15:33:02 <dgilmore> #startmeeting RELENG (2016-06-13)
15:33:02 <zodbot> Meeting started Mon Jun 13 15:33:02 2016 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:33:02 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:33:02 <zodbot> The meeting name has been set to 'releng_(2016-06-13)'
15:33:02 <dgilmore> #meetingname releng
15:33:02 <zodbot> The meeting name has been set to 'releng'
15:33:02 <dgilmore> #chair dgilmore nirik tyll sharkcz bochecha masta pbrobinson pingou maxamillion
15:33:02 <zodbot> Current chairs: bochecha dgilmore masta maxamillion nirik pbrobinson pingou sharkcz tyll
15:33:05 <dgilmore> #topic init process
15:33:07 <maxamillion> .hello maxamillion
15:33:08 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
15:33:13 * pbrobinson o/
15:33:17 <nirik> .hello kevin
15:33:18 <zodbot> nirik: kevin 'Kevin Fenzi' <kevin@scrye.com>
15:33:18 <nirik> morning
15:33:20 <dgilmore> hey all
15:33:53 * sharkcz is here
15:35:14 <dgilmore> lets get started
15:35:20 <dgilmore> #topic F24 Final
15:35:31 <dgilmore> so we are in final freeze, slipped a week
15:35:45 <dgilmore> we did finally get a RC compose request
15:36:43 <dgilmore> I really think I would like to propose that if we do ot have a RC request by Tuesday 23:59 UTC of the go/no-go week we just slip
15:37:16 <dgilmore> what do others think?
15:37:34 <pbrobinson> that's likely something we have to get accepted by FESCo no? But I agree overall
15:37:59 <nirik> we could... would save a meeting.
15:39:47 <dgilmore> thats my thought
15:39:53 <dgilmore> we push off the meetings
15:40:01 <maxamillion> +1
15:40:18 <dgilmore> at that point we have less than 48 hours to compose and test a release
15:41:17 <dgilmore> but the really late requests for RC's this cycle have made me think we need to have a point where if we do not have a compose we can't pretend like it may ship
15:41:32 <dgilmore> I will put together a proposal to FESCo
15:41:35 <nirik> I don't know that anyone was pretending it could ship
15:42:04 <dgilmore> nirik: having go no go meeting when we have nothing is kinda pretending we could ship something
15:42:05 <nirik> but sure, avoiding meetings where we know what will happen is good.
15:42:21 <pbrobinson> agreed, if we don't have all the blockers with fixes we're only fooling ourselves we can do it and then people try herculean efforts and that's overall unhealthy
15:42:35 <dgilmore> yep
15:42:47 <sgallagh> Well, there's a little leeway on the blockers.
15:43:10 <sgallagh> Every once in a while, we decide at the meeting that while it's technically a blocker according to the criteria, it's sufficiently unlikely.
15:43:13 <dgilmore> sgallagh: not really
15:43:14 <nirik> but if there's no rc...
15:43:21 <sgallagh> But as far as no RC, I'm all aboard.
15:44:35 <pbrobinson> I think the grub2 bug is a reasonable example of the issue
15:45:10 <dgilmore> sgallagh: and in this case I am saying that if we do not have a RC 40 hours before the decision we slip
15:45:34 <sgallagh> dgilmore: Right, I completely agree with that.
15:45:56 <sgallagh> I disagreed with "if we don't have all the blockers with fixes", because there's some small leeway on that.
15:45:56 <dgilmore> in part to not waste time on meetings, and in part to prevent people from trying to rush and put in an insane effort to try and ship
15:46:12 <dgilmore> sgallagh: if we have blockers we can not have a RC
15:46:31 <sgallagh> Sure we can, if blockers are proposed after an RC is cut.
15:46:31 <dgilmore> sgallagh: I think you are trying to pick cracks in things
15:46:49 <sgallagh> Anyway, I think we're in agreement on the major point, so I am going to stop talking about it.
15:46:50 <dgilmore> sgallagh: but we have a RC
15:46:58 <dgilmore> its possible it could shiop
15:46:59 <dgilmore> ship
15:47:13 <dgilmore> sgallagh: you are honestly not being helpful here
15:47:18 <sgallagh> dgilmore: We are *in agreement*
15:47:27 <sgallagh> We're just not using the same words.
15:47:42 <dgilmore> no
15:47:47 <dgilmore> anyway
15:48:02 <dgilmore> there is one clunky thing with the RC we have
15:48:19 <dgilmore> the installer media has _RC in the name and I do not think it should
15:48:29 <pbrobinson> no, that's not ideal
15:48:38 <nirik> yeah, that seems wrong
15:49:50 <nirik> I dont know if we need a rc2 yet, but if we want to fix that we should do so asap so it can be tested instead of rc1.
15:50:13 <dgilmore> it may be fixable witha  config change, but may need pungi code change
15:50:18 <dgilmore> not 100% sure yet
15:51:59 <dgilmore> does anyone have anything they want to cover about F24 in general?
15:52:16 * nirik has nothing
15:52:27 * pbrobinson doesn't think he does
15:52:31 <dgilmore> #topic Secondary Architectures updates
15:52:32 <dgilmore> #topic Secondary Architectures update - ppc
15:52:38 <dgilmore> pbrobinson: how is ppc looking?
15:52:55 <pbrobinson> with the grub2 fix which landed in -34 we're looking pretty good on f24
15:53:18 <pbrobinson> we've done a bunch of testing on today's branched compose and the grub2 issues are gone
15:53:20 <dgilmore> awesome
15:54:01 <pbrobinson> and I've spent the morning (and some of Sat) making sure we're ready to compose a RC, I was kind of holding off awaiting a confirmation of the RC naming issue
15:54:19 <pbrobinson> rawhide is looking pretty good as well
15:54:48 <pbrobinson> this morning we got an almost complete rawhide compose (cloud images crashed with a weird blivet bug)
15:54:52 <dgilmore> I asked adamw about it and he did not seem overly concerned, I did consider doing one that would have Final instead of RC
15:55:21 <adamw> dgilmore: one thing that i noticed yesterday is the composeinfo for RC-1.1 says 'final: false'
15:55:53 <pbrobinson> so does that mean we get the eat kittens pop still?
15:56:00 <adamw> i dunno what it means
15:56:22 <masta> it's just a flag passed to lorax to make anaconda show a scarry warning
15:56:54 <dgilmore> adamw: thats then a bug in pungi
15:57:28 <adamw> masta: i don't know if the 'final' status in composeinfo is the same as that setting.
15:57:28 <pbrobinson> masta: yes, exactly what I meant by the eat kittens popup
15:57:48 <adamw> since we've never used productmd/pungi4 for a release before, this is all new.
15:58:10 <pbrobinson> yep
15:58:23 <maxamillion> adamw: there be *new* dragons :)
15:58:32 <masta> it's a setting in pungi config
15:58:36 <dgilmore> adamw: pungi docs say if the LABEL starts with RC it sets the compose as supported
15:58:42 <adamw> in fact it doesn't look like it is, the openqa tests for the RC1 compose don't seem to show the timbuktu warning.
15:58:53 <adamw> dgilmore: is that the same thing as 'final'? :)
15:58:54 <dgilmore> which I guess is the internal wording for --is-final
15:59:04 * adamw checks the code. again.
15:59:07 <masta> yes
15:59:08 <dgilmore> adamw: I assumed so
15:59:30 <dgilmore> adamw: so we have a bug or two in pungi :(
15:59:38 <dgilmore> and will have to fix today and do a new compose
16:00:15 <dgilmore> #topic Secondary Architectures update - s390
16:00:23 <dgilmore> sharkcz: how is s390?
16:00:37 <dgilmore> I saw we disabled 31 bit support :)
16:00:50 <sharkcz> build-wise and from some one-off testing, we look good
16:00:54 * maxamillion still doesn't entirely understand how 31-bit is even a thing
16:01:34 <sharkcz> 31-bit addresses/32-bit data, and now enjoying no more multilib and other advantages
16:01:50 <maxamillion> :)
16:02:07 <dgilmore> maxamillion: it is special :) straight out of the 60's or 70's
16:02:10 <adamw> dgilmore: poking through the code, 'supported' and 'is_final' / 'isfinal' seem like separate concepts.
16:02:10 <pbrobinson> I worked on the final compose last week, I'm hoping I'm almost there, will be spending more time on it tomorrow
16:02:34 <adamw> hmm, or not.
16:02:39 <pbrobinson> ^^ is for s390x
16:02:51 <maxamillion> dgilmore: fun
16:02:54 <adamw> oh i dunno, you figure itout.
16:03:03 <sharkcz> pbrobinson: oh, sounds promising
16:03:06 <adamw> dgilmore: oh, there's also the thing i noticed on the mailing list
16:03:32 <sharkcz> pbrobinson: did the sshfs help a bit?
16:03:34 <dgilmore> adamw: I am a week behind in email
16:03:43 <adamw> dgilmore: the inconsistency in the Workstation filenames: https://kojipkgs.fedoraproject.org/compose/24/Fedora-24-20160611.1/compose/Workstation/x86_64/iso/
16:03:47 <dgilmore> so no idea what you are talking about
16:03:58 <adamw> dgilmore: note netinst has "RC-1.1" (the full label), live has just "1.1"
16:04:14 <dgilmore> adamw: that is expected and I told you that
16:04:21 <adamw> oh, i guess i forgot.
16:04:39 <dgilmore> I said only the install media has _RC in it
16:05:17 <dgilmore> anything else for s390?
16:05:39 <sharkcz> dgilmore: rawhide also looks good, and that's all
16:05:44 <dgilmore> #topic Secondary Architectures update - arm
16:05:49 <dgilmore> pbrobinson: how is aarch64?
16:05:59 <pbrobinson> we're looking good there as well
16:06:05 <dgilmore> nice
16:06:06 <pbrobinson> status is about the same as ppc
16:06:38 <pbrobinson> we didn't have the grub2 issue but given we've had issues with bootloaders earlier in the cycle it's being tested to ensure no regression
16:07:00 <maxamillion> +1
16:07:01 <pbrobinson> other than that bot f24 and rawhide packages are pretty close to mainline with no major blockers
16:07:13 <dgilmore> excellent
16:07:28 <pbrobinson> and we're ready for a RC compose there too
16:07:38 <pbrobinson> I'll wait for the new pungi version
16:07:39 <dgilmore> its probably a good point in time to merge aarch64, ppc64 and ppc64le into primary koji
16:07:44 <dgilmore> we will get it one day
16:08:35 <dgilmore> #topic Fedora 25 planning
16:08:38 <dgilmore> https://fedoraproject.org/wiki/ReleaseEngineering/PriorityPipeline#F25_Committed_Tools_Changes
16:08:57 <dgilmore> that has the list of things we have commited to for Fedora 25 and a few things we would like to have
16:09:37 <dgilmore> If there is anything people want we need to get it on the radar
16:09:40 <nirik> hum
16:09:54 <dgilmore> nirik ?
16:10:08 <nirik> so, that means I guess that releng wants a supported jenkins (since it would be release blocking right?)
16:10:17 <tyll> I can probably help with the autosigning change, but the information about what contents need to be signed and in which tools is missing for me
16:10:31 <dgilmore> nirik: maxamillion is working on that, and we are considering just using ansible
16:10:56 <nirik> dgilmore: ok, it says jenkins there, but good that the discussion is yet to happen. ;)
16:10:59 <dgilmore> tyll: thanks, I will try get it all documented this week
16:11:08 <maxamillion> nirik: it says Jenkins where?
16:11:11 <maxamillion> nirik: oh right
16:11:18 <maxamillion> nirik: that needs to be updated
16:11:20 <nirik> https://fedoraproject.org/wiki/ReleaseEngineering/PriorityPipeline#F25_Committed_Tools_Changes
16:11:24 <nirik> ok
16:11:32 <maxamillion> nirik: https://fedoraproject.org/wiki/Changes/ReleaseEngineeringAutomationWorkflowEngine
16:11:34 <dgilmore> nirik: well it was going ot be jenkins to align with internal. But turns out people are pushing for a few different options
16:11:42 <maxamillion> nirik: the "engine" bit is still "TBD"
16:11:48 <nirik> maxamillion: yes, read that. ;) Wasn't sure it matched up there, but good to know
16:12:10 <maxamillion> nirik: yeah, I'll update the PriorityPipeline page
16:12:23 <paragan> is there any plans for changing yum to dnf in koji source code so that when I install koji rpm, I will not get yum installed?
16:12:38 <dgilmore> paragan: that is off topic
16:12:46 <dgilmore> you would have to ask the koji devs
16:12:59 <dgilmore> we are only contributors to and users of it
16:13:07 <dgilmore> paragthough I doubt it
16:13:12 <dgilmore> paragan: I doubt it
16:13:36 <dgilmore> since koji has to run on everything from python 2.6 and rhel 5 up
16:14:12 <dgilmore> yum will also always be needed for rhel 5 6 and 7 buildroots in mock
16:15:00 <dgilmore> anything else for F25?
16:15:02 <paragan> dgilmore, thanks for this information.
16:15:23 <dgilmore> #topic Open Floor
16:15:33 <dgilmore> does anyone want to raise anything?
16:16:01 <nirik> dgilmore: was there any news on the macosx and windows deliverables for f25?
16:16:09 <nirik> or those folks just disappeared?
16:16:25 <dgilmore> nirik: we are waiting on the team requesting it to commit to providing hardware
16:16:49 <nirik> ok.I have heard nothing on it at all...
16:16:58 <dgilmore> There is an internal Releng guy that is tasked with figuring out how to do koji builds for OSX
16:17:06 <nirik> cool. thats good news.
16:17:11 <maxamillion> +1
16:17:21 <dgilmore> though  mikeb or mikem said they wanted it done as a content generator
16:17:29 <dgilmore> something that I personally think is not right
16:17:43 <dgilmore> the more I see content generators The more I think they are bad
16:17:58 <pbrobinson> +1 on that
16:18:15 <tyll> I am wondering about the accepted Freeze Exceptions that would fix broken deps - will they be processed soon?
16:18:15 <pbrobinson> well a poorly thought out rushed design/deliverable
16:18:30 <dgilmore> being decoupled makes it really confusinga nd hides the visibility
16:18:34 <nirik> adamw: ^ see tyll's question above?
16:18:49 <dgilmore> tyll: that is a question for adamw
16:19:27 <tyll> does he need to request them to be pushed to stable?
16:19:35 <dgilmore> tyll: he does
16:20:56 <dgilmore> I was wondering if anyone would object to making the meeting an hour or two earlier for the next 4 weeks?
16:21:14 <nirik> it would be early for me, but I could do it.
16:21:16 <tyll> I see, it is a little bit hard to be sure that we get the base repo broken dep free for F24 because without the updates pushed to stable it is hard to figure out which packages need to be retired still
16:22:06 <adamw> i thought i sent a push request the other day.
16:22:20 * dgilmore is going ot be in Brisbane, where the normal meeting start time is 01:30
16:22:23 <adamw> oh, i guess i didn't.
16:22:27 <adamw> i'll do one today.
16:22:58 <tyll> adamw: thank you
16:23:04 <pbrobinson> dgilmore: does that mean we just assign all deliverables to you in your absence ;-)
16:23:32 <dgilmore> pbrobinson: possibly :P
16:23:35 <maxamillion> dgilmore: +1 - I can make earlier meeting work
16:24:37 <dgilmore> lets shoot for an hour earlier
16:25:31 <nirik> ok
16:25:33 <dgilmore> #info meeting time will be 1 hour earlier for the next 4 weeks
16:26:01 <pbrobinson> can we make that a permanent move? An hour earlier for me would generally better
16:26:03 <dgilmore> nirik: that is 8:30am for you right?
16:26:08 <nirik> yeah
16:26:25 <nirik> I'm often putting out fires monday morning, but can try and make it. ;)
16:26:46 <maxamillion> dgilmore: will need to let acarter know since she often schedules backlog and retro stuff in/around that time slot
16:27:05 <dgilmore> maxamillion: right
16:30:02 <dgilmore> stipid networks
16:30:08 <dgilmore> stupid even
16:30:29 * dgilmore will contact acarter about changing the times of grooming and backlog meetings
16:30:44 <dgilmore> to see if we can switch permanently to an hour earlier
16:31:50 <maxamillion> +1
16:33:13 <dgilmore> lets wrap up unless someone has something else?
16:34:06 <dgilmore> #endmeeting