16:56:09 #startmeeting RELENG (2015-12-14) 16:56:09 Meeting started Mon Dec 14 16:56:09 2015 UTC. The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:56:09 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:56:09 The meeting name has been set to 'releng_(2015-12-14)' 16:56:21 #meetingname releng 16:56:21 The meeting name has been set to 'releng' 16:56:21 #chair dgilmore nirik tyll sharkcz bochecha masta pbrobinson pingou maxamillion 16:56:21 Current chairs: bochecha dgilmore masta maxamillion nirik pbrobinson pingou sharkcz tyll 16:56:24 #topic init process 16:56:26 * masta is here 16:56:51 .hello maxamillion 16:56:52 maxamillion: maxamillion 'Adam Miller' 16:57:20 morning 16:59:43 so we have no tickets for the meeting 16:59:59 epel7 is now building ppc64le by default 17:00:19 nice 17:00:29 there is bodhi work that needs to happen. but outside of updating the mash configs I am not sure what we need to do 17:01:25 yeah, could be some whitelist of arches to make sure ppc64le is there 17:01:57 may be some issues with mm 17:02:10 but not sure. hopefully it just picks it up and runs with it 17:02:27 * nirik nods 17:03:03 is there going to be a mass rebuild for ppc64le or has it been shadowing to pick up all the epel stuff already? 17:03:44 maxamillion: it was all scratch built with the same nvrs and imported 17:04:27 rgr 17:04:47 dgilmore: can those just be tagged/imported so that it goes out to the repos? 17:04:55 maxamillion: done already 17:05:16 we just need to add ppc64le to the mash configs, make sure they are all signed and do a push 17:05:19 ahhh ok 17:05:24 cool cool 17:05:58 we may not need to change anything in bodhi 17:06:03 its a bit up in teh air 17:06:37 is there somewhere we should file a ticket? or someone we should follow up with to make sure? 17:06:44 * sharkcz is somewhat here 17:07:08 maxamillion: I need to get with lmacken and see 17:07:13 which reminds me 17:07:18 dgilmore: rgr 17:07:29 we need to make sure we have docs from lmacken on how to do everything with bodhi 17:08:05 hopefully there are not too many exclusive/exclude-arch packages in epel7 that drop LE. Would be nice to have good parity. 17:08:19 "do everything" ... what specifically? (legitimately curious, I'd like to try and start writing down RelEng processes because I know there's plenty I'm still not familiar with and would love to have references for my use as well as for potential newcomers into releng space) 17:08:40 maxamillion: adding releases, doing teh eol process, 17:08:48 dgilmore: gotchya 17:08:55 maxamillion: basically everything that we need to do 17:09:37 dgilmore: right, but if someone asked me "what is everything that Rel Eng needs to do in bodhi?" I wouldn't know how to answer that question properly, and I'd like to have us as a group have those kinds of things written down in the releng docs 17:10:10 maxamillion: honestly I am not 100% sure 17:10:16 dgilmore: ah, rgr 17:10:19 which is paert of the problem 17:10:22 well we should sort that out :D 17:10:29 and why we need docs 17:10:36 dgilmore: what bodhi work needs to happen to add another arch? 17:10:45 lmacken: we do not know 17:11:01 lmacken: we know we need to update the mash configs 17:11:03 I am pretty sure the answer is: none 17:11:11 outside of that we have no clue 17:11:13 yeah, bodhi doesn't care about arches 17:11:34 only koji tags, so I'm almost positive it's just a matter of mash config changes 17:11:34 lmacken: that is great news then =) 17:11:46 lmacken: rocking +1 17:11:49 lmacken: okay. hopefully it does not cause things to puke by adding an arch 17:12:13 we can test it with updates-testing any time 17:12:21 lmacken: epel-testing 17:12:23 :P 17:15:17 so I guess this week Peter is on push duty, maybe we can get LE in the configs before he starts pushing (presumably) later today? 17:15:49 or perhaps we plan it out more after scoping? 17:15:50 masta: I will work with him on it 17:15:55 he already pushed stuff this morning. ;) 17:15:58 not sure where the signing stands 17:16:29 but it seems we just need to update the mash configs 17:16:46 lmacken: but we would really appreciate docs on how to do all the admin tasks 17:16:57 release management etc 17:17:05 dgilmore: yeah, I have a half-modified SOP from the last EOL that I need to wrap up 17:17:12 * lmacken makes a note to finish that this week 17:17:41 lmacken: thanks 17:17:57 #action lmacken to finish updating the sops for bodhi 17:19:46 #topic updates pushing 17:19:55 so pbrobinson is on duty this week 17:19:59 * nirik nods. 17:20:03 I am on duty next week 17:20:11 I'm going to be around for the holidays, happy to take over for anyone or whatever 17:20:13 but we have thursday and Friday off work 17:20:32 I will probably push on thusday but not Friday 17:20:48 not going to do anything on Chirstmas day 17:20:58 except open presents? ;) 17:21:00 we then have shutdown at Red Hat 17:21:09 nirik: well that and drink eggnog 17:21:20 masta: is on duty for the shutdown week 17:21:22 mmmmm... need to get some eggnog 17:21:52 as noted, I should be around, so I am happy to take over for anyone... masta: you want me to do that week? 17:22:03 * nirik doesn't expect there will be many updates really tho 17:22:21 yes, I plan to push during shutdown as best effort. That means my family would have to massively guilt me into not pushing, which I expect to resist. 17:22:26 I am wondering if for shutdown we do a day each, or just say no updates? 17:22:30 or something else 17:23:14 how about we look and if there's not that many we just defer to the next day (or no security updates or something) 17:23:36 from experience, the pushes are minimal during shutdown. So a few minutes of work signing mostly. 17:23:38 note also, that second week in jan I will be out at phx2, so likely will not want to be doing any pushing that week if thats when I am scheduled 17:24:14 nirik: I can cover you if needed, just let me know. 17:24:23 nirik: you ate the week of the 4th 17:24:29 ok. 17:24:35 are even 17:24:38 that week I should be around. 17:25:25 nirik: cool 17:27:34 #info nirik and masta will do best effort pushing during shutdown 17:27:52 #topic Open Floor 17:28:02 lets skip secondaries this week. 17:28:20 pbrobinson: is not here and sharkcz said he mostly wasn't here 17:28:33 does anyone have anything they want to discuss 17:28:34 I'm hoping to coordinate with pbrobinson this week and migrate arm hub/db... but if not, then we can do it in jan 17:28:47 nirik: :) would be nice 17:28:55 that replacement energycard should be arriving tomorrow... 17:29:03 That will be very nice 17:29:12 oh neat, then I guess after that we can buile epel7 for rhelsa =) 17:29:26 masta: maybe 17:30:18 I cannot currently powercyle arm04 due to crappy apc devices. 17:30:33 do we want to adjust what hosts are in what channels any? 17:30:39 or I guess we can do that out of meeting 17:32:23 nirik: we can do it out of meeting 17:32:38 but it would be nice to get a couple of arm systems with ssd 17:32:58 there is a pretty significant slow down on spinning disk 17:34:23 #info we will have a meeting next week, but not the week after, and probably not on the 4th unless we have something that needs discussing 17:34:28 we are getting 15more ssds, but not sure when 17:35:41 that will be nice 17:36:15 nirik: reminds me. I had a request last week for docker base images for arm 17:36:35 outside of the meeting I would like to discuss how we can do it short term 17:36:50 all options pretty much suck 17:37:28 does anyone have anything else? 17:37:48 sure. 17:39:15 I'm trying to help with old tickets. #4084 is really old and I'd like to get rid of it. 17:39:42 took me a while to fully appreciate the problem statement. 17:40:00 might take a bit more thought on how to solve. 17:41:07 I'm hoping it would not be too expensive to add a routine to mash to scope-out what was done in release repos for a multilib, and do the right thing for the scenario durring updates. 17:41:39 that's it for me, eof 17:42:30 masta: thats a really difficult problem to solve 17:42:57 pretty much you are going to have to feed it a second repo to use for resolving deps 17:43:04 dgilmore: I know, that's way it's been open forever and neglected. 17:43:18 but that does remind me 17:43:40 I am working on getting together a koji update with a bunch of bug fixes that we need for infra 17:44:01 i will be pushing out updates to all supported releases also 17:44:06 hopfully this week 17:44:26 nice 17:45:35 excellent 17:47:17 I hit bugs when I tried to do a real image build rather than a scratch one 17:47:34 oh yeah, did you get stg working? 17:47:53 also, is it ok to run ansible on koji/build* in stg? or do you have changes there? 17:49:49 nirik: did not get stg working 17:49:57 imag building there just does not work 17:50:13 humf. we will have to visit that when we get time/energy then 17:50:21 nirik: I think i have changes, but go ahead 17:52:10 nirik: I did have to change the redirects for kojipkgs to not use https 17:52:27 oz/imagefactory apparently do not do ssl 17:52:34 yeah, thats just sad. 17:52:40 but ok 17:52:50 apparently never tested 17:52:54 and it does not work 17:52:55 I am trying to get the delta for ansible changes as close to 0 as I can. 17:53:07 okay 17:53:39 nirik: we use http in prod for all teh koji communications 17:53:50 at teh least we do not redirect it to https 17:53:57 the image tasks use http 17:54:28 that may be some stg difference... 17:55:59 yeah 17:56:16 #info planning a koji update this week 17:56:31 need to be sure to get a fix for https://fedorahosted.org/rel-eng/ticket/6274 in 17:56:32 dgilmore: bold move right before the winter break 17:56:37 maxamillion: :P 17:56:42 +1 DO IT LIVE 17:56:57 maxamillion: I won't be here to fix it ;) 17:57:13 me either ... LET IT BURN 17:57:14 >.> 17:57:18 I suspect we will have a couple of koji updates in teh new year also 17:57:23 likely so 17:57:36 maxamillion: where is osbs at? 17:58:19 dgilmore: nowhere official yet, it will be in osbs.fedoracloud.org when I can get things working and request that DNS pointer, then I'll install the koji plugin in stg koji that will farm off builds to it 17:58:48 dgilmore: right now I have a cloud instance in fedora infra cloud that I'm fighting with making work, then I need to make sure I can duplicate making it work via ansible 17:58:49 maxamillion: hrrrm okay 17:59:05 maxamillion: one of the changes I am bringing into koji is content generators 17:59:15 dgilmore: cool cool 17:59:22 which is what they are going to use internally and not the plugin afaik 17:59:37 maxamillion: we need to make sure we line things up the same 17:59:46 maxamillion: I can setup a base playbook for you and dns, etc... just let me know the name and size... 17:59:49 dgilmore: I was under the impression that the plugin can do a meta-data import with content generators 17:59:59 nirik: that would be awesome, thanks 18:00:12 then you can just add to that 18:00:22 maxamillion: okay, talking to mikem he said the plugin did not use content generators 18:00:26 nirik: let me look at the size, but I'd like to name to be something like osbs.dev.fedoracloud.org or similar (I'm open to suggestion) 18:00:40 could be some confusion there about things as they are so much in flight 18:00:48 our convention on the others is "name-dev.fedorainfracloud.org' 18:00:50 dgilmore: right, I knew it could work without but I thought it was also supporting them ... :/ 18:01:09 nirik: ah, +1 osbs-dev.fedorainfracloud.org would be perfect 18:02:24 maxamillion: guess we need to be sure of what we need 18:02:46 I know I need some pykickstart fixes 18:02:46 nirik: I'm currently using a ms1.medium and considering the limited scope of the dev osbs, I think that should be sufficient 18:02:47 maxamillion: ok, can talk outside meeting on sizes, etc. 18:02:51 ok. 18:02:58 I can get that setup later today hopefully 18:03:02 and some other fixes for ssl issues and image building issues 18:03:05 nirik: awesome, greatly appreciated 18:03:35 okay lets be done for today 18:03:38 +1 18:03:42 #endmeeting