18:00:17 <nirik> #startmeeting Infrastructure (2015-03-05) 18:00:17 <zodbot> Meeting started Thu Mar 5 18:00:17 2015 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:17 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:00:17 <nirik> #meetingname infrastructure 18:00:17 <zodbot> The meeting name has been set to 'infrastructure' 18:00:17 <nirik> #topic aloha 18:00:17 <nirik> #chair smooge relrod nirik abadger1999 lmacken dgilmore mdomsch threebean pingou puiterwijk 18:00:17 <zodbot> Current chairs: abadger1999 dgilmore lmacken mdomsch nirik pingou puiterwijk relrod smooge threebean 18:00:17 <nirik> #topic New folks introductions / Apprentice feedback 18:00:25 <nirik> morning everyone. 18:00:35 <threebean> hey hey 18:00:43 * mirek-hm is here 18:00:48 <puiterwijk> (UGT) morning everyone 18:00:55 <pingou> Morning 18:00:59 <Cydrobolt> morning 18:01:02 * danofsatx is lurking about in the vicinity 18:01:03 * lmacken 18:01:24 <andreasch> hi 18:01:51 <nirik> we have a gobby document up. See https://fedoraproject.org/wiki/Gobby 18:01:54 * tflink is here 18:01:57 * Cydrobolt is here, but may be distracted 18:02:20 <nirik> also http://gobby.fedoraproject.org/cgit/.git/ 18:02:58 * oddshocks here 18:03:01 <Cydrobolt> installing gobby 18:03:01 <smooge> hello. sorry was playing with gobby/sobby/obby 18:03:04 <nirik> any new folks or apprentices with questions? 18:03:48 <nirik> In 2 minutes or so I will dump all the #infos from the gobby doc here for meetbot to digest. ;) 18:03:58 <nirik> feel free to add any before then... 18:04:03 <Cydrobolt> nirik, first time able to catch this meeting 18:04:10 <nirik> hey Cydrobolt. welcome. 18:04:17 <Cydrobolt> :) 18:06:15 <nirik> #topic Information / status updates from the last week 18:06:23 <nirik> <flood coming> 18:06:26 <nirik> #info Trying new meeting process this week with shared document - kevin 18:06:26 <nirik> #info Figured out why squid smp mode wasnt working and posted to list - kevin 18:06:26 <nirik> #info Ipsilon is now in staging, please test and give me your problems - puiterwijk 18:06:27 <nirik> #info A large number of bug/RFEs have been coming in on anitya, the-new-hotness, and fmn. Any help appreciated. - ralph 18:06:27 <nirik> #info reminder: we are in Fedora 22 Alpha Freeze - kevin 18:06:29 <nirik> #info worked on getting gobby back in shape - kevin 18:06:31 <nirik> #info made progress on new openstack cloud and kept old one stumbling along - kevin and puiterwijk 18:06:33 <nirik> #info nagios alerts this week - http://ur1.ca/jr7j4 - kevin 18:06:35 <nirik> #info moved osu hosts around, should help nagios alerting - kevin 18:06:37 <nirik> #info fixed issue with varnish caching old wiki pages - puiterwijk 18:06:39 <nirik> #info progit moving along but the more tester the better - pingou 18:06:41 <nirik> #info fedmenu prototype nearing completion, feedback requested http://threebean.org/fedmenu - ralph 18:06:56 <nirik> #topic New meeting process 18:07:06 <dgilmore> hola 18:07:12 * pbrobinson waves 18:07:15 <nirik> so this is all new, what do folks think? how can we adjust it or should we just go back to the old way? 18:07:20 <nirik> welcome dgilmore and pbrobinson. :) 18:07:25 * roshi lurks 18:07:30 * pingou ok with trying a little longer 18:07:36 <oddshocks> +1 18:07:44 <puiterwijk> yeah, +1 on trying at least some more time 18:07:45 <pingou> I thing I was thinking is that it makes it easier for people to not be present 18:07:52 <mirek-hm> nirik: +1 it will be faster (although maybe little bit async) 18:07:55 <nirik> It seems like folks don't remember to update the shared doc really... but not sure how to fix that 18:07:59 <pingou> I just dump my info on gobby and then I can go do something else 18:08:05 <threebean> heh, I think I like it already. 18:08:09 <nirik> pingou: yeah. unless there's a discussion item? 18:08:14 <nirik> that you care about 18:08:26 <pingou> nirik: yes, but easily missed, especially if added at the last minute 18:08:38 <smooge> no no no.. he just agrees to take the work from the item he missed 18:08:51 <nirik> true. Perhaps those items should be only added until the day before... 18:08:55 <Cydrobolt> it sounds interesting 18:09:24 <nirik> gobby isn't the best thing in the world either. ;( But it's free and we have it now... 18:09:35 * jsmith shows up late and lurks 18:09:59 <mirek-hm> ad OpenStack - all services are now available using SSL and only SSL ports are exported via endpoints. Everything works using command line tools, but when I start VM I get error, therefore it is still not finished 18:10:15 <pingou> arf 18:10:17 <pingou> so close 18:10:30 <nirik> mirek-hm: I was going to look at vnc later today perhaps... and I can look at the vm error. 18:10:45 <nirik> ok, so sounds like folks are ok to keep trying and adjusting this new format then? 18:11:15 * relrod here late, sorry! 18:11:23 <Cydrobolt> hi relrod 18:11:56 <nirik> ok, moving on 18:12:00 <nirik> #topic Fedorahosted 18:12:11 <nirik> sounds like we had some good discussion on list about this... 18:12:29 <nirik> and the concensus was mostly to just keep the current setup, but move progit forward as we can 18:13:16 <nirik> I guess we should move it to rhel7 though... 18:13:25 <Cydrobolt> nirik, do you have a link to progit? 18:13:37 <Cydrobolt> I can only find information about the book 18:13:57 <puiterwijk> nirik: I can take a look at trac on upgrading our current traci nstances to the new trac 18:14:09 <pingou> Cydrobolt: http://209.132.184.222/progit/issue/47 18:14:18 <puiterwijk> (so packaging requires extensions etc) 18:14:29 <pingou> btw, everybody is welcome to weight in that ticket as well ^ :) 18:14:37 <Cydrobolt> ah, we are developing it ourselves? 18:14:42 <nirik> puiterwijk: so, I looked into this. looks like 0.12.x is almost dead... they might do one more release, but thats it. 18:15:04 <puiterwijk> nirik: yeah, I saw the summary on your email. I think we want to look if 1.X is doable 18:15:05 <nirik> so, 1.0 seems like the best bet. It's unlikely to live for 10 years tho... 18:15:10 <Cydrobolt> interesting 18:15:32 <nirik> but I think we can just package it as 'trac' in epel7 and then when it dies we can package 1.2 as trac12 or whatever. 18:15:33 <puiterwijk> right, but I'm hoping the changes in the X-series will be minimal enough 18:15:38 <nirik> ie, not worry about the parallel right now 18:15:41 <puiterwijk> yeah 18:15:48 <pingou> \ó/ 18:15:56 <nirik> in which case, we don't need reviews, just branching and building and testing 18:16:13 <puiterwijk> yeah 18:16:34 <puiterwijk> so I'd be willing to look into that next week, and see if I can get an ansible playbook for hosted 18:16:49 <nirik> cool. I can build up the stack if you want to work on the playbook? 18:16:52 <nirik> or vice versa. 18:17:17 <puiterwijk> let's discuss that after the meeting, as it depends on where the stack is at this moment I guess 18:17:19 <nirik> I already do have the trac maintainers go ahead to branch 18:17:23 <nirik> sure. 18:17:41 <nirik> #info will move hosted to rhel7 and keep trac and current setup for now. 18:17:55 <nirik> #info will work on progit to groom it to replace hosted someday when it grows up. 18:18:18 <nirik> ok, any other discussion items before we move on to the 'learn about...' section? 18:19:03 <smooge> I do have a question 18:19:17 <nirik> fire away. ;) 18:19:28 <smooge> will we be putting the trac we use on the el7 boxes in epel or private hosted? 18:19:43 <nirik> I was assuming epel7... 18:19:53 <nirik> no real reason not to IMHO. 18:20:16 <smooge> OK but I saw that there were massive updates to trac from what we are using.. I didn't know if we later updated to that trac what the effect would be 18:20:33 <smooge> eg the mediawiki doom 18:20:50 <nirik> well, not sure how massive it is really... they changed their numbering... 0.11, then 0.12, then 1.0 18:21:09 <nirik> 1.x where x is odd is devel release, even is 'stable' release. 18:21:09 <smooge> or we make that some one else's problem if they want to update it (eg pingou because he didn't check into gobby) 18:21:50 <pingou> smooge: I am actually :) 18:21:55 <nirik> well, from 0.12 to 1.0 is not much: http://trac.edgewall.org/wiki/TracUpgrade#to1.0 18:22:12 <nirik> from 1.0 to 1.2? we don't know. they were talking about releasing 1.2 later this year sometime and try and do yearly releases. 18:23:12 <oddshocks> The devel/stable numbering system sounds odd to me :P 18:23:13 <nirik> so, I think it's ok to put 1.0 in epel7 as 'trac' and then if 1.2 is too much pain, we can make trac12, but hopefully we can just update a 18:23:40 <smooge> ok sorry to sidetrack this. I just wanted to get an idea of me to package it as track012 18:24:13 <nirik> smooge: yeah, I think the only reason we would do that is if they were going to keep maintaining it... which it sounds like... they are not. 18:24:26 <nirik> oddshocks: the kernel used to do that, but then abandoned it. ;) 18:24:36 <smooge> of course not 18:24:45 * oddshocks TIL 18:25:09 <nirik> #topic Learn about.... "the-new-hotness" 18:25:14 <nirik> threebean: you're up. ;) 18:25:16 <threebean> yay 18:25:33 <threebean> So, 'the-new-hotness' is one of the three parts of our new upstream release monitoring setup 18:25:45 <threebean> The project page is here https://github.com/fedora-infra/the-new-hotness/ 18:26:20 <threebean> And, in short, it's the "Fedora-specific" cohort to the "distro-agnostic" release-monitoring.org (called anitya) 18:26:48 <pingou> \ó/ 18:26:56 <threebean> the-new-hotness, like the fedbadges backend and the notifications backend, is a fedmsg 'consumer' plugin that runs as a part of the fedmsg-hub daemon. 18:27:08 * oddshocks thinks threebean just likes saying "the-new-hotness" 18:27:12 <threebean> I do 18:27:17 <oddshocks> :) 18:27:20 <threebean> It is.. fun. :) 18:27:28 <pbrobinson> :-P 18:27:29 <nirik> if something else hot comes along will this be the old hotness? sorry 18:27:40 <threebean> I can't wait.. 18:27:40 <nirik> :) 18:27:43 <dgilmore> nirik: the older new hotness 18:28:08 <threebean> at its core, this thing takes care of four distinct tasks: https://github.com/fedora-infra/the-new-hotness/blob/develop/hotness/consumers.py#L118-L132 18:28:32 <threebean> 1) when anitya notices a new upstream release is available, we try to file a ticket in bugzilla if pkgdb says its okay. 18:28:50 <threebean> if that succeeds, kick off a scratch build if possible and remember the task id. 18:29:16 <threebean> 2) when koji tasks complete, if we were the one to kick them off in the first place, then go followup on the bug we previously filed with the status of that 18:29:41 <threebean> 3) when "real" koji builds complete for rawhide, followup on open bugs if we can find them. 18:30:06 <threebean> 4) when a new package is added to pkgdb, try to add it to anitya with some reasonable defaults. 18:30:20 * mclasen_ considers any automated process that ends in filing a bug to be suspect 18:30:31 * jsmith saw steps 1-3 work today, and was very pleased with the results :-) 18:31:10 <jsmith> mclasen_: In many cases, there's already going to be an open bug saying that a new release version is available -- this just adds to that ticket, if it can find it 18:32:08 <threebean> mclasen_: sure, and we provide a flag in pkgdb to disable this whole process. it is the 'monitoring' tab on the upper right-hand side of the package view page. 18:32:11 <nirik> Ideally it would be nice if we could use our staging koji for the scratch builds... but we need to make that work again and add some arm builders to it first 18:33:09 <lmacken> #4, awesome. 18:33:22 <dgilmore> nirik: it would be kinda nice for it to commit to staging git and lookaside and do a real build 18:33:33 <threebean> I'd add that there are some dev instructions in the README and defaults that point at https://partner-bugzilla.redhat.com, so you can hack on it without fear of accidentally filing real bugs on the production bugzilla. https://raw.githubusercontent.com/fedora-infra/the-new-hotness/develop/README.rst 18:33:39 <dgilmore> nirik: then we will have content to test composes in stage 18:33:42 <nirik> dgilmore: yeah, even better 18:34:23 <dgilmore> the way we setup staging its easy to get builds 18:34:27 <nirik> or perhaps it could do both... stg stuff and the current prod stuff... most people wouldn't need to care about the stg part, but we could use that for test composes, etc 18:34:31 <dgilmore> but there is nothing for doing composing 18:35:26 <dgilmore> anyway just an idea 18:35:57 <threebean> yeah, auto-committing to dist-git (staging or not) might be sticky. 18:36:01 <pingou> threebean: did we turn off the stg instance? 18:36:06 <pingou> of the-new-hotness 18:36:19 <threebean> in that, we still rely on packagers to review the content of new upstream releases. what if the license changes? what if new non-free content is included, etc.. 18:36:37 <dgilmore> threebean: sure 18:36:52 <threebean> pingou: it is technically still running, it is just not listening to the production messages from anitya anymore.. and is therefore not doing anything. 18:37:02 <pingou> ok 18:37:52 <threebean> re-establishing that link would just involve uncommenting this block https://infrastructure.fedoraproject.org/cgit/ansible.git/tree/roles/hotness/templates/hotness.py#n15 18:38:19 * langdon just came here to say "dock-ah" 18:38:24 <nirik> ha. 18:38:55 <smooge> sorry not enough walsh in it 18:39:01 <smooge> please try again 18:39:15 <nirik> threebean: some more thoughts (might be too hard): could we have it add new stuff to packages? instead of the packages app regening? or I guess that could be done by revamping packages... 18:39:47 <threebean> yeah, that might start clustering too many varied responsibilities in this one service (imo) 18:39:54 <pingou> +1 there 18:39:55 * nirik nods 18:40:03 <threebean> we could rename it from the-new-hotness to the-do-all-the-things 18:40:13 <threebean> :P 18:40:18 <nirik> it's the new * 18:40:20 <pingou> but threebean were pondering how we could revamp packages and update it via fedmsg 18:40:21 <threebean> ha 18:40:24 <pingou> 42? 18:40:24 * threebean nods 18:40:36 <threebean> yeah, revamped fedora-packages should probably work in a similar way 18:40:36 <oddshocks> It's the Unix philosophy, after all, to create a program to do everything! 18:40:41 <pbrobinson> or all-the-shit ;-) 18:41:02 <threebean> we do have some open bugs, if people are interesting in hacking on it: https://github.com/fedora-infra/the-new-hotness/issues/ 18:41:03 <nirik> anyhow, thanks threebean. Anything else on this app? 18:41:13 <nirik> cool, 18:41:21 <threebean> this one is particularly nasty, and needs fixed sooner than later https://github.com/fedora-infra/the-new-hotness/issues/29 18:41:32 <pingou> http://fedoraproject.org/easyfix/ we have some on github 18:41:44 <threebean> i.e., we can report to packagers that the scratch build is all good, when really it mistakenly is a rebuild of an old version. 18:41:59 <nirik> yeah, nasty corner case 18:42:06 <threebean> (that's all I have) 18:42:18 <pingou> threebean: we could do a fedpkg source just after getting the spec file and see if it is different from the once we get after updating the spec 18:42:37 <pingou> or faster: check the content of the sources file 18:43:01 <nirik> well, the sources will have the old/current one always right? 18:43:01 <pingou> and see if the new sources are different from the content of the current sources file 18:43:05 <threebean> pingou: +1. that's the best approach I've got too. a hash of the sources. 18:43:09 <nirik> yeah. 18:43:34 <threebean> as mentioned in the ticket, github tarballs always have different hash sums, since they're generated on demand. which.. messes all that up. 18:43:44 <nirik> ah github, ;( 18:43:57 <threebean> a corner case for the corner case. 18:44:02 <puiterwijk> threebean: it gets better... they even have different sizes, though the content is the same! :D 18:44:06 <nirik> sharp corner. ;) 18:44:25 <nirik> reproducability? naw. don't need that 18:44:32 <pingou> threebean: ignore github ? :) 18:44:39 <nirik> I guess you could unpack it and diff -Nur ? :) 18:44:48 * threebean notes that. 18:45:44 <nirik> big hassle tho. I guess only for github 18:45:57 <nirik> ok, on to... 18:46:01 <nirik> #topic Open Floor 18:46:19 <dgilmore> nirik: I have one thing 18:46:19 <nirik> the floor is open. Just like Fedora and our default. :) 18:46:34 <nirik> go dgilmore 18:46:51 <dgilmore> I have built a docker base image for arm 18:47:16 <dgilmore> and as the upstream docker registry does not support arches 18:47:22 <nirik> :( 18:49:04 <dgilmore> I am thinking of setting up a registry 18:49:05 <nirik> you are suggesting we run a registry? 18:49:10 <dgilmore> since its all open source 18:49:20 <nirik> ok. Not at all sure whats involved there. I think it's actually a flask app 18:49:23 <dgilmore> I think it would be nice for use to do so 18:49:42 <dgilmore> but only for arm and maybe other arches if teh secondaries get support 18:49:59 <nirik> is there any plan for i686 ones? 18:50:09 <dgilmore> I probably can not commit to maintaining it. but maybe we can get someone to 18:50:16 <dgilmore> nirik: I think we shoudl 18:50:18 <dgilmore> should 18:50:44 <dgilmore> maybe we support x86_64 also, but still push things to upstream also 18:50:44 <nirik> ok. I'd say bring it up on the list and we can see if we can find someone to investigate it and propose a RFR, etc. 18:50:50 * nirik nods. yeah. 18:50:54 <dgilmore> It needs some investigation 18:51:01 <nirik> doesn't upstream have some other weird stuff like no signing? 18:51:05 <dgilmore> just wanted to put the idea out there 18:51:20 <dgilmore> not really sure 18:51:51 <dgilmore> upstream is kinda weird 18:51:58 <dgilmore> and we should be part of it 18:52:07 <nirik> ok. cool. 18:52:15 <dgilmore> and maybe us doing something will get them to be better 18:52:18 <nirik> #info we should investigate doing a docker registery. 18:52:42 <threebean> Unrelated question: When is the alpha freeze expected to lift? 18:52:57 <jsmith> threebean: When it's fully baked?!? 18:53:00 * jsmith ducks and hides 18:53:02 <langdon> nirik, (et al) .. envs&stacks is having much discussion about docker-build & docker-reg .. just fyi 18:53:30 <nirik> threebean: well, next wed if we ship next tuesday, but thats still up in the air. ;( 18:53:48 <nirik> langdon: cool. Good to know. Perhaps someone from there could tell us whats involved in running a registery? 18:54:20 <nirik> ok, anything else open floor wise? 18:54:28 <pingou> I'll be afk next week 18:54:32 <pingou> most of the week 18:54:47 <pingou> available by phone and I'll probably check my emails once or twice :) 18:54:48 <nirik> cool. :) enjoy. 18:54:52 <threebean> pingou: enjoy! :) 18:54:57 <pingou> will do :) 18:55:08 <langdon> nirik, well.. i think there are a lot of considerations on the 'what does it do" and "what is it for" ... as well as the "running it" 18:55:50 <nirik> langdon: indeed. we definitely want stakeholders involved. I mean we could set it up perhaps and let them decide other parts, but sometimes setting up feeds into what you can or can't do after it's setup 18:56:03 <langdon> nirik, right 18:56:34 <nirik> cool. 18:56:41 <nirik> ok, thanks for coming everyone! 18:56:43 <dgilmore> langdon: well we would want to do something to support what upstream doesn't 18:56:43 <nirik> #endmeeting