17:00:03 #startmeeting F28 Beta Go/No-Go meeting 17:00:03 Meeting started Thu Mar 22 17:00:03 2018 UTC. The chair is jkurik. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:03 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:00:03 The meeting name has been set to 'f28_beta_go/no-go_meeting' 17:00:05 .hello2 17:00:06 bowlofeggs: bowlofeggs 'Randy Barlow' 17:00:09 #meetingname F28-Beta-Go-No-Go-meeting 17:00:09 The meeting name has been set to 'f28-beta-go-no-go-meeting' 17:00:11 morning 17:00:16 #topic Roll Call 17:00:22 .hello jkurik 17:00:24 jkurik: jkurik 'Jan Kurik' 17:00:28 #chair nirik adamw sgallagh mboddu 17:00:28 Current chairs: adamw jkurik mboddu nirik sgallagh 17:00:44 hi everybody 17:00:50 .hello2 17:00:51 frantisekz: frantisekz 'František Zatloukal' 17:00:57 .hello2 17:00:58 kparal: kparal 'Kamil Páral' 17:01:22 #topic Purpose of this meeting 17:01:28 #info Purpose of this meeting is to check whether or not F28 Beta is ready for shipment, according to the release criteria. 17:01:34 #info This is determined in a few ways: 17:01:37 .hello2 17:01:38 sgallagh: sgallagh 'Stephen Gallagher' 17:01:41 #info * No remaining blocker bugs 17:01:46 #info * Release candidate compose is available 17:01:53 #info * Test matrices for Beta are fully completed 17:01:59 #topic Current status 17:02:12 As far as I am aware, the RC for F28 Beta is not yet ready as there are several blockers: 17:02:25 https://qa.fedoraproject.org/blockerbugs/milestone/28/beta/buglist 17:02:26 .hello mohanboddu 17:02:27 mboddu: mohanboddu 'Mohan Boddu' 17:02:33 As such, we do not have test matrices for the RC. 17:02:34 spoiler alert: we are not going to be go today. ;) 17:02:38 .hello adamwill 17:02:39 adamw: adamwill 'Adam Williamson' 17:02:45 Anyone wants to add something ? 17:03:03 .hello2 17:03:04 x3mboy: x3mboy 'Eduard Lucena' 17:03:10 .hello 17:03:11 mattdm: (hello ) -- Alias for "hellomynameis $1". 17:03:15 .hello2 17:03:16 mattdm: mattdm 'Matthew Miller' 17:03:19 it might be of use to do a quick blocker review so we can possibly make an rc in the next few days... 17:03:25 #info The RC for F28 Beta is not yet ready 17:03:33 #link https://qa.fedoraproject.org/blockerbugs/milestone/28/beta/buglist - F28 Blockers 17:03:43 #info As we have no RC there are subsequently no Test Matrices for the F28 Beta RC 17:03:44 yeah, i was assuming we're going to do blocker review here. 17:03:51 Let's do at least Mini-blocker review 17:04:03 adamw: may I ask you please to chair the mini-blocker review ? 17:04:05 #topic Mini-Blocker Review 17:04:06 sure. 17:04:10 did you #chair me ? 17:04:11 I think everyone knows its no go, better do a blocker review 17:04:16 yes 17:04:33 adamw: yes, you have the chair 17:04:43 thanks 17:04:53 #topic (1558906) AttributeError: 'DiskDevice' object has no attribute 'isDisk' 17:04:54 #link https://bugzilla.redhat.com/show_bug.cgi?id=1558906 17:04:54 #info Proposed Blocker, anaconda, ON_QA 17:06:01 this is basically a showstopper for intel firmware raid, +1 for me 17:06:06 +1 blocker 17:06:09 i'm building an iso with both fixes for testing 17:06:17 +1 Blocker I guess 17:06:24 I voted +1 blocker in the ticket 17:06:36 +1 17:06:41 +1 17:07:03 +1 if i get a vote 17:07:12 +1 Blocker 17:07:14 (do i get a vote? haha) 17:07:15 bowlofeggs: Everyone gets a vote at blocker meetings 17:07:18 ah cool 17:07:21 It's consensus-based. 17:07:34 there is a Highly Advanced Vote Counting Formula 17:07:44 HAVCF 17:07:47 +1 17:07:59 it exists entirely in my head 17:08:01 +1 frome me 17:08:02 and is impossible to document 17:08:36 proposed #agreed 1558906 - AcceptedBlocker (Beta) - accepted as a violation of "The installer must be able to detect and install to hardware or firmware RAID storage devices" 17:08:37 bowlofeggs: like we dont have enough abbreviations :D 17:08:50 ack 17:08:54 ack 17:08:59 ACK 17:09:08 #agreed 1558906 - AcceptedBlocker (Beta) - accepted as a violation of "The installer must be able to detect and install to hardware or firmware RAID storage devices" 17:09:16 #topic (1559180) hangs on launch 17:09:17 #link https://bugzilla.redhat.com/show_bug.cgi?id=1559180 17:09:17 #info Proposed Blocker, anaconda, NEW 17:09:25 just FYI, I'll secretalize (kparal said :D ) 17:09:28 so i'm just finishing my comment on this, but it's basically 'can't reproduce anyway' 17:09:32 anywhere* 17:09:36 thanks frantisekz 17:09:56 so -1 17:10:00 Yeah, I tried as well 17:10:01 -1 17:10:07 this is the memory thing? 17:10:14 -1 17:10:23 If its specific hardware or some scenario then -1 17:10:24 -1 17:10:25 he said he had 3 GB 17:10:38 not having tried it, i'm -1 due to failure to reproduce 17:10:45 the virt def has 2gb 17:10:57 2097152 17:11:17 he said later he tried with 3 17:11:23 openqa uses 2 anyway, iirc, and that works 17:11:32 huh. 17:11:33 I've hit a similar issue, on 2gig of RAM but I did some dnf installs on the live media 17:11:42 oh, yeah, that eats memory 17:11:51 anything that changes the filesystem is applied to a memory overlay 17:12:10 -1 17:12:15 If you install stuff in live media, they will go to memory, yes or yes 17:12:28 note, it *does* look to me like something in the package stack is using more memory in f28 17:12:35 from a quick look at some numbers this morning 17:12:41 still, -1. 17:12:43 the problem is probably packagekit running by default and downloading metadata 17:12:44 -1 based on what we know, but might be good to isolate 17:13:00 proposed #agreed 1559180 - RejectedBlocker (Beta) - no-one else is able to reproduce this in testing on various systems, so it does not appear to be a blocker 17:13:02 kparal: I was thinking it might be because we don't have the preseeded cache 17:13:05 ack 17:13:14 ack 17:13:16 ack 17:13:17 ack 17:13:18 ack 17:13:18 ack 17:13:39 kparal: even the installer images appear to use substantially more memory during early anaconda phase than f27 ones did 17:13:44 but we should look into that manually and confirm 17:13:51 * adamw just checked openqa records 17:14:03 is that anaconda memory graphing thing still in anaconda? 17:14:08 #agreed 1559180 - RejectedBlocker (Beta) - no-one else is able to reproduce this in testing on various systems, so it does not appear to be a blocker 17:14:12 nirik: yes. that's what openqa uses. 17:14:27 nice. should be easy to see what and how it's spiking 17:14:37 nirik: but the original logs from f27 final are lost unfortunately, i didn't have things set up to save them from garbage protection. some numbers can still be found in the compose check emails, though. 17:14:38 nirik: yeah. 17:14:45 well 17:14:51 the log just shows 'anaconda' as using the memory 17:15:10 this is because anaconda doesn't run dnf externally but imports it 17:15:13 (i think) 17:15:21 #topic (1557659) aacraid: Host adapter abort request 17:15:21 #link https://bugzilla.redhat.com/show_bug.cgi?id=1557659 17:15:21 #info Proposed Blocker, kernel, NEW 17:15:22 ah, thats not as helpfull as I remember it being 17:15:50 nirik: it's useful when it's things that run *during the package install* that use memory 17:15:53 as those things show up separate 17:15:59 ok 17:16:15 so, on this one, -1 since we appear to have a usable and document-able workaround 17:16:39 -1 yeah 17:17:01 -1 then 17:17:09 -1 17:17:11 -1 17:17:39 -1 17:18:07 -1 17:18:35 proposed #agreed 1557659 - RejectedBlocker (Beta) - this is specific to one particular type of RAID adapter, and we have a reasonable workaround, so we don't consider this a serious enough violation to be a Beta blocker 17:18:45 ack 17:18:50 ack 17:18:56 ack 17:19:07 #agreed 1557659 - RejectedBlocker (Beta) - this is specific to one particular type of RAID adapter, and we have a reasonable workaround, so we don't consider this a serious enough violation to be a Beta blocker 17:19:09 ack 17:19:15 #topic (1559174) Locally-changed booleans not preserved on upgrade from F27 to F28, cannot be set permanently after upgrade 17:19:15 #link https://bugzilla.redhat.com/show_bug.cgi?id=1559174 17:19:15 #info Proposed Blocker, policycoreutils, MODIFIED 17:19:23 i'm going to submit the update for this soon, fwiw. 17:19:49 +1 blocker 17:20:03 +1, as noted in the ticket I am concerned that this would have other effects beyond how it was discovered 17:20:12 +1 17:20:13 i'm a weak +1, approx. same rationale as sgallagh 17:20:13 +1 17:20:26 i think this actually hit me but i didn't realize it 17:20:46 i thougth maybe there was just a change to policy, didn't realize my bools had gotten switched 17:20:53 +1 Blocker 17:21:05 yeah, it'd affect an awful lot of folks IRL, i think 17:21:08 +1 17:21:14 +1 17:21:22 +1 17:21:37 proposed #agreed 1559174 - AcceptedBlocker (Beta) - accepted as a violation of the upgrade criteria applied to release-blocking server roles (as noted in the bug) 17:21:41 ack 17:21:43 we really should propose that criteria change, sigh 17:21:45 ack 17:21:52 ack 17:21:57 ack 17:22:04 adamw: Which change? 17:22:19 ack 17:22:21 sgallagh: to officially state that the upgrade requirements apply to release-blocking roles 17:22:34 sgallagh: at present that's not really stated, though we agreed in 2017 that we thought it should be 17:23:01 ack 17:23:06 #agreed 1559174 - AcceptedBlocker (Beta) - accepted as a violation of the upgrade criteria applied to release-blocking server roles (as noted in the bug) 17:23:25 should we also do the proposed freeze exceptions, or leave those for monday? 17:23:41 adamw: Probably not worth bothering with now, with server roles going away 17:24:13 sgallagh: eh...it's an easy tweak and it'd make us less likely to forget about it when revising the criteria for roles going away. 17:24:55 Could we do https://bugzilla.redhat.com/show_bug.cgi?id=1466093 if only because it's got a build ready? 17:25:08 i figure if we're gonna do one let's just do 'em all 17:25:10 there's only 4 17:25:17 #info doing proposed freeze exceptions quickly 17:25:22 adamw: ok. go on 17:25:24 4, 5...they're close 17:25:29 #topic (1559188) Failed to initialize security module 17:25:29 #link https://bugzilla.redhat.com/show_bug.cgi?id=1559188 17:25:29 #info Proposed Freeze Exceptions, abrt, NEW 17:26:10 seems reasonable to ease crash reporting from lives 17:26:24 yeah, +1 FE 17:26:26 we should have a criterion that bug reporting has to work, right? 17:26:27 +1 FE 17:26:37 +1 17:26:41 kparal: there's a criterion for reporting installer crashes 17:26:51 not for post-install, i think on the theory 'we can fix that with updates' 17:26:54 +1 FE 17:26:56 +1 FE 17:27:14 +1 17:27:25 ok 17:27:40 proposed #agreed 1559188 - AcceptedFreezeException (Beta) - this would ease reporting of crashes on live images, and immediately after install 17:27:42 so how do you report crashed anaconda from Live? 17:27:49 without FAF 17:28:00 kparal: anaconda uses a different flow, i think 17:28:06 it's still libreport, but...different 17:28:08 ack 17:28:11 ack 17:28:12 ack 17:28:12 alright 17:28:15 ack 17:28:16 ack 17:28:41 kparal: i mean, if you want to test it, please do :) don't think i have 17:28:45 #agreed 1559188 - AcceptedFreezeException (Beta) - this would ease reporting of crashes on live images, and immediately after install 17:28:53 #topic (1557511) annobin: failure building wxGTK3 in F28 17:28:53 #link https://bugzilla.redhat.com/show_bug.cgi?id=1557511 17:28:54 #info Proposed Freeze Exceptions, annobin, ON_QA 17:29:52 i ran into this weeks ago, in fact 17:29:58 thought i'd filed a bug but having trouble finding it 17:30:10 does this need an FE when we have BROs? 17:30:19 (build root overrides ☺) 17:30:21 Not particularly, no 17:30:36 well, in theory no, but then, i don't like the inconsistency bros cause 17:30:47 if we build something for beta with a package that's not *in* beta itself, that kinda sucks 17:30:52 i could see an argument for FE 17:30:54 well, also, can't we just do this after beta is in the can? 17:30:57 because we'll need a long-term BRO otherwise 17:31:07 this is kind of a generic issue with BROs, admittedly. 17:31:41 i do dislike the inconsistency i feel when i hang out with too many bros… 17:31:48 haha 17:31:57 i would +0 FE 17:31:58 If only we had some sort of mechanism that would allow us to set BROs for individual builds. We could call these builds "modules"... 17:32:02 so, i mean, i'm kinda a weak +1 just for consistency between buildroot and beta compose 17:32:05 sgallagh: =) 17:32:09 not opposed, but also i don't see a strong case for 17:32:13 sgallagh: (or, you know, tags) 17:32:14 sgallagh: :D 17:32:49 annobin isn't in any of the default installs, right? 17:32:56 It's really only a build-time dep 17:33:05 no idea. i'd guess not. 17:33:33 Which would make it functionally irrelevant if it was in stable vs. BRO 17:33:42 not really, it'd be in Everything for beta. 17:34:02 Right, but if it's only ever a build-time dep, what's the problem? 17:34:04 if anyone treats the Beta as an artefact, it's inconsistent. it's a small point. 17:34:26 it's not like our releases are very reproducible really anyway, but every little helps. :P 17:34:35 just vote something and let's move on? 17:35:18 so far nobody is -1 17:35:22 I guess weak +1 17:35:24 so we could just allow it a FE 17:35:58 bowlofeggs: oh, i forgot, the basic element of the Highly Sophisticated Voting Algorithm is that it always needs at least +3 17:35:59 I do not really know, so I am +0 17:36:00 (or -3) 17:36:21 so everyone gets to sit around till we have at least three votes in either direction, or i get bored and propose we just punt it. :P 17:36:43 adamw: How about +0.5 FE? :) 17:36:56 in theory we're supposed to require a vote from QA, one from releng and one from devel, i think, but several years ago i decided that we all plausibly represent all three anyway, so any old vote will do. :P 17:37:12 mboddu: welp, now we're at +2.5...:P 17:37:13 I'm +1 on the grounds that it will likely stay in the buildroot anyway 17:37:17 So might as well be consistent' 17:37:24 zenos paradox of FE... 17:37:30 2.999999999999 17:37:53 proposed #agreed 1557511 - AcceptedFreezeException (Beta) - this is accepted as a freeze exception on the basis it's needed to build some packages, and pulling it in as an FE rather than using a buildroot override keeps the contents of the composes more consistent 17:38:04 ack 17:38:06 ack 17:38:08 ack 17:38:14 ack 17:38:22 #agreed 1557511 - AcceptedFreezeException (Beta) - this is accepted as a freeze exception on the basis it's needed to build some packages, and pulling it in as an FE rather than using a buildroot override keeps the contents of the composes more consistent 17:38:22 ack 17:38:47 .fire sgallagh late acks will be punished 17:38:47 adamw fires sgallagh late acks will be punished 17:38:53 #topic (1466093) Decommissioning domain controller role fails when role deployed on Fedora 25 then system upgraded to Fedora 26 17:38:53 #link https://bugzilla.redhat.com/show_bug.cgi?id=1466093 17:38:53 #info Proposed Freeze Exceptions, firewalld, MODIFIED 17:39:14 hum, i should update that topic. 17:39:45 adamw: haha 17:39:52 The real problem here is a bug in firewalld that cannot handle removing firewall services with overlapping entries 17:39:54 true summary is "Decommissioning domain controller role fails if system is rebooted after deployment (due to firewalld bug)" 17:40:02 +1 FE. 17:40:09 In this case, "freeipa-ldap" and "freeipa-ldaps" 17:40:18 +1 FE, as I said in the ticket 17:40:27 +1 FE as well 17:40:30 well 17:40:33 +1 FE 17:40:37 there's an argument that, why does it need to go in the compose 17:40:41 +1 FE 17:40:45 since it requires a reboot, the system *really* ought to get updated 17:40:54 putting it in the compose basically helps openQA and not much else. 17:41:27 since the update is a whole new version of firewalld... 17:42:09 should we really pull it in? 17:42:46 adamw: https://www.firewalld.org/2018/03/firewalld-0-5-2-release 17:42:55 Looks like a pretty constrained bugfix release 17:42:56 adamw: I would say that for beta we can take the risk 17:43:55 ok, then 17:44:30 proposed #agreed 1466093 - AcceptedFreezeException (Beta) - decommissioning isn't actually part of the release criteria, but is a significant function of server roles, and pulling this in will improve openQA test coverage 17:44:38 ack 17:44:40 ack 17:44:42 ack 17:45:25 ack 17:45:50 ack 17:46:53 ACK 17:48:27 #agreed 1466093 - AcceptedFreezeException (Beta) - decommissioning isn't actually part of the release criteria, but is a significant function of server roles, and pulling this in will improve openQA test coverage 17:48:35 #topic (1546743) gfortran internal compiler error when compiling cp2k-5.1 on ppc64le 17:48:35 #link https://bugzilla.redhat.com/show_bug.cgi?id=1546743 17:48:35 #info Proposed Freeze Exceptions, gcc, NEW 17:51:33 hmm... I am not sure how many people will be affected by a bug in fortran compiler 17:51:34 well, everyone saw 'fortran' and immediately passed out, it seems 17:51:38 * adamw goes around with the smelling salts 17:52:13 but as it might affect c++ (if I understand the comment from Jakub) then I am +1 FE 17:52:24 :D (adam, go ahead and do rand(-1,1) for a few times and you have your votes) 17:52:30 i mean, i guess, to be consistent with the previous one, +1. 17:52:32 jkurik: And also only on ppc64le arch 17:52:55 ah, right; ppc64le only 17:52:57 they apparently haven't actually submitted an update for this, though. 17:53:36 So, I am soft -1, just because its fortran and just ppc64le 17:54:54 * mattdm notes that fortran is still very important to scientific computing 17:55:55 i mean, since this has a BRO already thus anything we pull in will be built with it anyway, i'd say +1 for the consistency argument 17:56:06 Yeah, I'm fine with adamw's perspective. +1 17:56:08 well...the bro expired 17:56:12 well, some people may want to try beta for building their stuff too 17:56:14 it was in force from 03-16 to 03-20 17:56:31 so anything built between those dates that we pulled in, was built with it 17:56:36 anything built since was built with the old one again 17:56:37 yay consistency 17:56:47 adamw: Yeah, just realized its BRO, so +1 17:56:48 so, yeah, +1 FBR 17:57:56 proposed #agreed 1546743 - AcceptedFreezeException (Beta) - accepted on the same principle as the annobin bug (1557511) - buildroot/compose consistency 17:57:59 ack 17:58:03 ack 17:58:06 ack 17:58:14 ack 17:59:06 #agreed 1546743 - AcceptedFreezeException (Beta) - accepted on the same principle as the annobin bug (1557511) - buildroot/compose consistency 17:59:14 last one: 17:59:14 #topic (1558510) "Star" feature missing in Nautilus 17:59:14 #link https://bugzilla.redhat.com/show_bug.cgi?id=1558510 17:59:14 #info Proposed Freeze Exceptions, nautilus, NEW 17:59:46 What's the critereon for this one? 17:59:46 eh, seems like there may even be no bug here? 17:59:56 jsmith: it's a proposed FE, there are no criteria for FEs. 18:00:11 just https://fedoraproject.org/wiki/QA:SOP_freeze_exception_bug_process#Freeze_exception_bug_principles 18:00:23 jsmith: every app needs to have at least one star 18:00:32 Ah, right... silly me. 18:00:35 -1 to FE 18:00:57 Can easily be dealth with after the freeze, if there's anything to be done 18:01:13 -1 18:01:20 -1, not even clear whether this is a bug, no fix ready 18:01:22 It's not even clear that this is a bug 18:01:23 is this touted as a gnome feature with the new release? 18:01:37 yeah, pretty unclear 18:01:54 -1 for now 18:01:57 -1 FE 18:02:19 -1 for now, probably fixed by an update 18:02:32 well, we can't fix it on the live image with an update, of course 18:02:42 and presumably the idea is people may use the live to showroom gnome 3.28 18:03:03 adamw: Ah right 18:03:17 if people are voting -1 on the basis that it's not clear if there's actually a bug, how about we just punt to clarify that? 18:03:25 it'll come up again at monday's meeting and hopefully should be clear by then 18:03:35 adamw: ack 18:03:53 * nirik nods 18:03:54 adamw: Better, +1 punt 18:04:07 fine 18:04:54 proposed #agreed 1558510 - punt (delay decision) - we suspect there's actually no bug here, but we agreed to just punt it till Monday's meeting, hopefully by then it'll be clear whether there's actually a bug and it merits FE status 18:05:00 ack 18:05:07 ack 18:05:08 ack 18:05:09 ack 18:05:12 ack 18:05:21 #agreed 1558510 - punt (delay decision) - we suspect there's actually no bug here, but we agreed to just punt it till Monday's meeting, hopefully by then it'll be clear whether there's actually a bug and it merits FE status 18:05:25 OK, that's all the proposals 18:05:26 thanks folks 18:05:33 #topic Test Matrices coverage 18:05:37 #info As there is no RC yet, Test matrices are not ready as well 18:05:43 #info We are skipping the Test Matrices coverage check 18:05:49 #topic Go/No-Go decision 18:06:01 proposed #agreed Due to missing RC for the F28 Beta release and presence of blocker bugs, the decision is “No Go”. The Beta release slips for one week to “Target #1” date (April 3rd). We are not going to slip the Final GA yet. 18:06:32 nirik, mboddu, adamw, and other folks ^^^ 18:06:44 yep. ack 18:06:45 +1 from me 18:06:47 ack 18:07:20 ack 18:07:36 ack 18:07:50 #agreed Due to missing RC for the F28 Beta release and presence of blocker bugs, the decision is “No Go”. The Beta release slips for one week to “Target #1” date (April 3rd). We are not going to slip the Final GA yet. 18:07:56 sure 18:07:59 #action jkurik to publish the Go/No-Go result 18:08:07 #action jkurik to organize second round of Go/No-Go meeting for F28 Beta on Thursday, March 29th at 17:00UTC 18:08:12 #topic Open floor 18:08:18 anything else to discuss today on this meeting ? 18:08:29 I'd like to note that the status on Blocker fixes is moving ahead really well, so next week seems favorable 18:09:12 sgallagh: Yeah, we are in a far better shape than we were 2 weeks before 18:09:12