18:01:06 #startmeeting third-party 18:01:08 Meeting started Mon Jun 9 18:01:06 2014 UTC and is due to finish in 60 minutes. The chair is anteaya. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:01:11 The meeting name has been set to 'third_party' 18:01:24 who is here for the third-party meeting? 18:01:24 o/ 18:01:29 o/ 18:01:43 that's a start 18:01:47 thanks for being here 18:01:58 #link https://wiki.openstack.org/wiki/Meetings/ThirdParty#Agenda_for_next_meeting 18:02:04 here is our agenda for today 18:02:19 #topic Welcome & Reminder of OpenStack Mission 18:02:25 so welcome 18:02:29 thanks for being here 18:02:42 #link The OpenStack Open Source Cloud Mission: to produce the ubiquitous Open Source Cloud Computing platform that will meet the needs of public and private clouds regardless of size, by being simple to implement and massively scalable. 18:02:55 the openstack misison, in case you forgot or are new 18:03:09 #topic Review of previous week's open action items 18:03:24 we had no open action items for last week 18:03:26 onward 18:03:39 #topic Announcements 18:04:08 o/ 18:04:21 third party documentation is here: Documentation: #link http://ci.openstack.org/third_party.html 18:04:44 and we have a new section on getting non-voting systems voting 18:05:04 o/ 18:05:25 this will become a staple as new folks integrate with our communication system #info How To Ask Questions The Smart Way: http://www.catb.org/esr/faqs/smart-questions.html 18:06:01 and I am working though items on this etherpad: #link https://etherpad.openstack.org/p/juno-infra-improving-3rd-party-testing 18:06:12 and one of them is about name formating 18:06:40 I am working on standardizing name formatting for third party accounts: #link https://etherpad.openstack.org/p/automated-gerrit-account-naming-format 18:07:09 currently the format guidelines are as follows: 18:07:12 Ci if it votes or ever might in the future, Bot if it will never ever vote 18:07:13 Format: 18:07:15 username --> lowercase, hyphens between words, matches Full Name, -ci or -bot at the end 18:07:17 Full Name --> capitalize, upper(minimal) and lowercase, matches username, CI, -CI, -ci or Bot, -Bot, -bot at the end 18:07:19 {company name}-{thing it is testing}-{ci|bot} 18:07:21 remove test, testing, jenkins, openstack, tempest, user from the names 18:07:41 we have an infra meeting tomorrow and I will get more feedback on these guidelines and then they will be published 18:07:52 but that is where they are right now, in draft form 18:08:04 any other announcements? 18:08:24 moving on 18:08:29 #topic OpenStack Program items 18:08:43 anyone have any items from an openstack program? 18:08:55 mestery: anything from neutron this week? 18:09:24 I'll go with no 18:09:27 and move on 18:09:31 #topic Deadlines & Deprecations 18:09:58 any deadlines or code heading for deprecation that we need to talk about this week? 18:10:24 okay 18:10:30 #topic Highlighting a Program or Gerrit Account 18:10:49 any third party accounts want or need to discuss anything about their account? 18:11:04 yes 18:11:14 sweston, you have the floor 18:11:30 what is the recommendation for multiple ci accounts in an organization? 18:11:43 that depends on what you are doing 18:12:04 vmware has one account, minesweeper that does both nova and neutron and it seems to work for them 18:12:08 salv-orlando: ping 18:12:24 ok 18:12:37 it depends on how you want it to return results and how many projects you intend on testing 18:13:09 maybe we can discuss more later 18:13:10 from a third-party account admin perspective, I would love it if all companies had one per company and then I knew who to ping for each company 18:13:31 but that doesn't tend to be how companies want to do it from their side 18:13:42 well we have 47 minutes left in the meeting 18:13:45 that would be optimal 18:13:46 IBM have the PowerKVM account who test openstack in general running in power, as well as another who test only openstack in db2 18:13:58 and hopefully others will read the logs, so it would be good to discuss now 18:14:04 ok, then we can discuss now, yay!! 18:14:10 yes 18:14:29 sweston: you are brocade, if I remember correctly 18:14:30 yes? 18:14:33 so, we have multiple projects we would like to upstream 18:14:38 yes 18:14:41 great 18:14:51 I remember talking to a brocade person at the summit 18:15:14 and told them that for my purposes, it would be optimal to have you, sweston, as point person for all the brocade accounts 18:15:19 yup, I wasn't able to make it because of another project 18:15:40 and then you and I can talk and you can ensure internally brocade is doing "the right thing" 18:15:46 understood 18:15:48 annegentle_: Nothing from Neutron (apologies for the delayed reply) 18:15:53 but that conversation happened 18:16:09 mestery: np, figured you had shown up and gotten pulled off on something else 18:16:42 are there other companies contributing multiple projects with only one gerrit account? 18:16:56 sweston: so yeah, if you are able to be brocade third-party account manager, that would be way easier for me, even if you do keep all three brocade accounts 18:17:07 sweston: ahhhh 18:17:10 * anteaya looks at list 18:17:13 hehe 18:17:21 smokestack 18:17:22 Yes, I will be third-party account manager 18:17:26 for Brocade 18:17:28 sweston: yes, PowerKVM is testing multiple projects 18:17:29 yay!! 18:17:32 thanks so much 18:17:45 smokestack tests many different projects 18:18:00 :-) 18:18:20 so what is the difference 18:18:31 between running smokestack and setting up your own ci 18:18:46 smokestack is dprince's ci 18:18:57 he decides what smokestack tests 18:19:05 the code is up somewhere 18:19:08 * anteaya looks 18:19:18 so for a reference implementation 18:19:25 that might be a place to start? 18:19:33 or even recommend 18:20:32 sweston: it is used on our pages as an example of how to set up a third party ci 18:20:59 understood 18:21:03 but I would suggest a workflow more along the lines of what we are currently using in openstack-infra now 18:21:19 are you testing propriatry code or just hardware? 18:21:46 or should I say propriatry hardware or just code 18:22:06 sweston: probably not recommended unless you like Ruby, it evolved into what was the first example of 3rd party testing in OpenStack and there are now many other examples of how to do that too 18:22:06 ok, just want to verify that I am on the right track, am using the infra manifests 18:22:33 sweston: yes that is probably a good place to start 18:22:40 dprince: everywhere, all the time :D 18:22:42 dprince: understood, I do like ruby 18:23:23 so moving on to next question 18:23:23 sweston: Our CI is very based on openstack ci, even the puppet files we use are based on openstack ones 18:23:33 I can give you some tips 18:23:51 ArxCruz: ok, thank you :-) 18:23:53 right now we have zuul, nodepool, jenkins, and even puppetboard running in our infra 18:24:06 just like upstream 18:24:26 our plans is to increase this to use logstash and other cool stuff 18:24:27 there are two ways to configure the communication between Gerrit and Jenkins, correct? 18:25:17 the jenkins gerrit trigger, and zuul 18:25:22 sweston: I use what upstream use, in this case zuul - gearman - jenkins 18:25:29 are those two methods exclusive? 18:25:59 ArxCruz: ok, gotcha. 18:26:24 jenkins have a gearman-job-plugin who connects to gearman 18:26:47 so the communication is jenkins <-> gearman-job-plugin <-> gearman <-> zuul 18:26:56 and zuul from and to gerrit 18:27:32 sweston: there are certain limitations when using jenkins with gerrit trigger plugin directly. No support for recheck for example 18:27:42 zuul is more powerfull 18:28:04 bookwar: thank you :-) was just going to ask about the differences 18:29:02 sweston: our ci is rather small thus we use simplified setup, but we already plan to move to zuul in next month or so because of better scheduling and features 18:29:18 anteaya: no proprietary hardware or code, AFAIK but I will need to check on a few things before making that official 18:29:47 sweston: if you have no proprietary hardware or code, why do you want to do third-party testing? 18:29:56 sweston: what would you be testing? 18:30:04 well, we have the vyatta plugin 18:30:13 sweston: what is that? 18:30:29 and we have adx 18:30:58 vyatta is a virtual router, and adx virtual switch 18:31:12 and brocade owns those? 18:31:22 excuse me, adx is a hardware switch 18:31:41 yes, Brocade owns them 18:31:42 is adx an opensource switch? 18:32:07 no 18:32:20 okay so you have propriatary hardware, adx 18:32:35 are you going to opensource the code for vyatta? 18:32:45 ok 18:33:00 so you have propriatary code and hardware both 18:33:07 anteaya: I need to check on Vyatta, there are licenses 18:33:17 so you are opensourcing the hooks to use the code 18:33:21 yes 18:33:38 since it is the hooks that will be opensource and you intend to merge them into openstack 18:33:40 right? 18:33:46 yes 18:33:49 great 18:34:05 sorry ... I know that was a long walk for a short drink of water :-) 18:34:16 so you use our test suite to test your hooks with the propriatary stuff 18:34:41 this is the third-party space, every conversation is a long walk for a short drink of water 18:34:54 hehe 18:34:58 need a workflow 18:35:03 right 18:35:12 talk with ArxCruz and bookwar 18:35:35 get a jenkins listening to the gerrit stream with a gerrit plugin to start 18:35:35 yes, own test suites 18:35:59 comment on the sandbox repo, test > make mistakes > ask questions 18:36:18 already have done that ... 18:36:20 once your jenkins is working with the gerrit plugin, move to zuul and friends 18:36:24 great 18:36:33 now doing ci "for realsies" 18:36:47 do you have zuul set up? 18:36:48 but want to get the implementation right the first time 18:36:57 stable logs, recheck capability 18:37:04 I have zuul bootstrapped with puppet 18:37:10 yes, that is what the sandbox is for 18:37:19 haven't started on configuring it yet 18:37:23 let's see some comments on sandbox patches this week 18:37:33 okay, so do that and ask questions in infra 18:37:33 zuul is easy to configure to not report 18:37:54 make sure you are only commenting once per patchset 18:38:04 get rid of the comment "staring jobs" 18:38:08 don't post that 18:38:21 ok 18:38:37 the post on the sandbox, ping me and we will go over it 18:38:45 ping ArxCruz and bookwar ask questions 18:38:51 =] 18:39:02 we will let you know if you are ready to comment on projects 18:39:03 sweston, our setup includes Zuul and Jenkins and using JJB, I can also assist you 18:39:09 ok, need to set some realistic expectations, for us and others 18:39:12 you won't be to start, start on the sandbox 18:39:17 sweston: yes 18:39:24 roeyc: ok 18:39:25 great, thanks roeyc 18:39:32 roeyc: which system is yours? 18:39:51 how much time should I give as an eta for when we can vote? 18:39:57 none 18:40:10 you don't know yet, since you have nothing to show us on the sandbox repo 18:40:12 Mellanox 18:40:28 I haven't seen the state of your logs, links, comment message, nothing 18:40:39 ok 18:40:57 get me that and we can start setting timelines, in collaboration with the ptls of the projects you want to test 18:41:05 sweston: tesing more than neutron? 18:41:09 roeyc: awesome 18:41:16 not right now, no 18:41:42 awesome, thank you, everyone for the information and helpful advice :-) 18:41:45 roeyc: hi, from Mirantis Fuel project, by the way :) 18:41:52 sweston: okay so once you have some testing to show me on sandbox, we will chat with mestery to see what he wants to see 18:42:07 bookwar: Hi :-) 18:42:13 sweston: thanks for using this space the way I had hoped it would be used 18:42:31 roeyc: I'm going to have to stick CI on the end of your name, Mellanox CI 18:42:43 roeyc: to fit with the new naming format 18:42:45 anteaya: Yay!! You're welcome :-) 18:42:49 sweston: :D 18:43:06 so anymore gerrit account discussion? 18:43:19 anteaya: fast status 18:43:22 not from me, all questions answered 18:43:25 sorry I was late 18:43:25 anteaya: Cool 18:43:32 luqas: np, let's hear from you 18:43:39 roeyc: thanks 18:43:43 well we fix our CI 18:43:46 sweston: great 18:43:53 it s still based on gerrit plugin 18:43:53 luqas: glad to hear those words 18:44:01 can you share what was wrong? 18:44:06 but tests neutron master correctly 18:44:09 it might help someone else 18:44:27 our plugin wasn't stacking correctly with the master branch of neutron 18:44:29 on devstack 18:44:41 did you find out why? 18:45:05 mmm my colleague tomoe was with that... 18:45:10 luqas is midokura 18:45:12 guess several things 18:45:23 go ahead 18:45:47 well basically we are reporting legitimate failures 18:46:08 we still need to add some logs, and more tests of the suite 18:46:30 and the recheck capability 18:46:40 but I guess this is zuul 18:46:53 which will be our next step 18:46:58 okay great 18:47:08 good work and thanks for the update on your status 18:47:22 anteaya: you're welcome! 18:47:26 :D 18:47:30 anyone else? 18:47:53 I'm going to change the topic to open discussion for the rest of the meeting 18:48:06 #topic Open Discussion 18:48:22 so now we can discuss anything related to third-party 18:48:33 does anyone have anything that want to talk about? 18:48:44 anteaya: are there any updates about account wiki pages? 18:48:44 yup 18:49:12 bookwar: this week I am going to create two seed pages, one for gerrit accounts and one for openstack programs 18:49:36 when is the latest that our system can be up and still be able to submit code to be included in Juno 18:49:41 if they pass review at tomorrow's infra meeting, I do believe I will be given permission to start informing people 18:49:52 anteaya: cool, thanks 18:50:04 I think I would like to start populating a few in coordination with others until we have a few good examples 18:50:10 and then announce to others 18:50:34 bookwar: if you would like to be one of the examples that is great, I will ping you after tomorrow's meeting and we can talk 18:50:38 bookwar: is that fair? 18:50:42 anteaya: yes, i'd be glad to help, so ping me if needed 18:50:49 bookwar: great thanks 18:51:01 sweston: that is a question that is project specific 18:51:19 so if you are gearing this question for neutron, we need to hear from mestery 18:51:24 if he is still here 18:51:30 anteaya: ok 18:51:42 anteaya: Sorry, I am back now, juggling a few too many things, apologies :( 18:51:45 if not, the neutron meeting is in an hour or two, get on the agenda 18:51:55 What can I help with? 18:51:59 I don't need an answer right now, but will need to have the discussion soon 18:52:05 great, so what are sweston's deadlines fo third party if he wants code in juno? 18:52:35 We just require the system to be voting correctly on the driver/plugin before it can be merged. 18:52:39 That's what we've done so far. 18:52:55 mestery: what is a realistic timeframe for includsion in juno 18:53:14 you want them commenting for 2-3 weeks before they are voting, correct? 18:53:14 Juno-2 would be the place to shoot for, after that FPF happens very fast. 18:53:18 mestery: optimally how much time would you like for review 18:53:20 right 18:53:20 Correct 18:53:24 2-3 weeks. 18:53:31 So, Juno-2 would be the ideal deadline 18:53:49 so ensure your system is stable by mid j-2, to enable you to be voting by j-2 18:53:56 and j-1 is in 3 days 18:54:05 :) 18:54:11 so you have about 3 weeks to get stable, sweston 18:54:23 so that is your system up and running, stable 18:54:30 anteaya: ok 18:54:33 stable logs, comments, gerrit history 18:54:48 so you need zuul up and running by the end of this week 18:55:06 or a running zuul for every gerrit account 18:55:27 anteaya: ok 18:55:36 that gives you two weeks to figure out how to keep it running while figuring out the rest 18:56:11 anteaya: gotcha, thank you so much for the wonderful guidance and advice 18:56:23 sweston: you ask good questions? 18:56:33 s/?/! 18:56:41 anything else for open discussion? 18:56:42 :-D 18:56:44 anteaya: so without zuul CI is not valid? 18:56:59 luqas: that is an incorrect statement 18:57:17 as long as the tests are running and we have logs the CI is valid 18:57:18 anteaya: I mean, without recheck capability 18:57:32 having recheck is one of our requirements 18:57:45 I know 18:57:49 so I would aim for having recheck ability 18:58:19 anteaya: ok, just to be sure 18:58:26 we are in j-1 18:58:42 so the conversation in j-1 is to aim for it and ask questions if you are having problems 18:59:05 luqas: make sense? 18:59:17 anteaya: completely 18:59:20 thanks 18:59:27 so it looks like we are done 18:59:32 anteaya: Can you please point me to where I can create a wiki page for our CI ? 18:59:48 thanks again, everyone for your time 18:59:49 roeyc: can I table that until next week? 19:00:04 roeyc: I will have a really good answer next week for that question 19:00:09 and a poor answer today 19:00:11 ok 19:00:13 thanks 19:00:17 we are out of time 19:00:26 thanks everyone for a great meeting 19:00:30 #endmeeting