15:56:59 #startmeeting hyper-v 15:57:00 Meeting started Tue Feb 12 15:56:59 2013 UTC. The chair is primeministerp. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:57:01 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:57:04 The meeting name has been set to 'hyper_v' 15:57:07 hi everyone 15:57:19 couple of things to discuss today 15:57:42 Hello 15:57:50 iben-mobile: how are things going on your end 15:58:17 hello people 15:58:19 alexpilotti: are you with us 15:58:30 hi there! 15:58:34 pnavarro: hi pedro 15:58:44 ok 15:59:07 alexpilotti: is tavi coming today? 15:59:14 yes 15:59:19 perfecto 15:59:30 shall we begin 15:59:53 yep 15:59:56 alexpilotti has been working quite hard at getting the resize feature done 16:00:19 alexpilotti: i know the first part got a +2 16:00:34 has there been any additional activity? 16:00:38 yes, everything is looking well 16:00:43 features are done 16:00:51 great 16:01:04 There are a lot of conductor related issues 16:01:20 alexpilotti: continue... 16:01:24 so we have to test them as well to help in fixing them 16:01:37 hi all, sorry for being late :) 16:01:41 resize is not tested by tempest 16:01:59 ociuhandu: no worries 16:02:01 so a lot of issues simply got through 16:02:03 alexpilotti: we'll have to write that 16:02:06 i'm assuming 16:02:13 the tempest bits 16:02:13 and they discovered them now that we started testing them :-) 16:02:17 ahh 16:02:19 nice 16:02:25 that's a good thing 16:02:40 things are going good - slow but steady - getting customer resources lined up to expand project internally 16:02:46 when I say missing resize on Tempest, I mean on KVM etc as well 16:02:58 alexpilotti: do you have specific bugs 16:03:07 alexpilotti: ahh 16:03:13 plenty :-) 16:03:16 alexpilotti: so it really wasn't be testing 16:03:22 er tested 16:03:44 https://bugs.launchpad.net/nova/+bug/1120402 16:03:44 iben: great 16:03:46 Launchpad bug 1120402 in nova "ConfigDrive uses metadata service which hits the database from within nova-compute" [High,In progress] 16:03:52 https://bugs.launchpad.net/nova/+bug/1119873 16:03:54 Launchpad bug 1119873 in nova "nova-compute crashes if restarted with an instance in VERIFY_RESIZE state" [High,Triaged] 16:04:02 https://bugs.launchpad.net/nova/+bug/1119817 16:04:03 Launchpad bug 1119817 in nova "Missing exception data for exceptions raised in resize methods" [Undecided,Fix committed] 16:04:05 alexpilotti: thx 16:04:11 https://bugs.launchpad.net/nova/+bug/1117882 16:04:12 Launchpad bug 1117882 in nova "Resize doesn't apply the hypervisor_type filter in choosing the migration target" [High,Triaged] 16:04:22 the last one is a big issue 16:04:39 looking now 16:05:30 it tries to migrate from Hyper-V to KVM :-) 16:05:38 o yes 16:05:41 we discussed 16:05:52 was taking a second to load 16:05:54 even if the image has a hypervisor property set 16:06:07 yeah that's to good 16:06:37 dansmith is working hard to fix them 16:07:04 ok 16:07:28 I did quite a lot of extensive testing on resize, so hopefully we won't have big surprises :-) 16:07:32 I think we'll be in a good position to start adding some value to the testing once we're back from redmond 16:07:43 alexpilotti: great 16:07:52 moving on 16:08:01 to the other issues that need to be addressed 16:08:03 But before the Grizzly release after G3, It'd be great to have some community help in testing all the feature 16:08:24 as always 16:08:31 beers for bugs 16:08:38 hahahah 16:08:43 I'm not putting that in writing 16:08:47 pnavarro: sure, plenty! :-) 16:08:59 ;-) 16:09:05 ... but see me at the summit 16:09:06 (of beers, not of bugs, hopefully) 16:09:25 so 16:09:31 actually 16:09:34 luis_fdez: ping 16:09:40 yep 16:09:45 luis_fdez: any chance you'll be able to check out the resize bits? 16:10:29 as I'm excited after reciving the mail of alexpilotti, I think I'll try it!! hehe 16:10:31 #topic missing features 16:10:36 so moving on 16:11:01 we have some area's that need to be addressed 16:11:09 specifically in terms of resize on boot 16:11:41 luis_fdez: is it important for you guys th eresize on boot? 16:11:43 and • ephemeral storage 16:12:05 alexpilotti: resize on boot is important for everyone i would imagine 16:12:11 I mean currently Hyper-V deploys the glance image w/o resizing it 16:12:22 alexpilotti: storing windows images can become quite horrible 16:12:28 alexpilotti: I think it would be useful as our environment could mix really different flavors 16:12:34 primeministerp: up to a point, private clouds might simply not care as they have tailored images 16:12:44 alexpilotti: nod 16:12:46 luis_fdez: cool 16:13:28 alexpilotti: but generally being able expand to a specific size at boot is useful 16:13:46 BTW the current version of resize expands the VHDs, but not the guest partitions 16:14:01 I'm thinking about adding a simple feature to Cloudbase-init 16:14:08 alexpilotti: I was going to say 16:14:15 that would have to be handled by the booting os 16:14:16 to simply expand the last partition ;-) 16:14:24 as a starting point is ok, from an unattend file is easy to expand the partition to the new size when the vm is deployed 16:14:24 alexpilotti: that might be a great workaround 16:14:27 yep, IMO is a guest os thing 16:14:58 luis_fdez: that would work too 16:14:59 I frankly don't like too much the fact that the Hypervirso fiddles with guest OS partitions 16:15:18 alexpilotti: seperation of church and state is key 16:15:20 ;) 16:15:21 It will be necessary for AKI/AMI/ARI 16:15:34 lol 16:16:07 In Italy we try to repeat it since a few centuries, but it doesn'r really work ;-) 16:16:13 hehe 16:16:16 alexpilotti: +1 16:16:58 so about resize on boot 16:17:16 a big advantage of VHDX over VHD is that it handles resize of differential disks 16:17:41 we could add resize on bot now, before closing Grizzly on VHD but: 16:18:13 if a CoW image comes in, we need to convert it to non CoW before resizing it OR, and here comes the trick: 16:18:37 we copy the base disk for each requested flavor and we resize that one. 16:18:42 before creating the diff disk 16:18:49 so the format will be e.g.: 16:19:06 _.vhd 16:19:20 and we can go on w CoW as usual 16:19:45 the only drawback is that it's not possible to change a flavor's size after the first deployment 16:19:46 and what's the main drawback in that case? 16:19:56 luis_fdez: ^ 16:20:03 hehe, you read my mind! 16:20:19 so if you need to change a flavor, you simply change ID as well :-) 16:20:33 alexpilotti: well strictly from a security perspective it's not a good thing 16:20:53 primeministerp: why? 16:21:25 alexpilotti: sorry 16:21:49 alexpilotti: was talking about the hypervisor controlling guest bits 16:22:04 primeministerp: ok :-) 16:22:05 alexpilotti: i was sidetracked for a moment 16:22:20 primeministerp: "a moment" ? :-D 16:22:28 alexpilotti: people talking to me 16:22:41 distracting my attention 16:22:49 so 16:22:54 back to the missing features 16:23:18 I think they are on the list for H 16:24:02 primeministerp: here's a quick list, Nova related: 16:24:29 - Nova dynamic memory 16:24:30 - Nova ephemeral storage 16:24:31 - Nova progress update 16:24:32 - Nova resize on spawn (currently images are not resized) 16:24:33 - Nova support for AMI / ARI / AKI 16:24:34 - Nova guest FS resize (at least: NTFS, EXT2-4) 16:24:38 -Nova rescue 16:24:49 I don't think i had rescue on my list 16:25:06 perfect thx 16:25:11 primeministerp: yes sorry, it's a new entry :-) 16:25:20 add it yours ;) 16:25:28 yours it more important at this time 16:25:56 it got added today, was going to send it w the rest later 16:26:14 great 16:26:23 so moving on 16:26:24 anyway, the good news is that no big features are missing 16:26:33 alexpilotti: that's what we were shooting for 16:26:37 alexpilotti: great work 16:26:42 merci :-) 16:26:57 resize was the last big thing 16:27:00 ociuhandu: ping 16:27:13 primeministerp: pong 16:27:15 #topic general updates 16:27:26 ociuhandu: got my scripts running out unattend.xml 16:27:34 ociuhandu: debugging puppet install 16:27:35 now 16:27:56 ociuhandu: i'll be back in the office tomorrow 16:28:19 ociuhandu: we can start testing the whole compute node deployment then 16:28:29 primeministerp: great 16:28:40 ociuhandu: I made good progress this am 16:28:54 vishy: hi, any chance you could take a look at: https://review.openstack.org/#/c/21695/ :-) 16:29:19 primeministerp: excellent 16:29:48 anyone have anything else they would like to add 16:30:40 ok 16:30:43 if not I'm closing it 16:30:51 ok! 16:30:51 #endmeeting