18:02:30 #startmeeting FESCO (2015-09-23) 18:02:30 Meeting started Wed Sep 23 18:02:30 2015 UTC. The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:02:30 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:02:30 #meetingname fesco 18:02:30 The meeting name has been set to 'fesco' 18:02:31 #chair ajax dgilmore hguemar jwb nirik paragan rishi thozza sgallagh 18:02:31 Current chairs: ajax dgilmore hguemar jwb nirik paragan rishi sgallagh thozza 18:02:31 #topic init process 18:02:34 .hello jkurik 18:02:35 jkurik: jkurik 'Jan Kurik' 18:02:36 hi all 18:02:42 sorry for the late agenda 18:02:45 hello teenage america 18:02:53 Hi 18:03:03 morning 18:03:07 hi 18:03:09 .fasinfo hguemar 18:03:10 number80: User: hguemar, Name: Haïkel Guémar, email: karlthered@gmail.com, Creation: 2006-07-18, IRC Nick: number80, Timezone: Europe/Paris, Locale: en, GPG key ID: 26613DF3, Status: active 18:03:13 jkurik: since you are here we have an item we need removed from the schedule. legal has told me that we no longer need to file for export compliance 18:03:13 number80: Approved Groups: +gitovirt cla_fedora cla_done fedorabugs ambassadors cla_fpca gitbeefymiracle +packager python-sig provenpackager 18:04:03 lets wait for a few more people 18:04:08 dgilmore: I am in the process of reworking the schedule, so I am just collecting this type of info 18:04:19 thanks for letting me know 18:05:57 I think we have qurom 18:06:10 * jwb is 1/2 here 18:06:46 #topic #1427 List of release blocking deliverables 18:06:46 .fesco 1427 18:06:47 https://fedorahosted.org/fesco/ticket/1427 18:07:36 dgilmore: #1427 (List of release blocking deliverables) – FESCo - https://fedorahosted.org/fesco/ticket/1427 18:07:43 I tried to make a more clear table of these. 18:09:14 I am +1 (even I am not a FESCo member) to use nirik's page without restricting access to it 18:09:31 well, if people are ok with the table I can fold it into the existing page... 18:09:59 and... I would suggest we also make a f24 one now (which we can adjust up to f24 alpha) 18:10:13 I feel like perhaps a better place for the data is a git repo on pagure 18:10:31 jkurik: everyone is free to express their opinion here :) 18:10:44 I would like to be able to write some tooling to verify that a compose has in it everything that is supposed to be there 18:10:51 dgilmore: is that a proposal? 18:11:02 number80: I guess 18:11:15 I don't mind using pagure, as people could submit easily pull requests 18:11:28 already voted in ticket, +1 to nirik's changes https://fedoraproject.org/wiki/User:Kevin/Fedora23_release_deliverables 18:11:34 ideally the data is presented in a human readable form. but can also be used by tooling to verify a compose 18:11:38 +1 to dgilmore proposal of using pagure to manage data 18:12:02 dgilmore: I don't mind the format as long as it's not xml :) 18:12:05 sure, that seems fine but until we have it... (I suspect there may be formatting bikeshedding and such) 18:12:19 number80: I think json would probably make sense 18:12:21 * jkurik is -1 for XML as well :) 18:12:55 as a start what nirik has done will work 18:13:08 but it will mean manual verification 18:13:20 I guess I just signed up to do some work 18:13:28 how about we use the wiki now and when we have a json one we switch to it. 18:13:40 that is fine 18:13:44 nirik: you were faster than me +1 18:15:17 we have +3 for nirik proposal 18:15:29 thozza, paragan, jwb? 18:15:29 +1 from me 18:15:30 so for F23 & F24 we will have wiki page and during F24 we will convert the wiki to json/pagure. Right ? 18:15:37 number80, already +1 18:15:43 paragan: sorry :) 18:15:49 np 18:15:54 Proposal #agreed: we will use nirik's wiki page until we can egt something in a better format that can be used to visualise and by tools to verify a compose 18:16:02 I was thinking of extending template for Changes with a question whether list of deliverables will be affected by a Change. Does it make sense ? 18:16:26 jkurik: I thought that was already part of the change process 18:16:29 I'll update wiki pages in the next few days and note it in the ticket. 18:16:35 thanks 18:17:05 jkurik: if it's not explicit already, just do it (there's no harm in adding the question) 18:17:08 dgilmore: not explicitly; there is a question whether an RCM work is needed 18:17:42 ok, I will explicitly mention it there 18:17:49 jkurik: okay, there really should be a clear list of new deliverables as part of a change 18:18:34 at least in the form of livecd's, cloud images, disk images, install media, layered image, docker base image, etc 18:19:01 or some new format 18:19:15 which requires much more coordination and tooling work 18:20:39 I am really not quite sure where we stand here and what exactly is our current proposal 18:21:14 I believe the proposal you wrote above is still valid 18:21:29 yes its still valid and can be approved 18:21:36 okay 18:21:48 #agreed: we will use nirik's wiki page until we can egt something in a better format that can be used to visualise and by tools to verify a compose 18:21:54 just going with that 18:21:56 sure 18:22:45 #topic #1452 F23 System Wide Change: Two Week Atomic 18:22:46 .fesco 1452 18:22:46 https://fedorahosted.org/fesco/ticket/1452 18:22:47 dgilmore: #1452 (F23 System Wide Change: Two Week Atomic) – FESCo - https://fedorahosted.org/fesco/ticket/1452 18:23:24 the automated testing is not ready yet 18:23:29 but is being worked on 18:23:45 I see mattdm's proposal looks good here 18:24:10 I guess I'm ok with the plan as listed. 18:24:22 the same 18:24:40 we produce f22 trees nightly for this today 18:25:06 so they need testing and f23 added? 18:25:21 automated testing should be coming soon (if that's what kushal and sayan were working on these last days) 18:25:28 well at some point we will switch to f23 based nightly composes 18:25:52 there was miscomunication over what was supposed to be tested 18:25:54 it looks reasonable 18:26:01 so some of it is not being done 18:26:21 we should be on production next week 18:26:28 the automated testing part. 18:26:57 historically we would have said this has to wait until f24 as it has missed the boat for f23 18:27:19 so i am going to vote 0 here as I feel my view may be a bit tainted 18:28:01 likely it will be ready sometime shortly after f23 is released 18:29:00 any other votes? where are we? 18:29:31 looks like we have 3 +1 nirik, paragan, number80 and my 0 18:29:44 and +1 more from thozza ? 18:30:00 i guess thozza's looks good could be a +1 18:30:08 looks reasonable 18:30:09 +1 18:30:10 right 18:30:12 not looks good 18:30:17 \o/ 18:30:42 well, I'd rather not postpone this to next meeting as it's an important topic for Cloud WG 18:30:53 #agreed mattdm's proposal is accepted (5,0,1) 18:31:24 #topic #1480 F24 Schedule 18:31:25 .fesco 1480 18:31:25 https://fedorahosted.org/fesco/ticket/1480 18:31:26 dgilmore: #1480 (F24 Schedule) – FESCo - https://fedorahosted.org/fesco/ticket/1480 18:32:25 I would like to have reviewed these milestones https://fedoraproject.org/wiki/Releases/24/Schedule 18:32:29 my only concern with it is 2015-12-29 Change Checkpoint: Proposal submission deadline (System Wide Changes) 18:32:32 well, I don't like the proposals deadline before the holidays. 18:32:37 that is when Red Hat is on shut down 18:32:54 I think we need to aim for the end of May not the start 18:33:10 and give a little time in the new year for proposals to come in 18:33:13 I can move it for one week to have it in first week of January 18:33:27 often folks rest up and have time to think over the holidays... 18:33:41 so then they might well come up with some change they had time to ponder on 18:33:43 also same date for Beta freeze and Beta release 18:33:49 i'd take either jan 5 or jan 12 for the swc deadline 18:33:51 +1 for delaying it 18:33:56 jkurik: what not release on may 24 or may 31? 18:34:02 why not 18:34:06 That would be really good. 18:34:51 paragan: yeah, that seems like a typo for beta release there. ;) 18:34:52 paragan: indeed, that is not possible 18:34:52 ok, no problem to delay the swc deadline to jan 5 or jan 12 18:35:08 12 for me, as 5 is new year eve in many eastern europe countries 18:35:09 jkurik: i would go for the 12th over the 5th 18:35:27 * nirik nods. 18:35:27 dgilmore: ok 18:35:29 number80: i think the 5th is too near to new years for the whole world 18:35:38 people will still be getting back into work mode 18:35:45 yeah 18:36:45 proposal: jkurik to adjust the schedule to have change submission deadline (system wide changes) be on Jan 12th 18:36:58 +1 18:37:03 +1 18:37:09 +1 18:37:10 sure. +1 18:37:23 do we want to revist next week? or just assume everything lines up ok after that? 18:37:26 I am just looking at the dates for Beta and it seems like a typo ... let me correct it ... 18:37:31 I think we should revist 18:37:44 yes we should revisit with new changes 18:37:55 +1 to my proposal 18:38:10 agreed: jkurik to adjust the schedule to have change submission deadline (system wide changes) be on Jan 12th (5,0) 18:38:34 ack for revisiting 18:38:35 #info we will revisit the schedule next week with the changes in place 18:38:38 Beta dates corrected 18:38:53 #topic #1481 Decision on distro-sync during upgrades with dnf-plugin-system- 18:38:56 upgrade 18:38:58 .fesco 1481 18:38:59 #topic #1481 Decision on distro-sync during upgrades with dnf-plugin-system-upgrade 18:39:00 dgilmore: #1481 (Decision on distro-sync during upgrades with dnf-plugin-system-upgrade) – FESCo - https://fedorahosted.org/fesco/ticket/1481 18:39:03 https://fedorahosted.org/fesco/ticket/1481 18:40:08 looks like this ticket is actually proposed for discussion in next week 18:40:39 I personally feel like distro-sync should be used, if you have a fully updated f22 system and a month after release of 23 you upgrade without updates enabled you will likely hit issues that distro-sync will take care of 18:40:55 paragan: yeah, it had the meeting keyword already 18:41:14 ah okay 18:41:16 In my rush i just looked to see if it was new or old business 18:41:19 why would you ahve a fully updated f22 but have updates disabled for an upgrade to f23? 18:41:29 jwb: I do not know 18:41:37 but I can see some people would 18:41:43 doesn't dnf system-upgrade default to whatever is enabled on the current instal? 18:41:59 I don't think so, it has it's own setting 18:42:11 does it default to updates enabled? 18:42:18 because if it doesn't that seems suboptimal 18:42:39 * nirik isn't sure 18:42:46 and yeah, i can see someone maybe being in that situation. i can also see someone manually removing files with 'rm' and then being broken too 18:43:17 The bigger question is 3rd party stuff... 18:43:38 do we remove it, or leave it installed but possibly broken. 18:43:40 well, system-upgrade is kind of broken all around there. i don't think distro-sync is going to help 18:43:44 nirik: that is a bigger issue, but we do not have much control on it 18:44:07 commonbug disable 3rd party repos 18:44:13 well, we could do one of those two things. ;) 18:44:20 I think "dnf distro-sync" offers a safer path than "dnf update" 18:44:26 Southern_Gentlem, that doesn't fix things. that still leaves broken systems 18:44:30 but it is far from perfect of problem free 18:44:47 s/of/or/ 18:44:50 we could also try and present to the user the choice. 18:45:01 * nirik leans toward removing. 18:45:18 i'm opposed to removing 18:45:52 someone installed that stuff for a reason. if they have the choice between "upgrade to the next release but ahve a broken system" or "don't upgrade", they are likely going to chose not to upgrade 18:45:58 how about removing with user ack? ie, 'these packages will be removed: ... ' ok/abort 18:46:18 that's fine 18:46:25 anyhow if this was for next week. perhaps we should invite the authors of the plugin and discuss it more then? 18:46:34 yes 18:46:37 nirik: yeah 18:46:56 +1 for inviting dnf folks 18:47:04 aye 18:47:20 #info ticket had meeting keyword but is worded for furture meeting, FESCo to invite stakeholders to the FESCo meeting next week to discuss 18:47:45 #topic Next week's chair 18:47:53 who wants to run things next week? 18:48:02 i will almost certainly miss next week 18:48:14 I've not in a while, I guess I can. 18:48:24 I'm still in the middle of finishing a release at $DAYJOB 18:48:29 nirik++ 18:48:30 #action nirik to run the next meeting 18:48:43 #topic Open Floor 18:48:45 jwb less brork system and a upgrade path is better than no upgrade path 18:48:56 nirik: want to go over the discussion in #fedora-devel 18:49:01 I had one quick item: 18:49:14 there's a dnf build that landed in todays rawhide that doesn't work. ;( 18:49:26 Southern_Gentlem: you may start a discussion on fedora-devel list (but please keep it civil) 18:49:32 Would fesco be ok if we untagged it so people who didn't update today might be ok? 18:49:59 nirik: as long as releng is ok, yes 18:50:16 nirik, +1 to untag it 18:50:19 the current policy is that we don't untag packages once they have gone out... unless fesco approves. 18:50:27 (which we might look at revisiting) 18:50:33 we probbaly should look at starting a discussion to consider changing our policy on things never going backwards, as a few people in the last 6 months have wanted to do it 18:50:48 nirik: consider this as +1 18:51:04 yeah, honestly for a lot of rawhide corner cases i think it's reasonable to expect "koji download-build" to be able to rescue you 18:51:06 I'm +1 to untagging it. Won't help that much, but wont hurt either. ;) 18:51:38 although tsk tsk how did this go out with no test catching it 18:51:48 ajax: indeed 18:51:53 no idea. It doesn't error, just doesn't do anything. 18:52:09 ajax: let us remain silent about dnf way of doing things :) 18:52:14 this is the second time in recent memory that dnf has broken silently 18:52:17 so, good chance to improve testing. ;) 18:52:28 +1 let's stay positive 18:52:35 any other votes for untagging? how many is that? 18:52:42 nirik: +1 18:52:45 +1 18:52:50 quorum reachead 18:52:51 I think it makes sense 18:53:09 I would rather we do not untag, but I do not feel so strongly about it that i would stand in the way of it happening 18:53:24 #agreed untag the broken dnf build 18:54:17 Does anyone have anything else to bring up 18:54:19 ? 18:54:27 can we ask QA to join the discussion next week? 18:54:36 they may help dnf folks to get some testing 18:54:44 adamw, roshi ^ 18:54:57 number80: the system-upgrade discussion? 18:55:00 yeah 18:55:17 yeah, I can be here for a discussion next week 18:55:24 roshi: thank you :) 18:55:24 number80: FESCo to invite stakeholders to the FESCo meeting next week to discuss to me includes QA as they are a stakeholder 18:55:24 sure, someone's gonna have to poke me though 18:55:28 chances of me remembering are 0 18:55:36 I'll poke you 18:55:42 adamw: well, I'll remind you 18:55:45 * dgilmore pokes adamw with a smelly trout 18:55:50 then, I'm done 18:56:16 if nothing else then lets wrap up 18:56:18 Thank you dgilmore for chairing and everyone else who attended the meeting 18:56:49 3 18:56:58 2 18:57:09 1 18:57:12 #endmeeting