:39 just starting, 13:55:53 should have the results in next meeting hopefully 13:56:02 sagar_nikam: Great! 13:56:53 my experience was that it died so fast i wasn't able to see it working at all :) hope it's better for you 13:58:17 standalone freerdp, not usin NLB cluster, worked for me, but i did not try keep the console open for long 13:58:48 i may hit the issue you are mentioning if i keep the console open for long and try some operations 13:59:01 looks like we have only a couple of minutes left, anything else? 13:59:26 I don't have anything :) 13:59:54 ok, thanks guys 13:59:54 nothing much from my side, i hope all the patches/features we are planning in newton gets merged 13:59:56 fast 13:59:58 #endmeeting