14:36:53 #startmeeting RELENG (2016-07-18) 14:36:53 Meeting started Mon Jul 18 14:36:53 2016 UTC. The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:36:53 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:36:53 The meeting name has been set to 'releng_(2016-07-18)' 14:36:53 #meetingname releng 14:36:54 #chair dgilmore nirik tyll sharkcz bochecha masta pbrobinson pingou maxamillion 14:36:54 The meeting name has been set to 'releng' 14:36:54 Current chairs: bochecha dgilmore masta maxamillion nirik pbrobinson pingou sharkcz tyll 14:36:56 #topic init process 14:37:21 #chair mboddu 14:37:21 Current chairs: bochecha dgilmore masta maxamillion mboddu nirik pbrobinson pingou sharkcz tyll 14:37:29 * sharkcz is here 14:37:34 morning 14:37:41 * pbrobinson o/ 14:40:41 * masta is here 14:40:58 #topic #6420 Hypothetical maximum speed mirroring 14:41:05 https://fedorahosted.org/rel-eng/ticket/6420 14:41:18 not sure we have much else to say 14:41:51 there was teh one reported issue where it seems that the filielist file was pulled while being updated 14:42:38 other than that we need to ensure that the filelist files are updated appropriately for each rsync module 14:42:47 I guess that file is written to in place 14:43:15 thats been fixed. 14:43:40 there's now a EOD marker at the end, so if you pull the file and it's not there, you know it's incomplete. 14:45:23 #info issue with pulling file while being updated has been addressed 14:45:51 have we identified all the areas that write to /pub? 14:45:55 do we need to? 14:46:44 it would be nice to know 14:47:06 I think we have... the only one that needs updating is the secondary stuff 14:47:39 #info secondary update processes need updating to update the filelist 14:47:41 rawhide, branched composes, bodhi for epel/fedora updates, secondary01 for alt stuff, seconday 14:49:30 alt stuff is written to from more places than secondary01 14:49:49 compose-x86-01 at the least also 14:50:05 yeah, it's a tough one because it's not all script driven 14:50:14 should we have some kind of messaging and a service that updates based on recieveing messages 14:50:16 the kernel-nodebug, etc could update anytime 14:50:23 yep 14:50:31 right now it's on a hourly cron 14:50:41 which isn't ideal 14:51:03 it is not 14:51:20 do we only update it if it changes? 14:51:29 do we publish messages when it is updated? 14:51:53 yes, only if something changes. 14:51:59 no, but we should do a fedmsg for it 14:52:26 we just need to pick a name/message and add it to the scripts 14:53:18 okay 14:53:42 it would be nice to have a message, the mirrors could then listen for it and pull when something changes 14:53:49 yep. 14:54:01 alt might need certs... but composers and bodhi have em already 14:54:15 they probably all need certs 14:54:33 since we probably want them for a mirror user or some such 14:54:39 or ftpsync 14:54:56 #info need to get fedmsg integration 14:55:01 yeah, I was going to use ftpsync/the same whatever sends the 'rsynced updates contents' one 14:55:09 but I haven't looked closely. 14:55:23 nirik: thats releng 14:55:31 i would rather not use that cert 14:55:52 actually I think branched/rawhide is releng and updates is bodhi 14:56:15 we can sort it out. perhaps file a ticket to discuss names/what cert out of meeting? 14:56:21 sure 14:56:28 lets move on for now 14:56:33 #topic #6437 Update documentation 14:56:43 https://fedorahosted.org/rel-eng/ticket/6437 14:57:03 mboddu: did you manage to get anywhere with adding the git hash of some other identifier? 14:58:41 yep, I figured a way to use hash for versioning 14:59:41 cool 14:59:49 and also I got the code which generates the docs on pagure, I will change it so that it can pick up the version # directly from the hash 15:00:26 first thing first, we need to identify what are all the changes and start documenting them, we can do it in our KT process 15:01:19 #info mboddu to update the docs to use git hash for versioning 15:01:28 mboddu: they all need review and updating 15:01:39 I do not think we need to indentify specifc ones 15:01:47 as we find issues we need to fix them 15:01:54 and add new docs for defiencies 15:02:28 KT- knowledge transfer 15:03:11 sure thing 15:05:43 #info we assume all docs need updated and as we go though things will update them 15:06:38 #info need a way to ensure all documents are reviewed 15:07:43 lets move on 15:07:44 #topic Secondary Architectures updates 15:07:45 #topic Secondary Architectures update - ppc 15:07:52 pbrobinson: how is ppc this week? 15:09:04 builds are blocked by 1356919 (emacs vs glibc malloc), but otherwise no problems 15:09:47 yea, not bad, there's a few minor issues 15:10:04 the one that was mentioned by sharkcz 15:10:45 #info build blocked by rhbz#1356919 15:10:52 cool 15:11:06 #topic Secondary Architectures update - s390 15:11:15 how is s390 sharkcz ? 15:11:49 better than ppc :-) still just hours behind primary from the builds point of view 15:11:59 cool 15:12:06 #info no blockers currently 15:12:08 and the introduction of s390x golang should bring us even closer 15:12:38 there was a side tag requested for golang, I created it this morning 15:12:45 #info s390x introduced golang so will remove more differences 15:12:52 okay 15:13:57 #topic Secondary Architectures update - arm 15:14:04 hows aarch64 pbrobinson ? 15:14:16 builds are mostly up to date 15:14:25 the composes were failing for the last week or so 15:14:47 I submitted a lorax pull request that should fix that shortly 15:15:06 okay 15:15:26 rawhide in general was failing due to lorax bugs in the last week 15:16:04 yea, not surprised there really looking at some of the recent commits/changes 15:17:29 #info composes have been failing due to lorax bug 15:17:46 #topic f25 status 15:17:49 moonshot is moving forward, discovered a firmware bug 15:17:56 #undo 15:17:56 Removing item from minutes: 15:17:58 awaiting confirmation around it 15:18:01 pbrobinson: fun 15:18:37 hoping to have confirmation/fix shortly 15:18:43 that's it for aarch64 15:18:52 cool 15:18:54 #topic f25 status 15:19:06 so lets quickly go over f25 15:19:21 https://fedoraproject.org/wiki/Releases/25/Schedule 15:20:42 branching in in 8 days 15:21:07 * nirik nods. 15:22:06 mboddu: will be handling the mass branching 15:22:17 pbrobinson: will be handling enabling branched composes 15:22:30 cool. Happy to help out with anything if anyone gets stuck on anything. 15:22:34 * dgilmore will be on PTO without computer access 15:22:42 you out all next week? 15:22:58 I will be most likely working monday 15:23:07 I'm around but will be on GMT+8 15:23:16 but I may have to take PTO to look after Mason 15:23:30 thanks nirik, I hope I wont face any issues 15:23:48 yeah, hopefully it will go smoothly. 15:23:54 pingu offered to help with pkgdb side of things 15:24:03 mboddu: co-ordination is key, and don't be afraid to ask questions 15:24:18 hopefully he fixed the bug that caused the pkgdb side of things to take about 10 hours to run 15:24:29 thanks pbrobinson 15:24:55 dunno, we should ask. ;) 15:25:01 he said he had 15:25:09 but not sure how well its tested 15:25:59 any releng features for f25 need to be deployed by monday next week 15:26:16 yeah, it's hard to test things that happen only 2x/year 15:27:48 maxamillion: so we have a week to ensure OSBS is deployed and whatever we are going to deliver in f25 is done 15:27:54 nirik: indeed 15:28:13 it does mean at least that it is one of the better documented pieces 15:28:39 dgilmore: yeah, there's still a few items on the list but we're mostly done 15:29:20 * dgilmore does not want us adding new features after alpha 15:30:08 dgilmore: it's not features, just finishing up some items like the BZ component and fedpkg blacklisting 15:30:17 maxamillion: okay 15:30:17 FYI, I have been updating builders slowly. arm01/arm02 are done. Going to do arm04 today and some buildvm's... then will hopefully finish buildvm's and buildhw in the next few days. 15:30:22 dgilmore: it's still in the same scope, we're not adding anything that we didn't originally say we would 15:30:38 * dgilmore wants to have all the deliverables done and defined for alpha 15:31:06 maxamillion: sure. we just need to have it done 15:31:20 dgilmore: indeed we do 15:31:23 nirik: is that updating them to F-24? 15:31:27 along with all the things 15:31:27 yes 15:31:39 seems to be working fine. I updated stg a few weeks ago... 15:31:44 cool, will start to do the same for ppc/aarch64 this week then too 15:32:58 #info nirik updating builders to Fedora 24 15:33:32 is F-22 last push today or tomorrow? 15:33:58 it is today 15:34:04 tomorrow is eol 15:34:25 nirik: right 15:34:36 the last push has to go out before then 15:34:39 so it will be today 15:34:42 I'm starting a push in a bit. You want me to do another f22 one after this one finishes? or just call it dead? 15:34:49 cool, once that's done I'll close out secondaries too 15:34:50 up to you 15:35:08 will see when it finishes and such... 15:38:05 anything else on f25? 15:38:30 #info nirik to do the last push for f22 soon 15:38:38 * dgilmore has nothing 15:38:53 #topic Open Floor 15:39:06 doea anyone have naything they want to raise? 15:39:11 or call it a day? 15:39:13 I was wondering about https://fedorahosted.org/rel-eng/ticket/6432 15:39:47 they will have to be done this week 15:40:00 * nirik nods 15:40:02 I am not sure that they are ready for it 15:41:04 dgilmore: yes, they are, I was going to use the partial script to push builds for the list tomorrow morning 15:41:07 well, I think rpm is ready. 15:41:26 this would done in a side tag? or ? 15:41:37 yep 15:41:39 pbrobinson: okay, to me at least its not clear that things were ready for the rebuild to happen 15:41:47 it will have to be a side tag 15:42:02 dgilmore: they asked about it on the internal rel-eng channel a few hours ago 15:42:03 * dgilmore thinks that it needs some better communication 15:42:14 pbrobinson: thats not very useful 15:42:32 not anyones fault 15:42:46 dgilmore: I don't disagree, I directed them in future to the external channel and they did update the ticket 15:42:48 just not very useful to ask in the wrong places 15:42:58 anyway 15:43:05 * nirik has nothing more off hand. 15:43:14 #info pbrobinson to do partial mass rebuild for python 15:43:19 dgilmore: you going to do the eol on f22 tomorrow? or want someone else to via the SOP? 15:43:35 pbrobinson: you going to do it on compose-x86-01 with the releng cert? 15:43:46 nirik: I was going to 15:44:05 ok. fine with me. 15:44:11 I will do it with mboddu 15:44:15 there's also the golang 1.7 rebuild that needs to be done by branch too 15:44:22 dgilmore: yes 15:44:42 we should send out an email today stating there is a week for any side tags to be done 15:45:21 I think it's currently just those two still open, haskell is merged in and done 15:48:16 yeah it is 15:48:20 just had a look 15:48:34 anything else? 15:48:53 oh, FYI, on the new buildhw's... we ran into a snag with rack sizes (again). Hopefully we will get them up in the next few weeks tho... 15:54:54 okay 15:54:59 what is the issue 15:56:39 they are too long to fit in our normal racks, so we had to order a longer one. 15:57:40 fun 15:57:46 what is the hardware? 15:58:15 nirik: sounds like the issue with the moonshot 15:58:15 they are dell ... uh, don't recall the model, but bladecenters... 8 machines per x 2 (16 total) 15:58:27 pbrobinson: yep, its the exact same issue. 15:58:38 my bad for not doublechecking it. ;( 15:58:48 anyhow, we will get em up and running. 15:59:59 and retire the old buildhw's finally. 16:02:07 okay 16:04:32 anything else? 16:04:59 #endmeeting