18:01:54 <smooge> #startmeeting EPEL
18:01:54 <zodbot> Meeting started Wed May 18 18:01:54 2016 UTC.  The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:01:54 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:01:54 <zodbot> The meeting name has been set to 'epel'
18:01:54 <smooge> #meetingname EPEL
18:01:54 <zodbot> The meeting name has been set to 'epel'
18:01:54 <smooge> #chair smooge nirik Evolution bstinson avij
18:01:54 <smooge> #topic aloha
18:01:54 <smooge> #topic Outstanding issues
18:01:54 <zodbot> Current chairs: Evolution avij bstinson nirik smooge
18:02:22 <smooge> Anyone able to make todays meeting?
18:02:24 <nirik> oh yeah, epel meeting. ;)
18:02:28 * nirik is here
18:02:51 <smooge> I am not sure we will have quorum
18:02:55 <smooge> Evolution is at a show
18:03:01 <smooge> avij probably hsa finals
18:03:08 <smooge> bstinson is in another meeting
18:03:17 * bstinson is here
18:03:25 <bstinson> and there
18:04:27 <smooge> ok we have a bare minimal quorum
18:04:31 <smooge> #info Orphans and removals
18:04:31 <smooge> #info EL-5/EL-6 build root changes via tibbs
18:04:31 <smooge> #info Recharter EPEL
18:04:31 <smooge> #info Removing EL-5 in 8 months.
18:04:32 <smooge> #info alternative arch issues for CentOS
18:04:48 <tibbs|w> Howdy.
18:05:01 <smooge> We have a large number of orphans in the list to be cleaned out.
18:05:55 <smooge> I need to find out how to remove them from EPEL this week
18:06:05 <nirik> I think tyll is traveling this week.
18:06:21 <smooge> ok
18:06:25 <nirik> there was a ticket in fedora releng for someone else to do the retires there... I think it had a SOP / instruction thing
18:06:40 <smooge> ok I will look at that and put that on my todo for tomorrow
18:07:15 <nirik> cool.
18:07:31 <smooge> tibbs has been doing a lot of work on the build root changes. I think he has finished them all
18:07:40 <tibbs|w> Pretty much.
18:07:51 <tibbs|w> Should be minor things from here on out.
18:08:28 <tibbs|w> I did find a couple of problems caused by the way I added %clean.  I may have a workaround, though.
18:08:48 <tibbs|w> Also, I never sent an announcement but the various Python macros are in EL6 now.
18:09:07 <tibbs|w> Including the py3 ones, though they just do nothing.
18:09:17 <tibbs|w> So... fewer ifdefs required now.
18:09:45 <bstinson> awesome
18:09:45 <nirik> well, could the fedora py3 ones tie into the epel python34 ones?
18:09:50 <nirik> or is that too complex?
18:11:25 <nirik> and yeah, many thanks for working on those.
18:11:36 <orionp> well, there is no python34 in EPEL6
18:12:01 <nirik> sure, I was taking epel7, but didn't mean to confuse.
18:12:01 <smooge> yet
18:12:59 <orionp> the py3 macros in epel7 work using whatever is default python3
18:14:10 <nirik> sure, but do they map to the fedora ones... ie, can I take python-foo from master and just build it fine on epel7 branch?
18:15:11 <orionp> yes, as long as you use python%{python3_pkgversion} in your package names
18:16:15 <smooge> anything else on this?
18:16:42 <smooge> tibbs, thank you so much for this work. Do you have anything else on your plans ?
18:16:57 <tibbs|w> If people want the python macros in EL5, there's that.
18:17:15 <tibbs|w> Otherwise if you want something, just file tickets against epel-rpm-macros and I'll have a look.
18:17:23 <smooge> ok I am not sure how much use they would be. We only have about 8 months on that
18:17:40 <tibbs|w> There's some stuff we're adding to Fedora which I'll try to push all the way down.
18:17:53 <tibbs|w> %gpg_verify and some more default filtering macros.
18:18:03 <tibbs|w> Though I don't think those can get down to EL6.
18:18:11 <tibbs|w> And... 8 months can't some soon enough.
18:18:18 <smooge> yeah :)
18:18:52 <smooge> I didn't get very far in reading the other Fedora Working Groups charters to get an idea of what to write up on the EPEL one.
18:19:06 <smooge> However I did get a bunch of cruft off my table so I can do so this wekeend.
18:19:35 <smooge> I have the CentOS alternative arch items on the list but I don't know what the status was other than Evolution might have something to say
18:19:41 <smooge> bstinson, did you know anything?
18:19:51 <nirik> I think aarch64 building/fixing is moving along...
18:20:05 <yselkowitz> yes, the aarch64 build is in progress
18:20:19 <nirik> someone asked me about armv7 in private email...
18:20:21 <yselkowitz> I pushed a bunch of updates to stable this week
18:20:40 <yselkowitz> the big question will be ghc
18:20:47 <bstinson> nirik: that's on the list of PoC builds also
18:20:58 * nirik nods.
18:21:38 <smooge> yselkowitz, thanks
18:21:47 <yselkowitz> np
18:22:01 <smooge> yselkowitz, is it a "why oh why?" or a "I need input from EPSCO?"
18:22:04 <smooge> on ghc
18:22:20 <yselkowitz> it needs to be coordinated
18:22:35 <yselkowitz> atm I'm thinking we should rebase to f21
18:23:05 <yselkowitz> same version, more patches
18:23:11 <yselkowitz> which we need for aarch64
18:23:19 <yselkowitz> but we'd have to be sure that doesn't break abi, which is very delicate
18:23:29 <yselkowitz> maintainer hasn't responded to needinfo
18:23:37 <smooge> ah who is that?
18:23:38 <smooge> .whoowns ghc
18:23:39 <zodbot> smooge: petersen
18:23:47 <smooge> .fasinfo petersen
18:23:48 <yselkowitz> there's also the fact that ghc_arches won't have aarch64 until 7.3
18:23:48 <zodbot> smooge: User: petersen, Name: Jens Petersen, email: petersen@redhat.com, Creation: 2005-04-13, IRC Nick: juhp, Timezone: Asia/Tokyo, Locale: en, GPG key ID: EF27EDD1, Status: active
18:23:51 <zodbot> smooge: Approved Groups: @haskell qa @gitliberation-fonts cvsredhat-config-language cla_done fedorabugs cvsfedora +packager sysadmin @cvslohit-fonts sysadmin-cvs cla_redhat cla_fedora @giti18n gitspin-kickstarts sysadmin-build @gitcabal2spec provenpackager sysadmin-web @githaskell-sig proventesters @gityum-langpacks gitreleng
18:24:23 <smooge> yselkowitz, so I would look at rebasing ghc with a BREAKING_STUFF alert around then.
18:24:37 <yselkowitz> so even if we build ghc we won't be able to build ghc-* libs until the macros are updated
18:24:52 <smooge> ah
18:25:00 <smooge> well then I would skip til then :)
18:25:24 <yselkowitz> probably, hopefully we can make plans to do it right after 7.3
18:25:38 <smooge> yep. maybe do an update across things
18:26:54 <smooge> ok anything else here?
18:26:59 <yselkowitz> why we're dependent on a rhel macro for %ghc_arches where ghc isn't in rhel is a different question...
18:27:32 <nirik> well, we could update them in epel-macros and drop that once they land in rhel
18:27:48 <smooge> yselkowitz, I am going to blame openshift
18:28:10 <yselkowitz> lol
18:28:23 <smooge> or openstack
18:29:09 <smooge> nirik, if they appear twice for any length of time would that cause problems with builds?
18:29:31 <smooge> as in they are in epel-macros and 7.3 macros
18:29:40 <orionp> last one wins...
18:29:43 <nirik> only if they are different... ;)
18:29:49 <nirik> and yeah
18:30:09 <smooge> ok. I couldn't remember if it was first one, last one, or "double defined. stop build"
18:30:24 <yselkowitz> all of the above? :-)
18:30:41 <smooge> .whoowns epel-macros
18:30:41 <zodbot> smooge: No such package exists.
18:31:04 <smooge> .whoowns epel-rpm-macros
18:31:04 <zodbot> smooge: No such package exists.
18:31:14 <smooge> .whoowns epel-release
18:31:14 <zodbot> smooge: orphan (stahnma in Fedora EPEL)
18:31:34 <smooge> hey I know what package I need to remove because its an orphan...
18:32:41 <yselkowitz> anyway first step is to agree on what to rebase to
18:32:43 <smooge> ok so if we put them into the epel-macros would that get you able to build.. or do you want to wait until 7.3 to go with a massive upgrade?
18:33:01 <smooge> I would rebase it to what works and would have a "long" shelf life
18:33:06 <yselkowitz> I'd *like* to talk with the maintainer about this
18:33:13 <smooge> yep. sounds good
18:33:23 <yselkowitz> if a major rebase is in order, then probably wait for 7.3?
18:33:48 <yselkowitz> but just rebasing to f21 (same version, more patches) may be doable sooner
18:33:54 <yselkowitz> if it doesn't break abi
18:34:09 <yselkowitz> if it will, then let's do a bigger rebase while we're at it
18:34:14 <yselkowitz> my .02
18:34:39 <smooge> yeah. sounds good. if you don't hear from the maintainer soon.. let us know and we can see if we can get something from them
18:34:40 <nirik> yeah. I haven't seen him around lately, but perhaps he's just traveling
18:35:12 <yselkowitz> bug 1200404, needinfo since 27 april
18:35:31 <nirik> thats a long while. ;( Hope he's ok
18:37:30 <smooge> me too
18:38:05 <yselkowitz> so maybe one of you can get the ball rolling?
18:38:17 <smooge> I will ping him and see if I can get any info from him if he is ok etc
18:38:36 <yselkowitz> thanks smooge
18:38:48 <smooge> #topic Current opportunities
18:38:48 <smooge> #info RHEL-6.8 released. Rebuilds of packages and upgrades
18:38:49 <smooge> #info dnf needs in EL-7
18:38:49 <smooge> #info mirrormanager problems caused issues with CentOS
18:38:49 <smooge> #info ???
18:39:06 <smooge> So 6.8 was released last week and CentOS has caught up and built things.
18:39:28 <smooge> This is the last major change release in the 6 series with it supposed to be stable only fixes from now on.
18:39:56 <nirik> cool.
18:40:01 <smooge> There was a request for a process for people who wanted to update major things and i need to come up with it
18:40:34 <smooge> if I put it out this wekeend, we can get it 'approved' next meeting and give people the go-ahead to do any major changes?
18:40:52 <bstinson> +1
18:41:39 <smooge> The next item that came up was that a bunch of changes went out for dnf in epel
18:42:47 <smooge> there was a request from the dnf maintainers that they wanted an exception granted that they could replace a shipped package in RHEL
18:43:22 <nirik> yeah, there's a fair bit to this saga
18:43:31 <nirik> dnf is in epel7, but not in rhel7
18:43:42 <nirik> libsolv and hawkey are in rhel7
18:43:57 <nirik> they were retired in epel7 but not correctly blocked in koji
18:44:11 <nirik> then someone came along and wanted to update dnf, so they did the entire stack
18:44:32 <nirik> anyhow, I don't know whats happening with this, but I think the dnf folks are pondering options.
18:44:44 <smooge> thank you for the summary. my version still has profanity in it.
18:44:55 <nirik> they may pull dnf into rhel, or do a copr or I don't know what
18:45:14 <tibbs|w> Oh, if Centos 6.8 is out, I need to take the autosetup stuff back out of epel-rpm-macros-8.
18:45:38 <smooge> bstinson, it seemed out when I did a yum update.. am I off the mark?
18:45:51 <bstinson> we're in CR at the moment
18:46:00 <smooge> CR?
18:46:05 <smooge> Critical Release?
18:46:15 <avij> hello, just came back from a real-life meeting that took a bit longer than I had thought
18:46:59 <bstinson> smooge: https://lists.centos.org/pipermail/centos/2016-May/159404.html
18:47:35 <smooge> continuous release
18:47:36 <smooge> got it
18:47:44 <bstinson> basically the packages get released to the CR repository for a period before it's called a full point release
18:47:46 <smooge> tibbs, so I was looking at CR and not the regular release
18:48:34 <tibbs|w> I already built the update; I'm just not pushing it to stable until 6.8 lands in centos.
18:48:40 <smooge> okie dokie
18:48:52 <tibbs|w> Otherwise we'd break mock builds.
18:49:03 <smooge> understood :)
18:49:17 <smooge> ok back to the main saga.. what to do with dnf
18:49:39 <nirik> nothing. Wait until they come up with another approach
18:49:43 <smooge> nirik, did those packages get blocked in koji finally? [and how would I check on this]
18:50:08 <nirik> I opened a bug about it, but they closed it recently... I can try and go block those packages now.
18:50:29 <nirik> oh, it is now
18:50:32 <nirik> koji list-pkgs --package=libsolv --show-blocked
18:50:41 <smooge> oooh thanks
18:50:47 <nirik> same for hawkey
18:51:22 <smooge> ok so with that.. they will need to do a copr. They will probably just drop it from epel since they can't get the update they needed
18:52:46 <smooge> or something
18:53:21 <smooge> ok last issue
18:53:27 <nirik> yeah, or pull dnf into rhel or something... but nothing we need to do anything with
18:53:55 <smooge> mirorrmanager has been having invalid problems which have been causing various issues on epel and copr
18:54:18 <smooge> some of the issues have been bad pkls but the others I don't know about.
18:54:40 * nirik doesn't know about this, haven't seen it...
18:54:51 <bstinson> smooge: pkl?
18:55:21 <nirik> bstinson: the way things work the backend generates a python pkl once an hour and syncs it to mirrorlist servers that serve the data in it out.
18:55:42 <nirik> the last bad one I recall was last monday or something (it's pretty rare and hard to pin down whats causing it)
18:55:52 <bstinson> ah, thanks
18:56:16 <smooge> I got notified about it last week from kbsignh when I had a migraine and I told him to help us figure it out or fork the repo locally. I wasn't the best person at that moment
18:56:40 <smooge> I didn't realize it had caused problems with a rebuild they were doing in their infrastructure.
18:57:36 <nirik> from time to time we have seen one mirrorlist server not update right (so it has old data)... we need to make a check for that
18:58:10 <smooge> the copr people were seeing a problem this morning which seemed to be similar. where the mirrors were outdated so the client is passed a bunch of info it can't build from
18:59:01 <nirik> do ask anyone seeing problems to file tickets so we can track these and gather info
18:59:02 <bstinson> i know we've seen it from time-to-time in the CI infra, happy to send more reports if we can catch it
19:00:30 <smooge> someone sent me some logs and a description of how they do a regular check in nagios to see if it is working or not
19:01:40 <smooge> which I promptly put in a folder somewhere
19:02:45 <smooge> and yes. I will tell them to file tickets from now on.
19:03:14 <smooge> ok we are over meeting time
19:03:31 <smooge> #topic Open Floor
19:03:40 <smooge> ok anything or I can close it out
19:04:07 <smooge> ok closing out
19:04:09 <smooge> #endmeeting