17:00:12 <johnthetubaguy> #startmeeting XenAPI 17:00:13 <openstack> Meeting started Wed Mar 27 17:00:12 2013 UTC. The chair is johnthetubaguy. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:14 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:16 <openstack> The meeting name has been set to 'xenapi' 17:00:32 <johnthetubaguy> hi, who is around for this weeks meeting? 17:00:40 <matelakat> me 17:00:44 * BobBall waves his hand 17:01:21 <johnthetubaguy> OK, cool 17:01:27 <johnthetubaguy> lets get cracking... 17:01:40 <BobBall> I was just saying to Mate, I might have to leave at 5 minutes past 5... 17:01:44 <johnthetubaguy> #topic Actions from previous meeting 17:01:51 <BobBall> But Mate thought that I should stay a little longer, so I will. 17:02:01 <johnthetubaguy> nothing to say 17:02:06 <BobBall> haha 17:02:07 <matelakat> No. 17:02:19 <johnthetubaguy> #topic blueprints 17:02:29 <johnthetubaguy> anything from anyone here? 17:02:38 <BobBall> well before we move on 17:02:41 <BobBall> there was an action on me 17:02:44 <johnthetubaguy> we all prep-ed for the summit? 17:02:46 <BobBall> that didn't get copied over from the week before 17:02:52 <BobBall> in terms of updating on the iSCSI timeout 17:02:55 * BobBall only just remembered 17:02:56 <johnthetubaguy> yes, true 17:03:04 <BobBall> what was the question again? Do you remember? or shall I go back through the logs? 17:03:14 <johnthetubaguy> you were working on a fix I think 17:03:18 <johnthetubaguy> to control the timeout 17:03:21 <matelakat> BobBall to update on iSCSI timeout when there is an update 17:03:35 <matelakat> #link http://eavesdrop.openstack.org/meetings/xenapi/2013/xenapi.2013-03-13-17.02.log.html 17:03:40 <BobBall> there were several iscsi timeout issues I've looked at over the last couple of weeks 17:03:57 <johnthetubaguy> it was the iscsi goes down, then it takes ages to timeout 17:03:58 <BobBall> oh yes 17:04:02 <johnthetubaguy> include the SR scan 17:04:03 <BobBall> the attach issue when the conn has gone away 17:04:06 <BobBall> we think so 17:04:18 <BobBall> oh hang on - that one was a different one (!) 17:04:59 <BobBall> Okay - the one where the attach was expecting a path to exist has been fixed - but that was a RS specific issue AFAIK 17:05:01 <johnthetubaguy> takes a while to detect the problem, there was a short term fix that uses the SR scan to detect failures 17:05:31 <BobBall> yes 17:05:37 <BobBall> that bug hasn't been progressed I'm afraid 17:05:44 <johnthetubaguy> we ideally need some way to specify a timeout 17:05:49 <johnthetubaguy> OK 17:05:50 <BobBall> (there is a separate OS bug raised talking about the 120s timeout) 17:06:01 <BobBall> although I'm struggilng to find that atm! 17:06:08 <johnthetubaguy> ah, you got that to hand? 17:06:46 <johnthetubaguy> #link https://bugs.launchpad.net/nova/+bug/1152401 17:06:49 <uvirtbot> Launchpad bug 1152401 in nova "xenapi: Detecting bad-volumes relies on 120 sec timeout" [High,Triaged] 17:07:01 <BobBall> ... I searched for xenapi 120 - how did that not come up?! 17:07:11 <johnthetubaguy> duno 17:07:18 <johnthetubaguy> shall I assign that to you? 17:07:23 <BobBall> odd. Anyway, that's the one. 17:07:27 <BobBall> Please 17:07:46 <johnthetubaguy> OK, should be yours now 17:07:52 <johnthetubaguy> action sorted, tracked by a bug 17:07:56 <BobBall> Joy. 17:08:14 <johnthetubaguy> OK, so anything blueprinty? 17:08:26 <matelakat> no 17:08:28 <BobBall> I had a review of the etherpad and I'm happy 17:08:32 <johnthetubaguy> cool 17:08:39 <BobBall> although since we're talking about it, do we have a time for the session? 17:08:40 <johnthetubaguy> #topic docs 17:09:04 <BobBall> I'd like to book it in ASAP and see if James B / Guy B want to come along to sit in on the session 17:09:12 <johnthetubaguy> BobBall: its not even approved yet, I think that happens much nearer the time 17:09:17 <BobBall> bah 17:09:19 <BobBall> okay 17:09:24 <matelakat> Regarding to the docs, Release notes. 17:09:34 <johnthetubaguy> OK 17:09:45 <matelakat> On the meeting Wiki Page, I included the stuff, that I want to flush to the release notes. 17:09:56 <matelakat> Could you please add any missed stuff? 17:10:00 <johnthetubaguy> ah, got ya 17:10:22 <matelakat> So far we have ConfigDrive, and XenAPINFS 17:10:45 <matelakat> THe other question. 17:10:58 <johnthetubaguy> I would have to read through all the blueprints on launchpad to check 17:11:07 <johnthetubaguy> is anyone fancying that? 17:11:26 <johnthetubaguy> I probably should take a look anyways... 17:11:39 <johnthetubaguy> #action johnthetubaguy to review blueprints for release notes 17:11:53 <johnthetubaguy> any more for docs? 17:11:57 <matelakat> I wanted to put only the things implemented by us / being involved in. 17:12:00 <matelakat> yes. 17:12:02 <matelakat> We are missing a step in the documentation: /boot/guest creation 17:12:15 <johnthetubaguy> I meant review for xenapi related things 17:12:31 <johnthetubaguy> hmm, yes, good point 17:12:38 <johnthetubaguy> for three part images 17:12:41 <matelakat> I haven't pushed changes to the docs this week. 17:12:57 <matelakat> Y, so the q is: should it be a symlink to a subdir within the Local SR? 17:13:21 <BobBall> _yes_ 17:13:23 <matelakat> devstack is just creating a directory: /boot/guest 17:13:38 <matelakat> Okay, so that's a devstack bug, and a documentation Todo. 17:13:41 <BobBall> I'm less fussed about devstack since the local directory could be used for lots 17:14:06 <BobBall> but any directory that could contain any non-trivial amount of data should be on the SR rather than in dom0 17:14:09 <matelakat> #action matelakat to fix /boot/guest in devstack, and document it. 17:14:13 <johnthetubaguy> yes, its not a very big deal, becuase the files there are quite small 17:14:22 <BobBall> /boot/guest coul contain a whole bunch of kernels, right? 17:14:23 <BobBall> +d 17:14:36 <matelakat> I remember, it caused pains for us for broken builds. 17:14:39 <johnthetubaguy> it could, yes 17:14:47 <johnthetubaguy> devstack should represent best practice 17:14:59 <johnthetubaguy> so I think it would be good to fix that, and make it match /images 17:15:10 <matelakat> Thanks. 17:15:11 <BobBall> OK - then yes, create a link. Or better yet, move it to a configurable SR 17:15:30 <matelakat> small steps are welcome. 17:15:36 <BobBall> hard coded paths like /boot/guest and /images are bad :) 17:15:37 <johnthetubaguy> of course, if you don't do three part images, then there is no need, but hey 17:15:37 <matelakat> So first, it will be a symlink. 17:15:51 <BobBall> fair enough 17:15:52 <johnthetubaguy> so, it turns out the kernel *must* be in /boot/guest 17:16:09 <BobBall> oh really? Why's that? 17:16:09 <johnthetubaguy> but clearly we could sym link somewhere sensible 17:16:20 <johnthetubaguy> its a xapi restriction on the kernel boot param, I think 17:16:20 <matelakat> I will try the symlink approach 17:16:29 <BobBall> oh right. Drat. 17:16:43 <johnthetubaguy> worth a double check though 17:17:09 <johnthetubaguy> I presume xen will follow the symlink correctly… hmm worth a test 17:17:41 <BobBall> okay - we'll look into that 17:17:43 <matelakat> You could leave that with me. 17:17:45 <BobBall> by we, I mean mate. 17:17:49 <johnthetubaguy> hehe, cool 17:17:50 <matelakat> ta 17:17:57 <BobBall> Welcome. 17:18:08 <johnthetubaguy> so I guess we our out of docs stuff 17:18:14 <matelakat> I believe. 17:18:16 <johnthetubaguy> #topic bugs 17:18:38 <BobBall> Can I talk about a bug? 17:18:45 <BobBall> #link https://bugs.launchpad.net/nova/+bug/1160323 17:18:46 <johnthetubaguy> OK, so I have been doing the odd bit of triage on the xenapi bugs 17:18:47 <uvirtbot> Launchpad bug 1160323 in nova "Cannot block migrate in xenapi with iSCSI cinder volumes attached" [Medium,In progress] 17:18:49 <johnthetubaguy> OK, sure 17:18:53 <BobBall> and particularly the review at: 17:19:01 <BobBall> #link https://review.openstack.org/25419 17:19:21 <matelakat> I was afraid, Bob will come up with the uuid one :-) 17:19:21 <johnthetubaguy> ah, yes, its on my todo list for this afternoon 17:19:30 <BobBall> *grin* 17:19:37 <BobBall> ahhhh good - if it's on the list, then that's fine, I'm happy 17:19:50 <BobBall> and the UUID one finally got in after a marathon number of comments 17:20:10 <johnthetubaguy> yes, its quite frequently like that 17:20:11 <BobBall> _AND_ the commit beat John's commit which probably conflicts in the test code, so I'm happy :) 17:20:26 <johnthetubaguy> hehe, probably 17:20:50 <johnthetubaguy> any other bugs? 17:20:58 <matelakat> I guess John does frequent rebasing, so he won't be hit by Bob's change... 17:21:07 <BobBall> oh yes, of course ;) 17:21:17 <johnthetubaguy> nah, don't both we that kind of thing 17:21:42 <matelakat> I have one. 17:21:47 <matelakat> It is not really a bug. 17:21:51 <johnthetubaguy> BobBall: it should prob have docimpact tag on there 17:21:58 <matelakat> I have it on the Other section. 17:22:02 <johnthetubaguy> OK, fire away 17:22:06 <johnthetubaguy> OK... 17:22:15 <johnthetubaguy> I see them 17:22:15 <BobBall> docimpact tag on the bug or the commit? 17:22:15 <matelakat> https://bugs.launchpad.net/nova/+bug/903445 17:22:17 <uvirtbot> Launchpad bug 903445 in nova "New xapi backend for Local Storage ISO SR" [Wishlist,Confirmed] 17:22:21 <matelakat> #link https://bugs.launchpad.net/nova/+bug/903445 17:22:31 <johnthetubaguy> #topic OpenDiscussion 17:22:40 <matelakat> I mean how we are doing with ISO files in general? 17:22:54 <johnthetubaguy> ah, very timely 17:23:01 <johnthetubaguy> I have just been fixing that up 17:23:17 <johnthetubaguy> https://review.openstack.org/#/c/24895/ 17:23:38 <johnthetubaguy> https://wiki.openstack.org/wiki/XenServer/BootFromISO 17:23:51 <BobBall> odd - wonder why I wasn't notified of that review... 17:24:02 <BobBall> Typo in the commit message btw :) 17:24:22 <matelakat> John, is that something, that should go to the docs? 17:24:44 <johnthetubaguy> yes, if it works and we are happy with it 17:24:55 <johnthetubaguy> done some quick testing myself 17:25:03 <matelakat> #action matelakat to add https://wiki.openstack.org/wiki/XenServer/BootFromISO to the install docs 17:25:13 <johnthetubaguy> OK, thanks 17:25:31 <matelakat> Are there any XS version restrictions on this tweak? 17:25:38 <johnthetubaguy> turns out the unit tests for spawn were quite broken, so didn't test the ISO code path 17:25:49 <johnthetubaguy> matelakat: not that I know of 17:27:14 <matelakat> Okay 17:27:18 <johnthetubaguy> so next thing 17:27:23 <matelakat> Other question is around Console. 17:27:26 <johnthetubaguy> ok 17:27:34 <matelakat> #link https://bugs.launchpad.net/nova/+bug/1004175 17:27:35 <uvirtbot> Launchpad bug 1004175 in nova "XenAPI text console support" [Wishlist,In progress] 17:27:37 <johnthetubaguy> text console 17:27:40 <matelakat> Seems to be orphaned. 17:27:59 <matelakat> No activity since december. 17:28:13 <johnthetubaguy> yes, its worth someone adopting that one I think 17:28:16 <matelakat> So it won't hit G 17:28:35 <matelakat> John, do you think, that it could hit G? 17:28:56 <BobBall> Too late for G isn't it? 17:29:00 <johnthetubaguy> nothing will make G anymore 17:29:06 <johnthetubaguy> only critical fixes right? 17:29:09 <matelakat> Okaz, my bad. 17:29:14 <matelakat> Okay, my bad 17:29:19 <johnthetubaguy> no worries 17:29:25 <matelakat> I am just overusing my time-machine. 17:29:48 <johnthetubaguy> this one is really a feature, needs the blueprint bringing back to life 17:29:56 <BobBall> Looks like quite a simple change I guess 17:30:10 <BobBall> although testing it is more trick 17:30:11 <BobBall> +y 17:30:13 <matelakat> I would say, let's pick up that change. 17:30:17 <johnthetubaguy> its a shame, XenServer added a feature for the serial console for OpenStack, bit it doesn't really work with what OpenStack needs. 17:30:19 <matelakat> Tempest has tests for it. 17:30:24 <johnthetubaguy> its worth a whirl 17:30:34 <BobBall> ahhh I see 17:30:41 <matelakat> Who wants to pick it up? 17:31:03 <BobBall> Well we can't schedule when it'll be done, but we can add it to the backlog for things to pick up. There are other higher priority things to pop off the top first 17:31:13 <johnthetubaguy> same here really 17:31:29 <johnthetubaguy> stuck in a pit of resize issues at the moment 17:31:31 <matelakat> Not good to hear these things... :-( 17:31:32 <BobBall> Particularly since it's missed G 17:31:55 <BobBall> Agreed... 17:32:02 <matelakat> I have another thing. 17:32:10 <johnthetubaguy> OK 17:32:15 <matelakat> I was looking at the plugin install in devstack 17:32:29 <matelakat> And I realised, that Quantum has changed... 17:32:34 <matelakat> https://github.com/openstack/quantum/blob/master/quantum/plugins/openvswitch/agent/xenserver_install.sh 17:32:39 <matelakat> #link https://github.com/openstack/quantum/blob/master/quantum/plugins/openvswitch/agent/xenserver_install.sh 17:32:59 <matelakat> So it seems, it no longer has the traditional xapi-plugin layout. 17:33:15 <johnthetubaguy> not sure what you mean 17:33:24 <matelakat> a sec 17:33:29 <johnthetubaguy> the quantum support hasn't hit trunk yet right? 17:33:44 <johnthetubaguy> there are no plugins in quantum yet, really 17:33:49 <johnthetubaguy> not for xenserver anyways 17:34:04 <johnthetubaguy> lol salvatore left quickly 17:34:13 <matelakat> So your saying, I cannot see that, cos it was never was there... 17:34:19 <johnthetubaguy> yep 17:34:26 <matelakat> How stupid I am. 17:34:40 <johnthetubaguy> https://review.openstack.org/#/c/15022/ 17:34:44 <BobBall> He's in and out - guess he has a network issue 17:34:53 <matelakat> :-) 17:35:22 <BobBall> Mate laughed. I appreciate the symapthy laugh. Thanks, 17:35:53 <johnthetubaguy> hehe 17:35:59 <johnthetubaguy> so what else we got on the list for today? 17:36:02 <BobBall> I see - that change linked deletes the script Mate pointed at. 17:36:07 <matelakat> Anyhow, I think I overlooked that Quantum stuff 17:36:12 <matelakat> I have one thing to note. 17:36:19 <johnthetubaguy> kernel lock up? 17:36:22 <matelakat> The good old kernel locking issue. 17:36:24 <matelakat> yes. 17:36:48 <matelakat> So if you don't use the loopback device, the pain goes away. 17:36:56 <johnthetubaguy> so you add an extra disk intead? 17:37:04 <johnthetubaguy> instead? 17:37:06 <matelakat> That's what I do in the CI. 17:37:18 <johnthetubaguy> hmm, OK 17:37:22 <matelakat> And the Jenkins weather became better. 17:37:42 <johnthetubaguy> cool, not sure if you are just changing the race condition, but if it works, thats all good 17:37:48 <matelakat> I had tempest running in an infinite loop for a day, and no nasty kernel messages. 17:38:00 <johnthetubaguy> sounds good enough to me 17:38:08 <johnthetubaguy> cool 17:38:21 <matelakat> That's all from me. 17:38:29 <johnthetubaguy> cool, thanks, some good updates 17:38:46 <matelakat> blogpost 17:38:47 <BobBall> Saw the CentOS update from yesterday too - which is looking positive 17:38:54 <matelakat> #link http://blogs.citrix.com/2013/03/18/virtual-hypervisor/ 17:38:56 <johnthetubaguy> virtual hypervisor post looks interesting too, 17:39:05 <johnthetubaguy> yes, took a quick look 17:39:20 <matelakat> Just some shell scripts, but might be handy. 17:39:31 <johnthetubaguy> its possibly a bit deep, but its good to advertise these things :-) 17:39:47 <johnthetubaguy> I install XCP using openstack to test my patch :-) 17:39:49 <matelakat> I am always too technical. 17:39:53 <johnthetubaguy> doing an iso boot 17:40:05 <matelakat> cool! 17:40:17 <johnthetubaguy> centos stuff is slowly moving forward 17:40:27 <johnthetubaguy> taking a break for the mo while stuff gets fixed 17:40:34 <johnthetubaguy> OK, I think we are all done? 17:40:43 <matelakat> y, John, thanks. 17:40:53 <BobBall> Think so 17:40:56 <johnthetubaguy> cool, thank you all 17:41:00 <johnthetubaguy> #endmeeting