19:01:28 <jeblair> #startmeeting infra 19:01:29 <openstack> Meeting started Tue Aug 26 19:01:28 2014 UTC and is due to finish in 60 minutes. The chair is jeblair. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:31 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:01:33 <openstack> The meeting name has been set to 'infra' 19:01:33 <nibalizer> o/ 19:01:36 <jeblair> #link agenda https://wiki.openstack.org/wiki/Meetings/InfraTeamMeeting 19:01:38 <krtaylor> o/ 19:01:43 <jeblair> #link last meeting http://eavesdrop.openstack.org/meetings/infra/2014/infra.2014-08-19-19.02.html 19:01:45 <jesusaurus> o/ 19:01:55 <jeblair> #topic Actions from last meeting 19:02:01 <jeblair> jeblair Publish python-jenkins 0.3.3 (last release without pbr) tarball to pypi 19:02:03 <jeblair> that happened 19:02:07 <zaro> o/ 19:02:09 <jeblair> so did 0.3.4 19:02:09 <anteaya> yay 19:02:15 <jeblair> because 0.3.3 was broken 19:02:19 <anteaya> :( 19:02:32 <fungi> if there were no broken releases we'd never have new releases ;) 19:02:38 <anteaya> ha ha ha 19:02:47 <jesusaurus> heh, true 19:02:50 <jeblair> but now i think that's all taken care of, and the next (0.4.0?) release can be done by zaro by pushing a tag in the normal way 19:03:12 <ianw> o/ 19:03:19 <jeblair> pleia2 create new mailing lists 19:03:20 <zaro> i think that will happen in about a month. 19:03:34 <pleia2> that was done, anteaya will talk about it later in the meeting 19:03:40 <jeblair> #link http://lists.openstack.org/cgi-bin/mailman/listinfo/third-party-announce 19:03:41 <anteaya> I can talk now 19:03:45 <jeblair> pleia2: i reordered :) 19:03:48 <anteaya> they are up 19:03:50 <pleia2> oh good :) 19:03:51 <anteaya> thanks pleia2 19:03:59 <jeblair> #link http://lists.openstack.org/cgi-bin/mailman/listinfo/third-party-request 19:04:07 <anteaya> this is my first time admining lists, feedback welcome 19:04:07 <krtaylor> yea! 19:04:10 <jeblair> #link https://review.openstack.org/#/c/116989 19:04:26 <anteaya> and that is the patch to redirect folks to use the lists 19:04:41 <anteaya> a speedy iteration prevents a week of transition 19:04:55 <jeblair> yeah, probably best to do this quickly to avoid confusion 19:05:05 <jeblair> are there outstanding requests from folks we have received on the infra list? 19:05:16 <clarkb> jeblair: there are a few yes 19:05:17 <pleia2> a couple 19:05:29 <anteaya> I'm standing by to offer new patchsets for 116989 if folks have -1s 19:05:44 <jeblair> okay, we should probably avoid asking people to resubmit on the new list, that's just silly 19:05:49 <anteaya> agreed 19:05:58 <jeblair> i think when anteaya's change merges, we should announce the new policy to the old list 19:06:02 <jeblair> ask everyone to subscribe to the announce list 19:06:07 <fungi> agreed 19:06:09 <jeblair> and new requests go to the request list 19:06:09 <anteaya> agreed 19:06:19 <jeblair> and explicitly mention that old requests don't need to be resubmitted 19:06:37 <anteaya> I can bulk import gerrit email addresses to the announce list 19:06:46 <anteaya> anyone think that is a bad idea? 19:07:11 <jeblair> anteaya: we could consider inviting them, but not actually subscribing them 19:07:21 <anteaya> okay, I can invite them 19:07:27 <jeblair> anteaya: mailman lets you do either in bulk 19:07:32 <pleia2> anteaya: that's an option in the interface 19:07:34 <pleia2> yeah 19:07:41 <anteaya> if we prefer I invite, I will invite 19:07:42 <jeblair> (they'll get a message saying 'click here to confirm' etc) 19:07:53 <krtaylor> ++ 19:07:54 <jeblair> yeah. i'd be very much opposed to subscribing without confirmation 19:07:54 <fungi> right, never bulk-subscribe people without prior consent or a very, very good reason 19:07:56 <anteaya> should reduce some cruft right off the bat 19:08:08 <anteaya> invite it shall be 19:08:21 <anteaya> I think that was all I needed here 19:08:27 <anteaya> anything I am missing? 19:08:41 <anteaya> okay thanks 19:09:05 <jeblair> #action anteaya invite third-party email addrs from gerrit to announce list 19:09:32 <jeblair> who wants to send the email to -infra announcing the new lists/policies after 116989 merges? 19:09:51 * jeblair is happy to if no one else loves that idea 19:10:26 <pleia2> I can do it 19:10:26 <fungi> i can as well 19:10:40 * fungi bows to pleia2 ;) 19:10:44 <anteaya> everyone wants that job 19:10:54 <jeblair> #action pleia2 send the email to -infra announcing the new lists/policies after 116989 merges 19:11:03 <jeblair> anteaya, pleia2: thanks! 19:11:05 <pleia2> anteaya: left a comment in the review 19:11:17 <jeblair> #topic Priority Specs (jeblair) 19:11:19 <clarkb> I left one too which expanded on pleia2's 19:11:27 <jeblair> #link https://review.openstack.org/#/c/100363/ 19:11:58 <fungi> i think that one's probably ready for approval unless more people want to go through it? 19:12:10 <clarkb> I was happy with it 19:12:19 * nibalizer here to answer questions 19:12:21 <fungi> or unless nibalizer wants to address any of the suggestions on that last patchset 19:13:01 <nibalizer> ill update it for those suggestions 19:13:14 <nibalizer> clarkb: did you want to talk more about the symlink? 19:13:31 <clarkb> nibalizer: not here, if you update the spec about what that is for I think that is enough 19:13:35 <nibalizer> okay sweet 19:13:36 <nibalizer> will do 19:13:44 <jeblair> nibalizer: also, "Set hiera.yaml appropriately to source both dirs in order" -- does that need more detail too? 19:13:58 <jeblair> that's an infra-root action, right? 19:14:15 <nibalizer> jeblair: no hiera.yaml will be managed by puppet 19:14:24 <nibalizer> it should be a file resource 19:14:31 <jeblair> ok nevermind then. it can be documented in the puppet 19:14:44 <wenlock> nibalizer, as a side note, we use hieradata strucutre on forj.io, we manage this with puppet, no symlinks required 19:14:54 <jeblair> nibalizer: so we'll wait for your update about the rationale for the symlink, then it looks like it's ready to approve 19:15:06 <nibalizer> okay cool 19:15:38 <jeblair> wenlock: maybe you could weigh in on https://review.openstack.org/#/c/100363 after nibalizer writes the update? 19:15:47 <nibalizer> wenlock: lets follow up after the nmeeting 19:15:49 <wenlock> yes, i'll read up on it today 19:16:09 <jeblair> wenlock: that way we'll know if we're doing something different than you are that requires it, or if we've missed something and don't really need it :) 19:16:21 <wenlock> nibalizer, sounds good, we can point you to our source on that 19:16:31 <jeblair> #link https://review.openstack.org/#/c/99990/ 19:16:58 <jeblair> i think this is probably ready modulo syntax errors? 19:17:08 <nibalizer> i think at this point that one is us fighting with restructured text parsing 19:17:10 <wenlock> jeblair, i'd like to propose that you use librarian-puppet as option in install_module.sh to manage the repos 19:17:19 <clarkb> I need to rereview it with the new problem statement but yes I think it is pretty close 19:17:30 <clarkb> wenlock: I will -2 that :) 19:17:31 <nibalizer> but i think for the ideas there we're largely getting consensus 19:17:35 <clarkb> wenlock: we can talk about why after the meeting 19:17:37 <jesusaurus> i think i cleaned up all the syntax errors, but the work items could use a re-review 19:17:41 <wenlock> jeblair +2 for subtree workflow too, this worked nice on our side 19:17:59 <wenlock> clarkb, yes, i'd like to understand how you will manage tags/revs, etc. 19:18:04 <jeblair> jesusaurus: it's currently failing tests 19:18:15 <anteaya> nibalizer: I agree with the content of 99990 if we can get the rst parsing figured out 19:18:34 <jesusaurus> jeblair: thats an old test, it should get through the check queue in like 6 hours or so 19:18:38 <jeblair> oh ok 19:18:58 <jeblair> #link https://review.openstack.org/#/c/110730/ 19:19:03 <jeblair> oh no, i have unaddressed comments 19:19:28 <fungi> jesusaurus: i added some comments to the latest patchset as well (i had them in draft for an earlier one) 19:20:15 <jeblair> anyway, i think resolution to the comments already in that one is straightforward, so please go ahead and review with that in mind 19:20:16 <jesusaurus> fungi: thanks 19:20:36 <fungi> jeblair: i have some more in draft i need to add too, though i can port them forward to the next iteration if i don't get to it before you update 19:21:28 <jeblair> #link https://review.openstack.org/#/c/110793/ 19:21:55 <jeblair> a few updates noted inline there 19:22:03 <clarkb> jeblair: I think you need to make at least one edit to 110793 as well 19:22:06 <clarkb> jeblair: but it is pretty close 19:22:24 <jeblair> probably the main thing is that there's a section where i list 3 options for openstack-manuals 19:23:21 <jeblair> option 2 is yucky. option 3 is probably the default option because it doesn't really change anything from the docs pov 19:23:54 <jeblair> option 1 actually simplifies the jobs somewhat, at the cost of some extra cpu time 19:24:00 <clarkb> I am a fan on 1 19:24:02 <clarkb> s/on/of/ 19:24:21 <jeblair> i should probably work with AJaeger on quantifying that and seeing what the impact would be 19:25:13 <jeblair> anything else on these specs? 19:25:37 <fungi> yeah, i think i'm really okay with any of the three options, keeping in mind that eventual optimization with afs will probably trump any of them for efficiency anyway 19:25:57 <jeblair> yup 19:26:07 <fungi> so simpler==better for now probably 19:26:26 <jeblair> #topic Translations demos active, sent to Daisy (pleia2) 19:26:51 <pleia2> this is mostly an FYI 19:27:04 <pleia2> we have demos up for both zanata on wildfly and pootle 2.6 19:27:32 <jeblair> pleia2: do you have an idea of the relative ease of pupetting those two options? 19:27:43 <clarkb> note the pootle 2.6 demo has not had translate-dev dns udpated to point at it because we lack openid and I want to keep it in a position where its really just a demo and not a dev server 19:27:53 <pleia2> we have step by step instructions for how they both were deployed in etherpads, but neither of them will be easy 19:28:23 <pleia2> zanata is a bit trickier since the wildfly support is not official yet, so it's a bit hacky as they continue to work on it 19:28:32 <clarkb> I think the biggest thing from pootle side is that we will have to run the various django managementy commands some of which expect human input (may need a graphite like hack for that) 19:28:32 <jeblair> pleia2: we do run one django app via puppet (graphite) 19:28:33 <fungi> pleia2: also a gut feel for which would require more ongoing care and feeding (infra core effort)? 19:28:57 <pleia2> fungi: zanata, since we don't work with jboss/wildfly today, and we do work with django 19:29:14 <clarkb> I do have a couple concerns about pootle. The first is allauth makes the openid stuff harder not easier :/ and second the UI is thoroughly unintuitive at least to me 19:29:26 <clarkb> but for the second thing I defer to the translation team(s) 19:29:46 <anteaya> how long do the translation folks need to complete their assessment? 19:29:57 <pleia2> I haven't looked deeply into zanata's openid support, but right now in our demo it is taking the https://launchpad.net/~lyz addresses as logins (we'll need to simplify and restrict this) 19:30:07 <jeblair> clarkb: what do you think needs to happen for openid? 19:30:25 <jeblair> pleia2: in either case, i think nothing short of openid sso is acceptable 19:30:31 <pleia2> anteaya: uncertain, we're getting to the point in the cycle where they will be really busy with translations work 19:30:32 <jeblair> clarkb: for pootle 19:30:40 <anteaya> pleia2: fair 19:30:46 <clarkb> jeblair: allauth needs to support a single openid provider (there is an open bug for this). also allauth and/or pootle need to learn how to do a single type of auth 19:31:08 <clarkb> jeblair: today it looks like you always get local auth in addition to whatever allauth other mechanisms you have enabled 19:31:26 <pleia2> as far as evaluation, Daisy has admin on both systems and I'll be working with her to help present the demos to the team on a schedule she determines 19:31:35 <clarkb> there is an open bug against pootle for the second thing which I need to follow up on and possibly file a bug against allauth for 19:31:38 <anteaya> yay Daisy 19:31:40 <pleia2> if anyone else wants admin, lmk 19:32:00 <anteaya> pleia2: you're a great admin :D 19:32:04 <jeblair> pleia2: do you have any idea how receptive the pootle dev(s) would be to this kind of work? 19:32:32 <pleia2> jeblair: they've really been leaning on django for all openid stuff, clarkb has some open bugs that they've responded to 19:32:50 <clarkb> jeblair: pleia2: ya I think they would be receptive but have already said go fix it in allauth 19:32:56 <pleia2> yeah 19:34:04 <jeblair> how about zanata? what's it take for openid there? 19:34:20 <pleia2> it works, haven't looked into restricting to only openid 19:34:37 <jeblair> i should say openid sso 19:34:48 <jeblair> cause it sounds like right now, it asks for your openid, right? 19:35:04 <pleia2> http://15.126.226.230:8080/account/register 19:35:24 <pleia2> yeah, so where it wants openid just put in your http://launchpad.net/~user address 19:35:34 <jeblair> and it also has the "and a local account" problem 19:35:41 <pleia2> yeah 19:36:22 <jeblair> well, i added openid sso to gerrit; i could probably add it to zanata too :) 19:36:25 <pleia2> I don't know what mechanism they're using for this, but they've been eager to help us get support for other things we need 19:36:34 <jeblair> even better if they do it :) 19:36:46 <jeblair> i also added it to pootle, but that's neither here nor there 19:37:51 <pleia2> I think that's it for this week, we'll tackle issues down the road as we come to them and the translations team gets a better idea of what works for them 19:37:56 <jeblair> i think from our pov, openid-sso is critical, and easy of installation/maintenance is the next most important 19:38:02 * pleia2 nods 19:38:24 <clarkb> ++ 19:38:26 <pleia2> zanata install steps: https://etherpad.openstack.org/p/zanata-install 19:38:34 <jeblair> #link https://etherpad.openstack.org/p/zanata-install 19:38:55 <pleia2> pootle install steps (down at line 47 and below): https://etherpad.openstack.org/p/pootle-250-upgrade 19:39:21 <jeblair> #link https://etherpad.openstack.org/p/pootle-250-upgrade 19:40:09 <jeblair> we probably have some anti-patters in puppet for gerrit that we can crib for some of the zanata stuff 19:40:13 <pleia2> oh, and zanata will probably be available via ansible..recipes? so hopefully we'd be able to convert them to puppet 19:40:14 <jeblair> anti-patterns 19:40:45 <jeblair> pleia2: cool, thanks 19:40:55 <jeblair> #topic Open discussion 19:41:08 <anteaya> so something has come out of the reviews so far on my patch 19:41:27 <anteaya> what are the expectations for thrid party folks regarding the two new lists 19:41:30 <fungi> the upcoming renames list has the dashboard puppet module on it, proposed to go to openstack-attic... that can come off the list right? 19:41:41 <anteaya> I want all third party folks to subscribe to both lists 19:41:54 <anteaya> to announce to keep alert for their system if it is disabled 19:42:00 <anteaya> to request to help me out 19:42:18 <jeblair> fungi: yeah, we should just merge a change to readme saying it's dead 19:42:28 <anteaya> https://review.openstack.org/#/c/116989 19:42:53 <fungi> jeblair: i'll propose a bunch of "it's dead jim" patches to projects in the same boat in that case 19:43:09 <jeblair> anteaya: i think that would be useful. i think announce should be considered a requirement (i don't intend on policing it); requests is a nice-to-have 19:43:12 <anteaya> only if I can call you bones, fungi 19:43:25 <anteaya> jeblair: I can live with that 19:43:39 <krtaylor> ++ 19:43:43 <anteaya> I'm not going to police either list 19:43:55 <anteaya> but if they don't know their system is disabled it is on them 19:44:05 <fungi> i think announce should be a very strong suggestion, since if we're making changes you need to be aware of or taking your systems offline, that's where you're going to find out about it 19:44:09 <jeblair> anteaya: exactly 19:44:34 <anteaya> fungi: I tried to capture that strong suggestion in the wording of my patch 19:44:34 <krtaylor> I vote requirement 19:44:40 <anteaya> krtaylor: no 19:44:40 <fungi> and if you aren't subscribed to the announce list or aren't paying attention, then it's your problem not ours 19:44:53 <anteaya> requirements are basis for disabling a system if they aren't met 19:45:01 <krtaylor> anteaya, why not? 19:45:15 <anteaya> if they aren't subscribed to announce disablying their system isn't going to be my action 19:45:39 <jeblair> anteaya: it will also be where we send notifications of policy changes (lack of adherence to which would get systems disabled too) 19:45:48 <anteaya> jeblair: agreed 19:46:11 <krtaylor> I am thinking more in terms of announcements that CI teams will want to know, that is important 19:46:19 <krtaylor> what jeblair said 19:46:22 <anteaya> fungi: also I have tried to capture the strong wording in teh message on the landing page for the list 19:46:32 <anteaya> krtaylor: right 19:47:21 <krtaylor> so why wouldn't it be a requirement? we need a global communication channel 19:47:38 <anteaya> requirements are what gets your system disabled if you don't have them 19:47:49 <krtaylor> or not granted in the first place 19:47:58 <anteaya> I am not going to disable a system if they don't subscribe to a mailing list 19:48:40 <anteaya> they are foolish if they don't, but that is on them 19:48:42 <krtaylor> I don't think that would be a problem, but listing it in the requirements section would be ok 19:48:49 <anteaya> I disagree 19:48:56 <fungi> however they may get disabled because they missed announcements on that mailing list, or they might not find out in a timely manner that they were disabled because we reached out to them via that list 19:49:22 <anteaya> fungi: true 19:49:28 <anteaya> again their responsibility 19:49:43 <anteaya> some of the onus has to be on them for their actions 19:50:24 <krtaylor> true, but not a good track record of that, at least initially 19:50:30 <anteaya> no 19:50:51 <anteaya> so anyway I hope taht clarifies my patch 19:51:19 <krtaylor> I am happy we have the lists, hopefully everyone will sign up 19:51:36 <jeblair> ++ 19:53:44 <jeblair> well, if that's it, i reckon we can end early 19:53:56 <clarkb> oh I did the tox and trusty stuff 19:53:58 <clarkb> it went mostly ok 19:54:07 <jeblair> oh cool 19:54:13 <fungi> it went remarkably well, i thought 19:54:21 <clarkb> glance hiccuped and so did a couple of our tools and stackforge puppet 19:54:22 <anteaya> yay tox and trusty 19:54:28 <clarkb> but considering we have several hundred projects it went well :) 19:54:53 <anteaya> :) 19:54:54 <clarkb> thats it 19:54:57 <fungi> and some review teams got a bit of a learning experience about prioritizing prerequisite changes for scheduled infra activities 19:55:06 <jeblair> haha 19:55:36 <clarkb> but we are future proofed until tox does their next release 19:55:50 <anteaya> great 19:55:53 <jeblair> clarkb: is there more to do with trusty? eg, zuul layout? 19:56:06 <hashar> hey. is that open discussion yet? :D 19:56:16 <clarkb> jeblair: nope, not unless the stackforge puppet team wants to move puppet 2.7 back to precise 19:56:19 <pleia2> hashar: has been for quite some time :) 19:56:23 <hashar> great 19:56:23 <nibalizer> clarkb: yeaaaaa 19:56:27 <nibalizer> so i think that gate is busted 19:56:35 <clarkb> nibalizer: even with mgagne's fix? 19:56:41 <nibalizer> did that land? 19:56:43 <clarkb> yes 19:56:43 <nibalizer> https://review.openstack.org/#/c/116915/ ? 19:56:45 <hashar> so for those that don't know me, I am the continuous guy at wikimedia foundation and I have basically copy pasted Zuul setup to our infra. 19:57:11 <jeblair> hashar: and improved zuul :) 19:57:13 <hashar> I would like to hereby publicly thank you in the name of me and the wikimedia foundation folks for all your hard work supporting third party installation 19:57:16 <nibalizer> clarkb: its like i was saying yesterday, 'double puppet install' isnt the problem, ruby 1.9.3 is the problem 19:57:25 <hashar> from Zuul (which is awesome) to kindly maintaining python-jenkins 19:57:41 <jeblair> hashar: oh, you're very welcome! 19:57:50 <fungi> hashar: you're welcome, and thanks for all your help too! 19:57:51 <mgagne> clarkb: so much lag, no result yet on my test :-/ 19:57:53 <jeblair> hashar: thank you for using it and helping us make it better 19:58:11 <zaro> hashar: thanks in turn for your contributions! 19:58:13 <hashar> our jobs are stressful, and little time is spent to say thanks. So here you have: merci beaucoup ! 19:58:35 <anteaya> thanks hashar 19:58:46 <mgagne> -> Thank you for helping us help you help us all. 19:58:52 <anteaya> :) 19:58:59 <jeblair> and we're going to start using hashar's zuul-cloner soon :) 19:59:00 <hashar> and I got Zuul cloner deployed in production last week. It is now voting as of today ! :D 19:59:04 <jeblair> mgagne: well put! :) 19:59:14 <fungi> hashar: also, in a broader context, thanks for keeping mediawiki and by extension wikipedia working. i use them both a lot 19:59:26 <hashar> (((that is on Wikimedia production, not OpenStack! ))) 19:59:28 <anteaya> yes 19:59:30 <ttx> I used that too 19:59:38 <clarkb> hashar: nice! 19:59:40 <jeblair> fungi, hashar: ++ 19:59:42 <mgagne> with GLaDOS' voice 19:59:47 <hashar> yeah MediaWiki is quite fun :] 19:59:53 <clarkb> hashar: and thank you! 20:00:08 <hashar> thought it can be scary and is not yet as tested as openstack can be. But we are working on it! 20:00:18 <jeblair> what a nice way to end a meeting :) 20:00:22 <jeblair> #endmeeting