16:32:12 <dgilmore> #startmeeting RELENG (2016-01-25)
16:32:12 <zodbot> Meeting started Mon Jan 25 16:32:12 2016 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:32:12 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:32:12 <zodbot> The meeting name has been set to 'releng_(2016-01-25)'
16:32:12 <dgilmore> #meetingname releng
16:32:12 <zodbot> The meeting name has been set to 'releng'
16:32:12 <dgilmore> #chair dgilmore nirik tyll sharkcz bochecha masta pbrobinson pingou maxamillion
16:32:12 <zodbot> Current chairs: bochecha dgilmore masta maxamillion nirik pbrobinson pingou sharkcz tyll
16:32:15 <dgilmore> #topic init process
16:32:21 <maxamillion> .hello maxamillion
16:32:22 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
16:32:40 <bochecha> hi
16:32:51 <dgilmore> I have two topics I want to go over first
16:33:01 <dgilmore> hey bochecha how are you?
16:33:06 <nirik> morning
16:33:31 <bochecha> dgilmore: I haven't planned on going, no
16:33:36 <bochecha> but then it's not that far...
16:34:16 <dgilmore> lets get started
16:34:35 <bochecha> dgilmore: should I go? :)
16:34:43 <dgilmore> #topic firefox addon signing
16:34:48 <dgilmore> bochecha: I will be there
16:35:05 * masta is here
16:35:18 * sharkcz is here
16:35:45 <dgilmore> so FESCo on Friday decided to write a letter to mozilla to raise some concerns over the policy of allowing only signed extentions to run
16:36:08 <dgilmore> it means that the extentions shipped in Fedora no longer work
16:36:59 <dgilmore> sgallagh: is crafting the email. A idea being tossed out is working with firefox to enable us to sign our own extentions
16:37:15 <dgilmore> To me there looks to be a really high cost to that
16:37:38 <dgilmore> as the buildroot needs access to the cert and key to sign them at build time
16:37:54 <dgilmore> and it will need to be implemented on all arches
16:37:55 <maxamillion> dgilmore: would this involve sigul?
16:37:57 <nirik> Looks like there's a api...
16:37:58 <nirik> https://developer.mozilla.org/en-US/Add-ons/Distribution
16:38:09 <nirik> https://olympia.readthedocs.org/en/latest/topics/api/signing.html
16:38:16 <dgilmore> https://www.mozdevgroup.com/docs/pete/Signing-an-XPI.html
16:38:21 <dgilmore> that is what i found
16:38:30 <dgilmore> maxamillion: maybe
16:38:43 <dgilmore> or some other private signing setup
16:39:10 <nirik> yeah it would have to sign in the buildroot... so it would need access to the keys _and_ network access.
16:39:34 <dgilmore> nirik: that involves network access and credentials in the buildroot
16:39:40 <nirik> yes.
16:40:26 <nirik> but it looks like you just ask the api to sign it, and it has a hash...so the clients just check the hash? so that could mean we could just extract the xpi from the rpm after the build and get it added.
16:40:32 * nirik isn't sure.
16:40:43 <dgilmore> nirik: to me reads asd you upload it
16:40:46 <dgilmore> they sign it
16:40:55 <dgilmore> then you can fetch the signed copy
16:41:06 <nirik> yeah, I guess so.
16:41:30 <dgilmore> not sure that I like the black box signing
16:41:31 <masta> Do we have an x509 cert in Fedora along the lines of the GPG signing keys?
16:41:44 <dgilmore> but anyway. something to put on teh radar
16:42:04 <dgilmore> masta: we do not have a cert and key to use for object signing
16:42:16 <masta> okay
16:44:46 <dgilmore> if anyone has suggestions on how we can implement I am open to ideas
16:44:47 <dgilmore> #topic buildroot issues
16:45:11 <dgilmore> last night pbrobinson pointed me at an issue with libsolv failing to build on s390, s390x and aarch64
16:45:26 <dgilmore> all the BuildRequires are installed
16:45:31 <maxamillion> heh, that should prove to be interesting
16:46:04 <nirik> couldn't be a dnf bug... :)
16:46:06 <dgilmore> but on further inspection what is installed in the buildroot differs a bit between the arches
16:46:23 <dgilmore> ruby looks to have made a huge mess for themselves
16:46:42 <dgilmore> they starting using soft deps for /usr/bin/ruby
16:46:56 <dgilmore> so you could have ruby or jruby satisfy things
16:47:45 <masta> right
16:47:50 <dgilmore> on primary koji and ppc rubygems and a few other packages get pulled in
16:48:07 <dgilmore> on arm and s390 kojis they are ignored
16:48:33 <dgilmore> if you explictly list the packages as BuildRequires they get installed okay
16:48:43 <masta> hum... I wonder if the (soft) dependency graph has some exclusive/exclude-arch happening
16:48:44 <dgilmore> sharkcz: quick question
16:48:57 <dgilmore> sharkcz: do you have koji patched to use createrepo_c
16:49:33 <sharkcz> dgilmore: yes
16:49:55 <dgilmore> sharkcz: primary is not using it
16:50:12 <sharkcz> yep, one more thing to try
16:51:07 <dgilmore> sharkcz: arm is using createrepo
16:51:12 <dgilmore> so it is not that
16:51:16 <sharkcz> ok
16:51:41 <dgilmore> though it has in the past
16:53:14 <sharkcz> I have a repo stored from the shadow build, only ~200 packages there, so dnf/libsolv guys can look into it
16:53:32 <dgilmore> I think that releng needs to take the lead in having a discussion over how softdeps work in buildroots
16:53:56 <dgilmore> there has already been a ticket filed to disable them, just because dnf's buildroot is bigger than yum's
16:54:30 <dgilmore> I think we need to get some hard data over the effects of different options
16:54:48 <masta> we need deterministic buildroot creation.
16:55:01 <dgilmore> we do
16:55:20 <maxamillion> +1
16:55:33 <dgilmore> if soft deps are turned off, likely people will need to add extra BuildRequires
16:56:00 <dgilmore> to account for soft deps not being installed
16:56:22 <maxamillion> let's write up some action items and assign them to folks to have them followed up on and get a status report at next week's meeting ... need some movement on this probably sooner rather than later
16:57:00 <nirik> well, does this mean FPC would need to change guidelines around them?
16:57:04 <dgilmore> maxamillion: not sure what tehy are right now
16:57:08 <nirik> or this is all seperate from that?
16:57:10 <masta> so we need to reproduce scenarios that have different outcomes.
16:57:15 <dgilmore> nirik: they probably do
16:57:51 <sharkcz> there should be difference between build deps in a buildsystem (install all) and runtime deps (actually use weak deps)
16:58:30 <dgilmore> https://fedoraproject.org/wiki/Packaging:WeakDependencies
16:59:42 <dgilmore> nirik: the guidelines have no mention of how they should work in buildroots
16:59:58 <dgilmore> tibbs: ping, are you around?
17:01:21 <dgilmore> #action work with fpc to have guidelines over expected buildroot behaviour
17:02:02 <dgilmore> #action work with koji devs to ensure we can configure the expected behaviour
17:02:30 <dgilmore> #topic Secondary Architectures updates
17:02:30 <dgilmore> #topic Secondary Architectures update - ppc
17:02:37 <dgilmore> sharkcz: how is ppc?
17:02:55 <sharkcz> no issues afaik
17:03:14 <sharkcz> and for the record, pbrobinson updated the builders today to f23
17:04:34 <dgilmore> #info builders have been updated to f23
17:04:40 <dgilmore> #topic Secondary Architectures update - s390
17:04:50 <dgilmore> sharkcz: did rawhide get re-enabled?
17:05:02 <sharkcz> yes, runs now daily
17:05:09 <dgilmore> good :)
17:05:19 <sharkcz> and I'm watching the disk space consumption
17:05:24 <dgilmore> okay
17:05:42 <dgilmore> we should be freeing up a bunch of space next week
17:06:01 <sharkcz> are you moving the old releases?
17:06:03 <dgilmore> though with snapshots it will likely take a couple of weeks to kick in
17:06:07 <dgilmore> sharkcz: yes
17:06:13 <sharkcz> good :-)
17:06:35 <dgilmore> sharkcz: all hardlinked to http://dl.fedoraproject.org/pub/archive/fedora-secondary/
17:06:48 <dgilmore> just need to rm from /mnt/koji/tree
17:07:00 <dgilmore> which should happen next week
17:07:09 <sharkcz> ok
17:07:19 <dgilmore> there will only be 22 23 and 24 left
17:07:32 <nirik> cool.
17:07:47 <sharkcz> how does mirrormanager work with such moved releases?
17:08:08 <dgilmore> it may need repointed
17:08:18 <dgilmore> but will work
17:08:31 <sharkcz> ok
17:09:18 <dgilmore> just pinged adrianr on it
17:09:33 <dgilmore> #topic Secondary Architectures update - arm
17:09:45 <dgilmore> I think it is a little behind build wise
17:10:05 <sharkcz> one more item to s390 - there is a ticket to allow nfs to the fedora storage from the builders
17:10:22 <dgilmore> sharkcz: pwhalen: are you guys aware of any major issues?
17:10:31 <dgilmore> #undo
17:10:31 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x3669890>
17:10:56 <dgilmore> #info sharkcz working to get builders access to /mnt/koji
17:11:10 <dgilmore> sharkcz: okay, hopefully something can be worked out
17:11:18 <dgilmore> it will make things easier
17:11:31 <dgilmore> #topic Secondary Architectures update - arm
17:11:31 <sharkcz> dgilmore: yep, they allowed http, so we need one more port
17:12:06 <pwhalen> dgilmore, for aarch64 waiting on binutils 2.26 to make its way to arm.koji
17:12:14 <dgilmore> pwhalen: okay
17:12:50 <dgilmore> #info aiting on binutils 2.26 to make its way to arm.koji
17:12:54 <dgilmore> #undo
17:12:54 <zodbot> Removing item from minutes: INFO by dgilmore at 17:12:50 : aiting on binutils 2.26 to make its way to arm.koji
17:12:58 <dgilmore> #info waiting on binutils 2.26 to make its way to arm.koji
17:13:06 <dgilmore> #topic open floor
17:13:15 <dgilmore> I have a few things.
17:13:24 <dgilmore> I will not be able to make next weeks meeting
17:13:29 * nirik had a few things as well. will wait for dgilmore.
17:13:41 <dgilmore> so if someone wanted to run it, that would be awesome
17:14:10 <dgilmore> nirik: go ahead, just went blank
17:14:49 <nirik> so, mass rebuild starts next tuesday, right? I'd like to get all builders updated before then... so might try and do that later this week if there's no objections.
17:15:01 <dgilmore> nirik: yeah  it does
17:15:10 <dgilmore> as I will be in Brno then
17:15:15 <dgilmore> reminds me
17:15:44 <dgilmore> I created a releng fas user and releng@fedoraproject.org bugzilla account
17:15:59 <dgilmore> the releng user will be doing teh mass rebuild and filing the bugs
17:16:19 <nirik> yeah, so another thing was that acarter wanted me to roll out some pungi4 thing if you were gone this week... but I have no details. :) So, if you could meet up with me today and let me know what you need me to do that would be lovely. Or an email. or is it still needed?
17:16:33 <nirik> are you going to fire the mass rebuild from brno? or is someone else?
17:16:46 <dgilmore> nirik: my only objection would be if updating introduces the regressions, issues with teh buildroot seen on arm and s390 kojis
17:17:03 <maxamillion> need to step afk, apologies
17:17:12 <dgilmore> nirik: I am going to do it from brno
17:17:15 <dgilmore> just not as me
17:17:22 <nirik> yeah, I can do stg first and scratch build whatever failed? which package was that?
17:17:32 <dgilmore> nirik: libsolv
17:17:42 <nirik> ok. can do that today and see if I see it.
17:17:45 <dgilmore> nirik: stg koji is a bit special right now
17:17:51 <nirik> oh?
17:18:19 <dgilmore> it has koji builds that have livemedia-creator support and signed-repo support
17:18:31 <dgilmore> which is what acarter wanted rolled out
17:18:32 <nirik> ok. Hopefully scratch builds should still work ok.
17:18:36 <nirik> ah ha.
17:18:40 <dgilmore> I am hoping to geta  cleaned up patch set today
17:18:50 <dgilmore> they should still work yes
17:18:55 <nirik> so you may still get it rolled out before you leave?
17:19:12 <dgilmore> we still have one issue to fix to get livemedia-creator working right in koji
17:19:18 <dgilmore> nirik: I hope so
17:19:30 <nirik> cool. well, do let me know if you need me to do anything...
17:19:36 <dgilmore> will do
17:19:52 <dgilmore> need to get the compose report tool packaged up
17:19:59 <dgilmore> and test it
17:20:54 <dgilmore> I have kickstart changes ready
17:22:28 <dgilmore> nirik: will let you know
17:22:37 <dgilmore> nirik: anything else you wanted to bring up?
17:22:54 <nirik> one last thing... need to get with pbrobinson and get arm hub migrated.
17:23:01 <nirik> hopefully can do that soon. thats all
17:23:13 <dgilmore> :) awesome that would be good
17:25:05 <dgilmore> okay lets wrap up
17:25:07 <dgilmore> thanks all
17:25:10 <dgilmore> #endmeeting