16:01:15 <sgallagh> #startmeeting FESCO (2016-09-30)
16:01:15 <zodbot> Meeting started Fri Sep 30 16:01:15 2016 UTC.  The chair is sgallagh. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:15 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:01:15 <zodbot> The meeting name has been set to 'fesco_(2016-09-30)'
16:01:15 <sgallagh> #meetingname fesco
16:01:15 <zodbot> The meeting name has been set to 'fesco'
16:01:15 <sgallagh> #chair maxamillion dgilmore jwb nirik paragan jsmith kalev sgallagh Rathann
16:01:15 <zodbot> Current chairs: Rathann dgilmore jsmith jwb kalev maxamillion nirik paragan sgallagh
16:01:15 <sgallagh> #topic init process
16:01:20 <kalev> morning
16:01:22 <jwb> hi (again?)
16:01:22 <maxamillion> .hello maxamillion
16:01:22 <jsmith> .hello jsmith
16:01:22 <sgallagh> Sorry, bad copy-paste
16:01:22 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
16:01:25 <zodbot> jsmith: jsmith 'Jared Smith' <jsmith.fedora@gmail.com>
16:01:26 <nirik> .hello kevin
16:01:28 <zodbot> nirik: kevin 'Kevin Fenzi' <kevin@scrye.com>
16:01:42 <sgallagh> .hello sgallagh
16:01:43 <zodbot> sgallagh: sgallagh 'Stephen Gallagher' <sgallagh@redhat.com>
16:01:52 <jsmith> FWIW, I'll only be here for the first 30 minutes or so, and then I'm driving to the airport
16:01:59 <jsmith> Sorry about that :-/
16:02:37 <t8m> can you please start with the OpenSSL 1.1.0 rebase change ticket - I'd need to leave early
16:03:45 <t8m> .hello t8m
16:03:46 <zodbot> t8m: Sorry, but you don't exist
16:04:03 <t8m> t8m = Tomas Mraz <tmraz@redhat.com>
16:04:21 <sgallagh> t8m: zodbot expects your FAS ID
16:04:27 <t8m> ah ok
16:04:33 <t8m> .hello tmraz
16:04:34 <zodbot> t8m: tmraz 'Tomáš Mráz' <tmraz@redhat.com>
16:05:12 <sgallagh> OK, we have six people, so I guess we can start.
16:05:38 <sgallagh> #topic #1629 F26 System Wide Change: OpenSSL 1.1.0
16:05:38 <sgallagh> .fesco 1629
16:05:41 <zodbot> sgallagh: #1629 (F26 System Wide Change: OpenSSL 1.1.0) – FESCo - https://fedorahosted.org/fesco/ticket/1629
16:06:19 <sgallagh> t8m: Is this going to need a mass-rebuild?
16:06:26 <sgallagh> (or a large side-tag?)
16:06:41 <kalev> I think the proposal makes sense to me; F26 timeframe sounds about right and the compat package which avoids a flag day with switching over sounds like a good plan as well
16:06:50 <sgallagh> Oh, sorry. I see it under Release Engineering now
16:06:58 * nirik is in favor, +1
16:07:01 <t8m> the question is what are the requirements for the side tag
16:07:21 <kalev> I don't think there's a need for a side tag as long as the compat package and new openssl are built at the same time
16:07:33 <jsmith> +1 from my side -- I don't see anything concerning about this change
16:07:37 <sgallagh> t8m: a side-tag is a good way to do things if you want all of the rebuilt packages to land together
16:07:40 <kalev> +1 from me as well
16:07:49 <t8m> I mean - my current plan is to not have -devel subpackage for the compat package
16:07:52 <sgallagh> But if there's a compat package that will work without a rebuild, then it's not strictly needed
16:08:28 <sgallagh> t8m: Hmm, well that does mean that anyone who wants to update their package is forced to move to 1.1.0
16:08:32 <t8m> and that means there will be a window when some packages can not be rebuild because they would need patch to be buildable with 1.1.0
16:08:38 <t8m> yes
16:08:38 <sgallagh> Which, for minor updates, might not be ideal.
16:08:40 <kalev> yeah, makes sense to do it without a -devel subpackage to make sure everything switches over
16:08:47 <sgallagh> So it may be worth discussing.
16:08:59 <sgallagh> kalev: Depends on how hard the switchover is
16:09:02 <nirik> t8m: do you have any sense of how hard it will be ?
16:09:15 <nirik> is it just changing some calls/names? or is everything different?
16:09:20 <t8m> it depends on the package's use of openssl
16:09:29 <t8m> for most the changes are pretty trivial
16:09:29 <sgallagh> For example, if some package needs to patch a CVE, they won't be able to do it without fixing up to support 1.1.0
16:09:35 <t8m> yes
16:10:11 <t8m> if it uses some long-ago deprecated things like krb5 did - it is a little bit harded to switch to new API
16:10:41 <sgallagh> t8m: OK, so what's the common case? Is it that >50% of packages would work with a trivial rebuild?
16:11:04 <sgallagh> Will most packages work with very minor patching?
16:11:14 <sgallagh> Or is it genuine porting work for everyone?
16:11:46 <t8m> sgallagh, active upstreams are already patched, for most packages - I'd say >50% deps it is trivial
16:11:58 <sgallagh> t8m: Please define trivial
16:12:07 <sgallagh> Small code edits or simple rebuild?
16:12:15 <t8m> small code edits
16:12:19 <sgallagh> ok
16:12:49 <sgallagh> And the OpenSSL maintainers are volunteering to assist where needed, so I suppose I'm +1 to the plan as-is
16:12:51 <t8m> simple rebuild would work only in case I'd ship deprecated apis in 1.1.0 which I do not want to.
16:13:31 <jwb> I'm +1
16:13:38 <nirik> is there any planned day/time to land this?
16:14:03 <t8m> I'd like to do that once the compat package is reviewed
16:14:06 <sgallagh> I'd recommend landing it in Rawhide as soon as possible (if not sooner) to give it maximum time to shake out
16:14:11 <t8m> yeah
16:14:23 <nirik> I'd recommend picking a date and announcing in advance...
16:14:27 <kalev> t8m: I can help with reviewing the compat package if you want, I've got some experience with those
16:14:32 <nirik> and have that date not be friday or the weekend.
16:14:32 <t8m> nirik, will do
16:15:07 <sgallagh> t8m: Assuming kalev helps you get the compat package ready in time, maybe announce for Wednesday?
16:16:02 <t8m> sgallagh, you mean wednesday next week? I am not sure I am fully ready for that. But the next wednesday would work for me
16:16:38 <sgallagh> t8m: Sure
16:17:07 <sgallagh> I count +4 right now, nirik, kalev, jwb and myself.
16:17:14 <sgallagh> jsmith, maxamillion?
16:17:20 <jsmith> I already said +1
16:17:24 <t8m> I'll submit the compat package for review some time next week
16:17:25 <maxamillion> +1
16:17:32 <t8m> and announce the plan too
16:17:41 <sgallagh> jsmith: Thanks, I missed it
16:17:42 <maxamillion> sorry, was catching up on the backlog .... attempting to multitask
16:18:12 <t8m> btw any preference on the compat package name? - openssl110 would continue the naming of older openssl compat packages
16:18:17 <sgallagh> #action OpenSSL 1.1.0 Change is approved, will be landed alongside 1.0.2 compat package in Rawhide as soon as possible (+6, 0 ,-0)
16:18:31 <t8m> but compat-openssl would be more aligned with other packages
16:18:49 <sgallagh> t8m: compat-openssl110 ?
16:19:07 <sgallagh> (not joking, I think that would be the most clear name)
16:19:12 <nirik> is it intended to keep that around after things move? or drop it?
16:19:15 <kalev> t8m: compat-something I think. this is usually the naming for ABI compatibility libraries that don't ship a -devel subpackage
16:19:21 <t8m> ok
16:19:50 <t8m> nirik, I think we should keep it for third parties
16:20:09 <nirik> ok. if you like. ;)
16:20:09 <kalev> yeah, I agree, should keep it for at least one release so that 3rd party repos have time to switch over
16:20:27 <kalev> same as we did for gnutls a few releases back
16:20:37 <jwb> should be longer than one release
16:21:03 <t8m> we'll have to keep it in rhel "forever" but that is not too relevant to Fedora
16:21:39 <t8m> I would not have problem keeping it at least until upstream discontinues 1.0.2 support
16:21:39 <sgallagh> I'd suggest keeping it until there is no longer a Fedora to upgrade from that shipped 1.0.2 as the primary choice
16:21:53 <t8m> sgallagh, that works for me
16:22:06 <sgallagh> So keep it in F26 and F27, then drop it for F28?
16:22:20 <maxamillion> sgallagh: makes sense, +1
16:22:22 <kalev> sounds like a good plan, +1
16:22:27 * nirik is happy to leave it to t8m, was just curious. ;)
16:23:06 <t8m> thanks for the approval
16:24:01 <kalev> t8m: I think I would call it compat-openssl10 actually, as the sonames are libssl.so.10 and libcrypto.so.10, but up to you of course
16:24:11 <sgallagh> #info The compat package will be retained at least for the duration of F27. After that, it will be at the maintainer's discretion.
16:24:43 <sgallagh> (Is that fair?)
16:24:53 <kalev> yup
16:24:56 <t8m> no problem with that
16:25:07 <sgallagh> Great. Anything else on this topic, then?
16:26:25 <sgallagh> #topic #1626 Release blocking deliverables for Fedora 25
16:26:25 <sgallagh> .fesco 1626
16:26:26 <zodbot> sgallagh: #1626 (Release blocking deliverables for Fedora 25) – FESCo - https://fedorahosted.org/fesco/ticket/1626
16:26:29 <t8m> bye
16:27:17 * nirik shakes his head at this one. ;(
16:27:18 <sgallagh> So, this is very concerning to me.
16:27:24 <jsmith> Worried that we still haven't heard from the Cloud WG...
16:27:25 <sgallagh> Only one WG has bothered to reply.
16:28:08 <jsmith> (Half-serioius) Proposal: Threaten not to produce their deliverables if they don't respond.
16:28:10 <jwb> kushal said last time we met that the list that was sent out was correct
16:28:15 <jwb> for Cloud
16:28:17 <nirik> well, Workstation did too
16:28:27 <jwb> however, nobody actually followed up in the ticket or with jkurik
16:28:55 <jwb> jsmith: empty threats don't go over well.  and if we turned them off, we'd have to turn them back on in a crisis state later
16:28:58 <sgallagh> Proposal: Starting with F25: any image not approved by the appropriate group before Alpha Freeze is *not blocking* for that release.
16:29:12 <sgallagh> Let's avoid this next time around and/or force the issue.
16:29:20 <maxamillion> sgallagh: +1
16:29:31 <sgallagh> err, that should read F26
16:29:54 <maxamillion> right yes ... I transposed it in my head for some reason
16:29:56 <jsmith> jwb: While I typically don't agree with the "name and shame" idea for individuals, I have no qualms about doing that for WGs/SIGs
16:31:10 <nirik> I know a while back when we first started generating that release deliverables page we set some deadline for it...
16:31:17 * nirik looks for the ticket
16:31:30 <sgallagh> nirik: It was Alpha Freeze, I'm 90% sure
16:33:02 <nirik> We agreed to have a list of release deliverables for F23 at least two weeks before the Alpha Freeze with release blocking deliverables marked
16:33:11 <nirik> https://fedorahosted.org/fesco/ticket/1427#comment:12
16:33:19 <jsmith> Anyhoo, I have to run to the airport.  Please consider me +1 on whatever you decide on this issue, +1 on the DNF ticket, and +1 to jkurik's proposals on the Changes not in ON_QA ticket.
16:33:39 <sgallagh> nirik: Right, but I don't think we specified a penalty. Hence my proposal above.
16:34:27 <nirik> I guess I am in favor, but I think it might lead to some anoyance... people asking for image blah even though it's not release blocking...
16:34:31 <nirik> but we can give it a go
16:34:54 <sgallagh> nirik: Well, if it's not release-blocking, rel-eng can just go ahead and say "no, try next release"
16:35:41 <maxamillion> with citation of FESCo's decision (pending there is one)
16:35:47 <nirik> sure.
16:36:55 <sgallagh> Alright, so let's talk in present-day terms.
16:37:23 <sgallagh> I for one am not confident that anything that didn't get a formal approval is being maintained well enough to treat as blocking, personally
16:37:32 <jwb> i'm not in favor of this
16:37:33 <sgallagh> But that's likely going to prove an unpopular position :)
16:38:09 <jwb> in reality, what will likely happen is rel-eng will say no, they'll get cast as the bad guy, there will be a scramble to turn stuff back on, and then they'll be left with even more work
16:38:28 <jwb> i'd rather FESCo just hound people continuously until someone gives a definitive answer
16:38:56 <nirik> who is our cloud liason currently?
16:39:22 <maxamillion> kushal?
16:39:34 <jsmith> jberkus?
16:39:56 <maxamillion> I'm in the Cloud WG but don't consider myself a representative of it in any official capacity ... I supposed I can become that if necessary
16:40:07 <maxamillion> I basically just show up and try to help out when I can
16:40:14 <kalev> can you bring it up at next Cloud WG meeting?
16:40:22 <maxamillion> kalev: certainly
16:40:39 <kalev> let's just leave it at that and move on and give Cloud WG some more time?
16:40:48 <nirik> well, when we started the working groups, we agreed to have a fesco member as liason to each of them... to ask them stuff fesco wanted and bring up things to fesco they wanted.
16:40:50 <sgallagh> kalev: We're in beta freeze right now
16:41:01 <sgallagh> If it comes up, what (if anything) do we slip on?
16:41:27 <kalev> just assume same things are blocking than what was in last release?
16:43:58 <kushal> I am iirc.
16:44:13 <kushal> sorry in between too many things, thus being late.
16:44:36 <maxamillion> kushal: +1
16:44:56 <kushal> Cloud base image is the only blocking deliverable.
16:44:59 <kushal> Atomic is not.
16:45:01 <maxamillion> kushal: I'll follow up with you off-meeting, but basically we need a Cloud WG response to https://fedorahosted.org/fesco/ticket/1626
16:45:17 <nirik> https://fedoraproject.org/wiki/Fedora_Program_Management/ReleaseBlocking/Fedora25
16:45:24 <kushal> Oh, I thought I replied last week.
16:45:38 <kushal> I was pointed out to that in the last week
16:45:44 <kushal> not sure what happened to my mail
16:45:50 <kushal> will reply once again to him.
16:47:03 <sgallagh> OK, and jwb: did you want to speak for Workstation? (Are you still the liaison, or did we ask stickster to step in there?)
16:47:10 <jwb> that's stickster
16:47:19 <jwb> kalev could likely cover too?
16:47:37 <nirik> AGREED: The list as it currently stands is fine for Fedora 25. The getfedora.org website should not offer 32-bit x86 media for Server.
16:47:42 <nirik> ^ there
16:47:49 <nirik> oops. thats the wrong one
16:47:53 * nirik needs more coffe.
16:48:56 <nirik> they basically said the list was fine but then started asking about workstation ostree.
16:49:02 <kalev> Workstation's deliverables for F25 are exactly the same as we had for F24. the atomic workstation image isn't planned to be release blocking
16:49:05 <sgallagh> oh, right
16:49:33 <nirik> so, I think hopefully we are ok on this and should move on?
16:49:38 * kalev agrees.
16:49:54 <Southern_Gentlem> ? 32bit is no longer release blocking correct?
16:50:08 <kalev> correct
16:50:11 <nirik> Southern_Gentlem: it hasn't been since f24
16:50:36 <sgallagh> OK, let's move on, then
16:50:49 <sgallagh> #topic #1628 F26 System Wide Change: DNF 2.0
16:50:49 <sgallagh> .fesco 1628
16:50:51 <zodbot> sgallagh: #1628 (F26 System Wide Change: DNF 2.0) – FESCo - https://fedorahosted.org/fesco/ticket/1628
16:51:43 <nirik> so, they built dnf 2.0 in rawhide yesterday.
16:51:55 <nirik> but there was at least lorax that was not at all ready for it.
16:52:20 <nirik> There is a lorax patch now (and I think a build just went by)
16:52:28 <nirik> and anaconda has a patch and should be doing a build.
16:53:01 <nirik> did we already approve this in a previous meeting?
16:53:04 <maxamillion> I'm not the biggest fan of this change, but given this statement from the wiki page "DNF-2 is the upstream DNF version, the only version actively developed." ... there's not really much option to not roll with it at least at some point
16:53:15 <maxamillion> nirik: we might have?
16:53:21 <maxamillion> nirik: it does sound a little familiar
16:53:28 <maxamillion> I think we at least discussed it
16:53:39 <kalev> I think we did, yep
16:53:46 <jwb> we approved it last meeting
16:54:30 <jwb> oh, wait
16:54:50 <nirik> well, hopefully we will have a rawhide tomorrow... I'm not sure what all else will break
16:55:09 <nirik> (we didn't have one today due to the lorax changes not having landed)
16:55:33 <jwb> yeah, we approved it last meeting
16:55:37 <jwb> 16:35:01 <jwb> #approved DNF2.0 Change for F26 is accepted (1:8, 0:1, -1:0)
16:55:46 <jwb> https://meetbot.fedoraproject.org/fedora-meeting/2016-09-16/fesco.2016-09-16-16.06.log.html
16:55:51 <nirik> also, dnf 2.0 broke for me right away not understanding the "type=rpm" in repofiles... which they quickly fixed
16:56:06 <nirik> but it bodes ill to me that they tested. ;)
16:56:30 <kalev> I think it has a test setup that's based on F24
16:56:57 <nirik> could be. that type thing only landed in rawhide recently I think
16:58:40 <nirik> anyhow, I guess we will move forward and try and get everything working.
16:58:49 <nirik> it's a neverending battle. ;(
16:59:07 <sgallagh> OK, so nothing to see here, move along?
16:59:13 <sgallagh> #info this was approved last week
16:59:23 <sgallagh> #topic #1630 F25 Changes not in ON_QA status (<100% completed).fesco 1630
16:59:23 <kalev> anyway, I'm glad that we have a compose that breaks when things aren't really working, as opposed to end users getting a dnf that doesn't work at all
16:59:27 <sgallagh> #undo
16:59:27 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x462616d0>
16:59:31 <sgallagh> #topic #1630 F25 Changes not in ON_QA status (<100% completed)
16:59:31 <sgallagh> .fesco 1630
16:59:32 <zodbot> sgallagh: #1630 (F25 Changes not in ON_QA status (<100% completed)) – FESCo - https://fedorahosted.org/fesco/ticket/1630
17:00:26 <sgallagh> Hmm, in retrospect I think this one was talking about maxamillion rather than adamw.
17:00:28 <sgallagh> What's the word?
17:00:29 <maxamillion> The Rel-Eng Atuomation Workflow Engine has been proposed to the Infrastructure Team and is currently undergoing code review and security audit before deploying but I have demonstrated it in working fashion
17:00:35 <maxamillion> Automation*
17:01:19 * kalev doesn't have any insight into any of the three changes.
17:01:21 <maxamillion> puiterwijk is currently looking into it and I'm awaiting the review to complete before we can deploy ... from there to the best of my knowledge it's just a matter of getting it deployed
17:01:37 <maxamillion> I'll update the BZ for the change, I should have done so last week
17:03:34 <maxamillion> is that good enough for an ON_QA status?
17:04:15 <sgallagh> maxamillion: If it's not deployed, I'd kind of say probably not. That being said, where is the interest in this to end-users?
17:04:40 <sgallagh> /me wonders why this is a Change, exactly.
17:04:42 <nirik> yeah, I am not sure this even really needs to be a change?
17:05:05 <nirik> but hey, we approved it a while back so...
17:07:00 <sgallagh> I mean, if it was going to replace some part of the release pipeline, I could see it
17:07:05 <sgallagh> But it looks additive
17:07:59 <maxamillion> it's currently additive, eventually it will replace things ... it won't have any real user facing anything, the main idea for filing the Change was for the sake of visibility ... a lot of RelEng stuff flies under the radar
17:08:28 <sgallagh> maxamillion: Fair enough
17:08:45 <sgallagh> Let's call it ON_QA, then
17:09:10 <maxamillion> +1 thanks
17:10:40 <sgallagh> So for the GHC one, I propose we wait a week for updates, as jkurik recommends.
17:10:51 <maxamillion> sgallagh: +1
17:10:55 <sgallagh> And also that we enact the Contingency Plan on the RPM autoprovides
17:10:59 <nirik> yep.
17:11:04 <nirik> +1 for that
17:11:42 <maxamillion> +1 for Contingency Plan
17:12:13 <kalev> the contingency plan is to just not announce the change, right?
17:12:30 <sgallagh> kalev: Not to put it in the guidelines, yes
17:12:40 <sgallagh> Or at least, the guidelines should say F26+
17:13:43 <kalev> yeah, that may be a good call as it pretty much requires a mass rebuild for all package to pick up the autoprovides
17:13:48 <kalev> as I understand it.
17:13:58 <kalev> and we'll have one for F26
17:14:15 <sgallagh> Oh right, it's going to need macro changes.
17:14:27 <sgallagh> But that didn't land either, so this is just plain not done.
17:14:53 <sgallagh> Do we still have quorum, BTW?
17:14:55 <nirik> well, all python packages... but yeah
17:15:00 <jwb> sgallagh: i think so?
17:15:14 <sgallagh> jwb: Ah ok. I wasn't sure if you were still around.
17:15:29 <jwb> not much to add yet
17:15:38 <maxamillion> jsmith: left but he said he was +1 to everything left
17:15:41 <maxamillion> errr
17:15:50 <maxamillion> jsmith left but he said he was +1 to everything else *
17:15:56 <jwb> i'm +1 to everything that's been discussed on this topic thus far
17:15:57 <maxamillion> words and typing is hard
17:16:33 <sgallagh> #agreed Wait one week for information about GHC. Enact Contingency Plan on python autoprovides (+6, 0, -0)
17:16:50 <sgallagh> #topic Next Week's Chair
17:17:30 <sgallagh> Who wants it?
17:17:37 <kalev> I could do it next week
17:17:54 <sgallagh> #info kalev to chair 2016-10-07 meeting
17:17:58 <sgallagh> #topic Open Floor
17:18:34 <kalev> how's the release coming along? any blockers that would need fesco's poking?
17:18:35 <sgallagh> Just a general note: the conversation that cmurf started on devel@ regarding Cloud and Server positioning is Important(TM). FESCo members should probably follow and chime in
17:18:49 <nirik> sgallagh: I have it marked, but have not yet had time to reply.
17:18:54 <sgallagh> Thanks
17:19:05 <nirik> https://qa.fedoraproject.org/blockerbugs/milestone/25/beta/buglist
17:19:21 <sgallagh> kalev: Blocker list is in remarkably good shape at the moment
17:19:30 <nirik> yeah, nothing too crazy
17:19:32 <nirik> yet
17:19:37 <sgallagh> (Which probably means no one is testing anything yet ;-) )
17:19:52 <kalev> sounds great :)
17:20:13 <kalev> ohh, could people vote on https://bugzilla.redhat.com/show_bug.cgi?id=1377741 FE status, please?
17:20:32 <kalev> as in, those that usually do that kind of voting :)
17:20:48 <sgallagh> kalev: Will do
17:21:05 <kalev> thanks
17:22:44 <sgallagh> Anything else?
17:23:10 <kalev> nothing from me
17:23:16 * nirik has nothing else off hand
17:24:34 <maxamillion> nodda here
17:24:41 <sgallagh> Thanks for coming, folks
17:24:45 <sgallagh> #endmeeting