17:01:46 #startmeeting third-party 17:01:47 Meeting started Tue Aug 18 17:01:46 2015 UTC and is due to finish in 60 minutes. The chair is krtaylor. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:48 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:01:51 The meeting name has been set to 'third_party' 17:02:03 hey everybody, its that time again 17:02:13 anyone here for CI working group? 17:02:37 o/ 17:02:45 hey krtaylor 17:02:51 hey 17:03:06 hi rfolco mmedvede patrickeast 17:03:17 patrickeast, havent seen you in a while 17:04:19 \o 17:04:21 here's the agenda for today: 17:04:24 hi, I'm half here 17:04:27 #link https://wiki.openstack.org/wiki/Meetings/ThirdParty#8.2F18.2F15_1700_UTC 17:04:38 hi sweston asselin_ 17:04:46 hi krtaylor 17:05:09 any quick announcements? deadlines? 17:05:19 none on the agenda 17:06:32 #topic Common CI 17:06:46 asselin_ I understand if you are too busy to discuss 17:07:06 LOTS of patches 17:07:18 #link https://review.openstack.org/#/q/topic:downstream-puppet,n,z 17:07:35 no updates from me. Still working through patches and reviews. There's progress which is good news. 17:07:50 several look really close to merge 17:09:31 ok, next then 17:09:37 #topic Spec to have infra host scoreboard 17:09:51 this is moving along 17:10:32 although not NEARLY as fast as we were hoping when jeblair and I first discussed this idea 17:10:57 the idea was to push this out fast, then start working on radar 17:10:58 sigh 17:11:15 the main block is that it would not be as temporarily as initially was intended 17:11:16 anyway, it should be fairily close if we can et reviewers 17:11:34 why couldn't it be mmedvede ? 17:11:41 what can folks do to hurry it along? 17:11:42 ill try and take another look at it today 17:11:53 * patrickeast has been busy with cinder things :( 17:12:30 because we are asking infra team to deploy it, there needs to be infrastructure for maintaining it 17:12:54 #link https://review.openstack.org/#/c/194437/ 17:12:57 I am writing puppet module to deploy scoreboard. Almost done 17:13:43 it would be good to get more reviews so I could stop having to push a refresh for every nit 17:14:00 can this be set up in such a way that when we are ready to suggest radar hosting, it can be easily dropped in? 17:14:38 patrickeast: should we name puppet module puppet-ci_scoreboard or puppet-scoreboard? 17:14:47 mmedvede, for the naming, there are so many dashboards now proposed, I felt it added clarity without too much overhead 17:15:00 but will change it if you feel strongly about it 17:15:34 sweston, that is the intention, at least the url, vm, etc would be there 17:15:49 krtaylor: good enough 17:16:19 krtaylor: no strong feelings. If people agree with naming, need more opinions 17:16:45 yea i don't really have strong feelings on the puppet module naming either 17:17:15 darn white space, I checked it too (then added that line) sigh 17:17:26 anyway, new patchset pushed just now 17:17:52 check it out and review, the more +1's we get on it, the more likely that the infra folks will assist 17:18:32 krtaylor: my -1 was not about whitespace. I try not to -1 for style 17:18:37 Will do 17:18:50 it was for section still missing about gerrit account requirement 17:19:01 mmedvede, understood, all good, I didn't think so 17:19:30 oh, I didn't add that, crap 17:19:32 will do 17:20:03 mmedvede, what do you think? I feel like it should have it's own id 17:21:11 anyone else have any comments on the hosting spec? 17:21:14 krtaylor: it might be a necessity 17:21:22 mmedvede, agreed 17:21:33 patrickeast: do you think the scoreboard would be able to handle the load once it is used by more people? 17:21:55 mmedvede, I based this spec on other hosting specs, and it was not mentioned in them, it may be a "given" 17:22:21 patrickeast: is uses flask. I know it is possible to also use apache along flask to make it more resilient 17:22:23 mmedvede: maybe, my biggest concern would be how it is serving static files we should at some point switch it to using apache or something 17:22:32 mmedvede: yea exactly 17:22:50 or put that work into radar 17:23:03 mmedvede: it should be ok, the one i have in a little aws vm uses like <2GB of ram peak and like .1 cpu load on average 17:23:17 krtaylor: hmm, maybe. I wanted to make the account requirement explicit, so infra team knows what they are getting into :) 17:23:23 mmedvede: i don't anticipate a ton more folks would start hitting it 17:23:32 .1, that much :) 17:24:03 mmedvede, I'll add it right after this, thought I did 17:24:16 too many tasks atm 17:24:39 patrickeast: I remember you aws instance was down sometimes, did you figure out the reason? 17:24:49 ' 17:21:11 up 169 days, 16:59, 1 user, load average: 0.10, 0.10, 0.13 ' 17:24:50 from uptime 17:25:26 mmedvede: its something wrong with flask/python socket handling, i haven't tracked it down yet but also haven't had much time to look at it 17:26:08 patrickeast: I think after official announcement, the ci-dashboard.o.o might get more traffic. We can try to harden it, but it would be easier once it is running. 17:26:09 mmedvede: would probably be fixed if we migrate towards an apache integrated solution 17:27:45 mmedvede, I am proposing a Work Item to create a user account for 'ci-dashboard' 17:27:52 see any problems with that? 17:28:16 thought a generic account name could be reused for whatever solution is deployed in the future 17:29:00 krtaylor: the thing is, it would need to be account managed by infra team 17:29:23 e.g. they would need to manage private ssh key 17:29:48 mmedvede, wouldn't that be a work item? 17:30:05 so you are thinking a dependency? 17:31:34 krtaylor: Are you talking about work item inside the spec? 17:31:54 yes, in the Work Item section 17:32:22 I can note that it would need to be created and maintained by the infra team 17:32:30 +1 17:32:41 since none of us have that acls 17:33:09 ok, cool, I'll finish that asap 17:33:25 any other comments on hosting the dashboard spec? 17:34:37 BTW, lightning session topic: "Using CI Dashboard to check on a CI system's health" 17:34:45 for Tokyo 17:34:54 I'm just sayin.... :) 17:35:42 ok, let's move on 17:35:45 #topic Radar spec 17:36:06 sweston graciously moved the spec to our third party tool repo 17:36:18 #link https://review.openstack.org/#/c/211713/ 17:37:08 so, there are two ways to address this 17:37:22 1) is to merge it then patch it for spec changes 17:37:48 2) is to wait till we all agree on its content them merge it meaning that the design is complete 17:37:54 I am leaning toward 1 17:37:57 comments? 17:38:14 the (2) kind of defeats the purpose of moving it 17:38:31 I am leaning toward 1 as well, this spec has been hanging too long for approval, and it is blocking progress 17:38:31 we could work on it at its original location 17:38:54 well, moving it always implied that it would be re-proposed to infra after we worked on it 17:39:26 mmedvede, yes, long history here, it was agreed to move it 17:40:12 mmedvede, that would have been the preferred approach, but it was too confusing for some reason 17:40:41 we can't wait to improve CI system trust 17:41:04 the sytems that are busting their behinds to push reliable results need to be trusted 17:41:21 and in order to do that we have to show developers the test results 17:42:02 hence, the tactical/strategic approach 17:42:26 krtaylor: +1 17:42:27 * krtaylor gets off his soapbox 17:43:12 so, do we agree on merging it first, then patching design ideas and corrections? 17:44:15 if no one disagrees, then I'll merge it this afternoon 17:44:32 krtaylor: I vote for merging it as it is now. I will be writing Gerrit queries and integrating some data into Radar over the next two weeks, and I would prefer to have the spec approved 17:44:34 +1 for merging it 17:45:23 hard for me to have a good opinion, I did not work on many specs to understand what is best, so I abstain :) 17:45:46 +1 merge 17:45:50 done 17:46:06 so, next up 17:46:31 #topic Patches 17:46:36 #link https://review.openstack.org/#/q/project:stackforge/third-party-ci-tools+status:open,n,z 17:47:07 * krtaylor looking 17:47:32 i need to update my FC passthrough one 17:48:17 asselin_: did you and hemnafk figure out how to get the offline check one to work? 17:48:57 patrickeast, that's on hold a bit 17:49:00 patrickeast, looks like some minor changes 17:49:18 patrickeast, I'd like to get your landed first, and then look at detaching at the end of the job 17:49:30 patrickeast, explain offline check? 17:49:43 and then include the offline check 17:49:43 krtaylor: the HBA's can get into an 'offline' state 17:49:51 krtaylor: and then the passthrough will fail 17:50:05 krtaylor: the idea is to have some notification early on that it is happening 17:50:18 asselin_: that makes sense 17:50:24 ah, interesting, thanks for the education 17:50:37 patrickeast, or include that check adter the detach before the attach 17:51:28 patrickeast, perhaps we can make it more general, when fc fails altogether, send an e-mail (if configured) 17:52:01 asselin_: yea i was thinking it might be a better nagios kind of check since they don't seem to recover automatically 17:52:08 needs some manual intervention (right now) 17:53:10 ah, it is well documented in the next patch, my bad 17:54:09 so a quick open discussion then 17:54:26 #topic Open Discussion 17:54:27 krtaylor: there was another topic, about stackforge migration 17:54:41 small thing - we need to remember to add our repo to the list 17:54:57 #link http://lists.openstack.org/pipermail/openstack-infra/2015-August/003069.html 17:54:59 ah, just added, I needed to refresh 17:55:34 that is all for that :). I do not believe there is a wiki page yet 17:56:46 mmedvede, yeah, not seeing a page yet 17:57:02 but it is good to keep that on the agenda, we don't want to miss the "move" 17:57:07 thanks mmedvede 17:57:55 any other topics? 17:58:10 I've been running into a recent issue with nodepool and juno openstack clouds. 17:58:47 asselin_: i have been having issues with it too, what problem are you getting? 17:58:48 don't update unnecessarily...still working on the fix 17:59:12 {"error": {"message": "Project ID not found: admin (Disable debug mode to suppress these details.)", "code": 401, "title": "Unauthorized"}}[ 17:59:31 {"error": {"message": "User e082a15d2e6b490ba8329e60e7f092ea is unauthorized for tenant admin (Disable debug mode to suppress these details.)", "code": 401, "title": "Unauthorized"}}[ 17:59:57 thanks everyone! 18:00:08 Thanks!! 18:00:13 #endmeeting