14:00:31 #startmeeting interop_challenge 14:00:32 Meeting started Wed Jan 4 14:00:31 2017 UTC and is due to finish in 60 minutes. The chair is topol. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:34 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:37 The meeting name has been set to 'interop_challenge' 14:00:40 o/ 14:00:43 o/ 14:00:47 Hi everyone, who is here for the interop challenge meeting today? 14:00:47 The agenda for today can be found at: 14:00:47 #link https://etherpad.openstack.org/p/interop-challenge-meeting-2017-01-04 14:00:48 o/ 14:00:49 We can use this same etherpad to take notes 14:01:04 Happy new year everyone! 14:01:26 first topic: 14:01:36 #topic Welcome everyone to our 2017 Kickoff 14:01:46 WELCOME and Happy New Year!!! 14:02:11 next topic 14:02:22 thanks Brad 14:02:26 #topic Refresh everyone on the decisions we made in December 14:02:42 AGREED: Kubernetes and NFV will be our top priority workloads for this round of the interop challenge 14:02:53 #link http://eavesdrop.openstack.org/meetings/interop_challenge/2016/interop_challenge.2016-12-14-14.00.html 14:03:38 So that's gets us all refreshed 14:03:52 #topic Review last meeting action items 14:03:52 #link http://eavesdrop.openstack.org/meetings/interop_challenge/2016/interop_challenge.2016-12-14-14.00.html 14:04:14 eeiden to push up the NFV workload to new repo as soon as possible 14:04:28 Is eeiden here? Has this been done? 14:04:47 I don't recall seeing any patches on this 14:04:56 @topol, I did not see any patches. 14:05:21 K. We'll keep that action goin then 14:05:37 #action eeiden to push up the NFV workload to new repo as soon as possible 14:05:49 next action. 14:06:04 tongli to start on kubernetes 14:06:17 tongli, how is that going? 14:06:24 @topol, yes, already started working on that. 14:06:51 Willing to share some details? Enquiring minds want to know 14:07:28 When might you have a patch for folks to review? 14:07:51 probably in two weeks. 14:08:02 tongli, excellent! 14:08:08 Keep going! 14:08:10 then we can add more things if it is too basic. 14:08:18 Very cool 14:09:04 next action 14:09:07 tongli to create lauchpad for our repo 14:09:27 Looks lije hogepodge did this correct? 14:09:49 maybe docado? 14:09:55 @topol, that is for bug tracking, right? 14:10:02 tongli yes 14:10:11 if it is, then it is there already, and I have created a bug. 14:10:12 here is the link #link https://bugs.launchpad.net/interop-workloads 14:10:22 right. 14:10:25 wasnt you hogepodge? 14:10:32 it was docado? 14:10:39 it is all working, Chris created that for us. Thanks Chris. 14:10:39 THANKS docado!!! 14:10:58 topol: yes, docadeo did it 14:11:24 docado gets a Gold Star! 14:11:50 @topol, yes, 14:11:55 looks like we even have a bug posted 14:12:38 tongli looks like you triaged it 14:12:54 tongli were you going to try and fix it 14:12:59 yes, I have been trying that for awhile. 14:13:07 withno success. 14:13:12 #link https://bugs.launchpad.net/interop-workloads/+bug/1649949 14:13:13 Launchpad bug 1649949 in Interop Challenge Workloads "wordprocess failed to run wp core install" [Undecided,New] 14:13:22 something went in actually break the lampstack workloads. 14:13:32 I can not successfully run it any more. 14:14:14 so we get a new version of wordpress and that one no longer installs? 14:14:14 I went all the way back (probably 10 patches back), it worked for ubuntu env. 14:14:35 I think that the patches added to support fedora maybe the problem. 14:14:53 but I can not be sure, spend quite few days on this already. 14:15:30 tongli is there someone you think can help us to debug this? 14:16:31 @topol, anyone from the last summit can actually help. 14:16:50 thinking who added the fedora support probably can fix it. 14:17:51 So this was a patch on the old repo? can we do a git blame check and see who added it? 14:17:55 would be good indeed if the "old" workload stays in a working state ... 14:18:04 sure. 14:18:07 garloff +++ 14:18:19 Thats putting it mildly 14:18:23 :-) 14:18:37 I think we need to fix the old workloads. 14:18:50 it bothers me a lot having the latest stuff not working. 14:19:05 tongli yes we most certainly do. Let's see if we can track down who added it 14:19:28 #action tongli, topol to track down who added the fedora patch 14:19:43 #action all please help trying to get wordpress back working 14:20:26 Next item: 14:20:28 all review and add stuff to the lessons learned doc 14:20:56 is there a way for us to find a ftp/http repository so that we can put something there such as the superuser theme, content. 14:21:10 just so that these content can be used by workloads. 14:21:38 tongli Im not sure I understand your question 14:21:51 rely on some external site is sometimes difficult, when the content changes, the workloads may break. 14:22:23 currently our workload download stuff such as wordpress code, theme, and content from external sites 14:23:16 for example, we get the latest wordpress from wordpress site, however, when they release new ones, the new release of wordpress may no longer work in our workloads. 14:23:33 due to changed deployment methods or changed format. 14:23:45 so we could do that, but then we live in our own insulated bubble 14:24:15 using the latest wordpress release, we can no longer import the example content because of the format change. 14:24:29 true, I do not know what is the best way. 14:24:51 let's see how often wordpress continues to break us 14:25:08 I found that one of the problems contributing to the current workload breakage is the wordpress content format change. 14:25:09 ideally, they dont break us too often and we stay up to date 14:25:10 If this is meant to be a test environment, pinning to known working releases makes sense, since we're not testing wordpress, we're testing OpenStack 14:25:29 @hogepodge, exactly. 14:25:47 Or, just accept that if external dependencies break you fail and test for them (which is more of the 'OpenStack way') 14:25:54 different wordpress release requires different way of deploy it, 14:26:06 and the content could be different as well. 14:26:37 the current sample wordpress content was downloaded from another site , which has to be modified slightly to import to new wordpress release. 14:26:55 hogepodge, that is true. But not working on them over the long haul results in us convincing oursleves that "this is fine" but for real world folks we dont work 14:27:09 I was thinking if we can have a ftp or http repository just to save some of these static content (do not want to put these in git hub) 14:27:23 that probably can help. But I am open for other good ideas. 14:27:59 It all depends on the model of updates you want to have. Fix on fail means you're up to date, but have unexpected outages. 14:28:14 Fix on schedule means you can fall behind, especially if you don't honor the schedule. 14:28:18 @hogepodge, yes. 14:28:25 my fear is that if we can't show workloads running with the latest wordpress and others run into that we have done them a disservice 14:29:10 @topol, even with that, we still need to have a repo for content like openstack superuser award site. 14:29:20 Since were not in "important demo mode next week" my gut says seeing failures and fixing them is most valuable 14:29:38 I suspect for kubernetes and nfv we will need something as well. 14:29:51 for wordpress itself, we can pin the version by downloading from: https://wordpress.org/download/release-archive/ instead of the latest tarball 14:29:57 for context, I've spoken a bit with some of the kubernetes folks about testing on OpenStack. They do not want to test using OpenStack master, and will only accept a stable release 14:30:02 tongli, why cant we use our own repo to store anyting you want static? 14:30:18 @skazi, yes, we can do that for the wordpress itself, but not for the content. 14:30:58 tongli: true :o| 14:31:13 @topol, I was thinking about that but do we want to store these tars , zips in the code stream? 14:31:41 tongli if we put them in a separate folder I think its fine 14:31:56 and then they get reviewed and controlled like everything else 14:32:08 is everyone ok with that? 14:32:22 I do not mind doing that. 14:32:34 sounds great 14:32:37 do we have an agreement? 14:32:43 Let's try it 14:32:58 ok. sounds good to me. 14:33:09 this helps , folks, thanks. 14:33:11 any concerns? actually folks can voice concerns when they see your patch 14:33:24 sure, sure 14:33:59 are there any size limitations on those repos? 14:34:30 @skazi, good question. I do not really know. let me do a search. 14:35:05 K, let's see what happens. I see the benefits both ways. Pinning the wordpress version gives us stability. Eventually somone might complain we arent working with latest wordpress 14:35:27 1 GB per repo. 14:35:35 100 MB per file. 14:35:44 that is the strict limit. 14:35:54 so hopefully are zips and tars are under these limits? 14:35:59 any file bigger than that will be rejected. 14:36:01 err our zips and tar 14:36:19 topol: mu suggestion would be to just schedule mainteance to bring it up to date, but I know some would throw things at me for that suggestion 14:36:36 hogepodge I like that idea 14:37:22 tongli markvoelker, thoughts? 14:37:48 you want to try pinning the old wordpress for now? 14:38:18 well, we do want to specify a release for sure. 14:38:29 Seems like it would be relatively easy to provide option to try the latest wordpress or use a local copy. Best of both worlds? 14:38:32 blindly using the latest release will be an issue regardless. 14:38:50 E.g. maybe have a variable for the tarball location which can be a url or file:/// ? 14:38:53 markvoelker, good point 14:39:06 tongli: with such limits I would suggest to pin as much as possible from external resources and keep only our/modified content on the repo 14:39:07 we already have the option (in configuration file) for you to select a release (where to get the wordpress). 14:39:18 but the code to deploy them is the same. 14:39:45 @skazi, we can try that. 14:40:06 skazi that plan sounds good to me as well 14:40:13 smaller file can go in to the git hub repo, bigger file, other means. 14:40:50 Ok, let's try that 14:41:24 tongli with that approach will we be able to get wordpress bck runing soon? 14:41:47 @topol, I hope so, 14:42:03 Ok, lets hope for the best 14:42:11 it is just the time. I really really want to spend some time fixing the existing workload first. 14:42:17 then dive into the kubernetes. 14:42:22 tongli, thats fine 14:42:26 priority thing. 14:42:29 go ahead 14:42:37 your priority makes sense to me 14:43:09 most folks arent back to work for another week anyway :-0 14:43:23 next item: 14:43:34 topol create a short blog article to update everyone on the status of the interopchallenge 14:43:41 I will get this done this week :-) 14:44:09 Will include some pictures I took of everyone at the summit 14:44:42 #topic open discussion 14:44:57 anything else we need to discuss? 14:45:24 @topol, couple of things. 14:45:29 go ahead tongli 14:45:38 there is a comment to the lesson's learnt patch. 14:45:49 I copied the comment to the etherpad, 14:45:49 o/ that was me 14:45:55 please take a look. 14:46:18 @zhipengh, please let everybody know what you mean. I am not sure I can address your comment. 14:46:57 okey. I do like the current content of the lesson leanrt doc, however it might just be too techy for some users 14:47:14 therefore what I suggested is to have an overview section 14:47:35 describe in a broad stroke that what interop chanllege has found/fixed 14:47:48 then go into the details like what we have now 14:48:21 zhipengh Having an overview section makes sense to me. Will you push a patch with a first draft? 14:48:39 sure no problem :) 14:48:45 @zhipengh, I would like to have the current patch in, then you can change based on this. 14:48:56 tongli okey 14:48:58 otherwise, it will be just sitting there. which is what happened. 14:49:01 zhipeng, cool 14:49:19 tongli does your patch need some +2's? 14:49:27 tongli got a review link? 14:49:36 in the etherpad. 14:49:44 @topol, yes. 14:50:18 https://review.openstack.org/#/c/409847/ 14:50:22 Folks it s #link https://review.openstack.org/#/c/409847/ 14:50:25 sorry, I lied, it was not in the etherpad. 14:50:43 Im gonna +2 it even though tongli lied and made me check the etherpad 14:51:08 @topol, most forgiven leader I ever had. 14:51:14 @topol, thanks. 14:53:06 anyone else able to +2 and +A and then zhipeng can add to the merged patch? 14:53:16 tongli :-) 14:53:42 Mark can, luzC can, zhipengh can 14:54:02 * markvoelker is looking again as I haven't read it since before the holiday break...stand by 14:54:45 we can keep adding patches that revise it. 14:54:52 @markvoelker, please take a look again. 14:55:12 @topol, yes, that is the point of having the github so that we can always update it. 14:55:13 done 14:55:23 cool! Thanks 14:55:51 thanks guys. 14:56:14 #action zhipengh to add overview section to lessons learned doc 14:56:21 4 mins left. anything else? 14:56:40 will we have ptg sessions ? 14:57:24 zhipengh, we definitely should. Do we want informal sessions or do we want to lock in a time 14:57:45 I believe there will be plenty of rooms/space for us to meet 14:58:16 is it required to book a room through Thiery ? 14:58:57 I asked him and he told me we could wither use some of the refstack room or their would be informal meeting space as well 14:59:17 cool 14:59:28 I logged his response in one of our last meetings 14:59:34 so space is covered. 14:59:53 its more how much time folks want to carve out and use for interop? A day? 15:00:24 lets discuss this at next weeks meeting 15:00:33 half day maybe enough. 15:00:41 #action all discuss PTG time allotment 15:00:46 THANKS EVRYONE 15:00:49 agree tongli 15:00:49 #endmeeting