15:03:53 #startmeeting XenAPI 15:03:54 Meeting started Wed Oct 15 15:03:53 2014 UTC and is due to finish in 60 minutes. The chair is johnthetubaguy. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:03:55 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:03:58 The meeting name has been set to 'xenapi' 15:03:59 hello hello 15:04:04 hows is everyone? 15:04:11 #topic XenServer CI 15:04:20 so I guess we start with a CI update 15:05:05 BobBall: hows things going with CI? 15:05:28 I've just asked Mate to join too 15:05:38 but CI is going well ATM 15:05:43 failure rate is way down 15:05:55 awesome 15:06:00 http://www.rcbops.com/gerrit/reports/nova-cireport.html seems higly broken though :) 15:06:05 well the run rate is probably lower too I guess 15:06:14 Hello 15:06:36 Failure rate percentage wise 15:06:39 it's very small now 15:06:49 I suspect due to the coalescing fix 15:07:00 {'': 37, 'Failed': 15, None: 15, 'Passed': 238, 'Aborted: No result found': 2} 15:07:08 ah, thats possible 15:07:11 Failure rate used to be around 15% 15:07:19 but that's what... 6-7%? 15:07:22 at a rough guess 15:07:30 BobBall: does it include the failures due to invalid code? 15:07:36 yes matel 15:07:37 invalid code? 15:07:40 bad patches 15:07:48 ah, right 15:07:50 gotcha 15:08:22 so I think all's good on the CI front 15:08:26 Do we know the failure rate for other CI systems? 15:08:30 No 15:08:35 because the nova-cireport is broken :) 15:09:54 But anyway, we're not seeing many failures now 15:10:02 and that's all I care about :) 15:10:06 (in that little world) 15:11:04 ok 15:11:15 I was playing with neutron the last two days. 15:11:23 ah, interensting 15:11:32 is that for moving the CI to neutron? 15:11:50 We'll see, first we want to see in what state we are in. 15:11:57 OK, gotcha 15:12:05 just thinking about CI stuff 15:12:15 I'm logging my work here: https://github.com/citrix-openstack/blogentries/blob/master/neutron_on_xenserver.md 15:12:18 are we just seeing errors that are related to general issues? 15:12:25 We're making our internal CI on neutron work 15:12:31 cool 15:12:53 At the moment I got to the point where I can deploy something that's using neutron. 15:13:00 ====== 15:13:00 Totals 15:13:00 ====== 15:13:00 Run: 614 in 1484.443647 sec. 15:13:00 - Passed: 500 15:13:01 - Skipped: 55 15:13:03 - Failed: 59 15:13:35 Excellent 15:13:36 So I guess a lot of work needs to be done. 15:13:52 I'm assuming that the 500 jobs are the ones that don't use VMs ;) 15:13:53 yeah, probably not starting instances 15:13:54 I will propose several devstack changes. 15:14:15 OK, makes sense 15:14:17 I haven't looked at the particular test cases yet. 15:14:38 first things first: 1.) Be able to deploy it 15:15:00 That's really all from me. 15:15:11 Tis good tis good 15:15:33 are you repeating are you repeating yourself yourself? 15:15:52 Perhaps... perhaps... 15:15:57 Anyway - move on? 15:16:01 sure 15:16:05 (oh, we already have! not talking about the CI now!) 15:16:50 So I'd like to refresh the list of XenServer modifications OpenStack would find useful. 15:17:07 Do either of you have suggestions for that? 15:17:47 Do we have a list to start with? 15:18:18 I think so - but it's an old + internal list 15:18:38 BobBall: should we put some ideas on OS wiki? 15:18:39 https://info.citrite.net/display/Partner/OpenStack+-+Required+PR+list 15:19:07 erm, external ones maybe? 15:19:24 yeah, I gues we all said that 15:19:34 I'm trying to form an internal request list - certainly some of those aren't things that should be externally listed 15:19:55 But what thoughts do you have? I can make sure they are on the internal list 15:20:29 matel: that list does need to be updated, I know. 15:20:43 hmm, OK 15:20:48 so requests for XenServer features? 15:20:52 BobBall: I am looking at the list, but at the moment can't even comment on that. 15:20:56 features or modifications yes 15:21:06 import / export is the biggest pain 15:21:25 live-migrate and hooks for VIF stuff keeps coming up 15:21:30 but thats the main bits 15:21:49 python2.7 in Dom0 wouldn't hurt 15:22:02 usual things really 15:22:12 BobBall: while I'm playing with neutron - a more recent version of OVS? 15:22:21 Well we've upgraded python in Creedence 15:22:30 hooks for VIF scripts? 15:22:36 OVS is also upgrading in Creedence 15:22:38 whats creedence? 15:22:43 XenServer 6.5 15:23:06 ah, I thought that was happening last I heard, cools 15:23:45 oh 15:23:46 hang on 15:23:49 I thought we did 15:23:55 I'll have to check on that 15:24:03 latest build I've just checked has 2.4.3... 15:24:40 hmm 15:24:49 2.7.x is my hope, one day 15:25:01 along with CentOS 6.x or ideally 7.x 15:25:14 We'll cross our fingers for that john :) 15:25:18 OVS is definitely upgraded though matel 15:25:19 indeed 15:25:33 upgraded to 2.1.3 15:25:40 looking at: https://github.com/xapi-project/sm/blob/xs64bit/tests/install_prerequisites_for_python_unittests.sh - it seems to use 2.4 15:25:42 ah, thats quite new, cool 15:26:05 Yes matel - I was mistaken 15:26:33 anyways, import / export is probably the big one for reliability 15:26:46 just thinking of what plugins and hacks we could remove 15:27:01 Indeed. I'm working on import/export 15:27:04 (or at least what we could simplify, then properly unit tests) 15:27:10 (working as in trying to persuade - not coding yet) 15:27:28 BobBall: OK, we are looking at possibly doing partial snapshots and stuff soon 15:27:37 so the requirements are probably growing a little 15:27:38 Very interesting. 15:27:43 Yes - let's talk about that 15:28:11 I want to split our base images into two, then have customer images on top of that, and store the differences in glance, with links to the parent 15:28:13 basically 15:28:35 I need to look at the stats to see how much it will help, but thats pencilled in 15:28:47 *nod* 15:28:52 Sounds like a plan. 15:29:54 yeah, idea being to speed up builds of snapshots 15:30:09 i mean it will probably slow down the XenServer CI, but hey ho 15:30:20 why would it slow it down? 15:30:44 well your image contains very few blocks that were in the original base image 15:31:17 well, maybe its got some, but you know, its anoyingly different 15:31:18 anyways 15:31:28 #topic Open Discussion 15:31:36 I clearly didn't change the topic 15:31:41 any more for any more? 15:32:10 not from me. 15:32:24 nah 15:32:28 cool 15:32:37 a 30 mins back 15:32:40 thanks all 15:32:43 #endmeeting