18:00:01 <smooge> #startmeeting Infrastructure (2018-03-15)
18:00:01 <zodbot> Meeting started Thu Mar 15 18:00:01 2018 UTC.  The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:00:01 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:00:01 <zodbot> The meeting name has been set to 'infrastructure_(2018-03-15)'
18:00:01 <smooge> #meetingname infrastructure
18:00:01 <zodbot> The meeting name has been set to 'infrastructure'
18:00:01 <smooge> #topic aloha
18:00:01 <smooge> #chair smooge relrod nirik pingou puiterwijk tflink
18:00:01 <zodbot> Current chairs: nirik pingou puiterwijk relrod smooge tflink
18:00:07 <smooge> Hello all
18:00:39 <nirik> morning
18:00:41 <puiterwijk> .hello2
18:00:42 <zodbot> puiterwijk: puiterwijk 'Patrick "マルタインアンドレアス" Uiterwijk' <puiterwijk@redhat.com>
18:01:15 <cverna> hello o/
18:02:00 <tflink> .hello2
18:02:01 <zodbot> tflink: tflink 'Tim Flink' <tflink@redhat.com>
18:02:50 <smooge> hmm seems like its going to be a quiet meeting
18:03:03 <smooge> I guess nirik is trying to find coffee
18:03:17 <smooge> #topic New folks introductions
18:03:17 <smooge> #info This is a place where people who are interested in Fedora Infrastructure can introduce themselves
18:03:18 <bowlofeggs> .hello2
18:03:20 <zodbot> bowlofeggs: bowlofeggs 'Randy Barlow' <rbarlow@redhat.com>
18:03:42 <nirik> I've got coffee. ;)
18:03:49 <nirik> everyone should
18:03:59 * bowlofeggs only has water in a dirty glass
18:04:12 * pingou partly here
18:04:22 <bowlofeggs> pingou party here
18:04:27 <pingou> so close
18:04:32 <bowlofeggs> haha
18:04:44 <smooge> #topic announcements and information
18:04:44 * relrod checks in
18:04:44 <smooge> #info Infrastructure hackfest in april: https://fedoraproject.org/wiki/Infrastructure_Hackathon_2018
18:04:44 <smooge> #info We are in F28 Beta Freeze! All changes to frozen hosts must get +2 on list before being applied - kevin
18:04:44 <smooge> #info If you maintain any applications that parse NVRs, see the email on modularity on the mailing list - patrick
18:06:13 <smooge> Any other announcements? Everything seems pretty quiet otherwise
18:07:28 <nirik> it's nice and frozen.
18:07:57 <relrod> brrr
18:07:58 <pingou> seems to be a quiet freeze :)
18:08:01 <pingou> for once :)
18:08:30 <smooge> #topic Ticket cleanup
18:08:30 <smooge> #info none this week.
18:08:31 <smooge> #topic Turning off pkgdb
18:08:31 <smooge> #info It has been a depracated service but some things still refer to it
18:08:31 <smooge> #info what needs to be done to finish it off?
18:08:36 <bowlofeggs> not really, modularity has broken bodhi a bunch :/
18:08:52 <pingou> bowlofeggs: true there
18:08:54 <smooge> oh my condolences on that
18:09:10 <smooge> .whoowns pax-utils
18:09:10 <zodbot> smooge: rathann
18:09:15 <puiterwijk> I think I'm almost through the Bodhi side of things.
18:09:17 <pingou> we need to figure out how to populate that new pdc endpoint
18:09:57 <smooge> .whoowns clamav
18:09:57 <zodbot> smooge: sergiomb (robert in Fedora EPEL)
18:09:58 <nirik> right, pdc needs to work for that case and then we need to switch everything to using it. ;(
18:10:04 <nirik> sadly that will take a while I bet.
18:10:16 <nirik> BTW, whoows is still using pkgdb. :)
18:10:19 <smooge> I would like to talk about pdc after this
18:10:51 <smooge> currently who owns pkgdb and what will still break if it is turned off? (zodbot seems to be one)
18:11:09 <tflink> I think we might still be using pkgdb to get a list of critpath packages
18:11:15 <nirik> yes, it will also break apparently a bunch of qa stuff
18:11:31 <nirik> adamw's fedfind also
18:12:32 <smooge> I am not sure what the replacement api is for pkgdb
18:13:01 <smooge> pkgs? pdc? abcdefghij?
18:13:05 <puiterwijk> for most things, pdc.
18:13:16 <puiterwijk> For the other things, it will be pdc, once they get added to it
18:14:06 <pingou> zodbot, mdapi
18:14:28 <pingou> gnome-software is going to be the most annoying one
18:14:47 <pingou> pkgdb is how gnome-software is made aware of pre-release and stable releases
18:15:08 <smooge> ouch
18:15:14 <pingou> puiterwijk: note that we have the new endpoint to populate the release stuff in pdc
18:15:23 <pingou> we just need to populate it :)
18:15:32 * nirik still thinks it could use the same json file that FMW uses.
18:15:50 <pingou> nirik: might be worth
18:16:02 <nirik> but I guess we could also move FMW to use pdc that would be better. ;)
18:16:12 <nirik> less files is better
18:16:22 <smooge> so what drove this is what should zodbot say in the future when someone says .whoowns clamav
18:16:22 <pingou> +1 there
18:17:01 <smooge> since in the past EPEL had different owners and some package owners want NOTHING to do with EPEL
18:17:05 <pingou> would zodbot get that info from pdc?
18:17:17 <adamw> nirik: the FMW JSON files are, I believe, generated with fedfind. :P
18:17:21 <pingou> I think it would get it from pagure + the override repo
18:17:24 <relrod> adamw: they are
18:17:30 <pingou> adamw: problem *solved*
18:17:32 <adamw> so, in a roundabout way, we're back at the same source.
18:17:56 <nirik> adamw: yes indeed
18:17:59 <pingou> adamw: https://xkcd.com/978/
18:18:03 <adamw> once PDC actually has the necessary data, and someone's convinced me there is a process in place by which it will be updated promptly when it ought to be, i can change fedfind to use it.
18:19:03 <nirik> https://github.com/fedora-infra/supybot-fedora/issues/58 <- whoowns bug if anyone wants to fix it... should be able to pull the src.fedoraproject.org/pagure extras data to use.
18:19:23 <adamw> pingou: heh, yes.
18:19:36 <relrod> nirik: I can look at it
18:20:16 <nirik> relrod: cool. Thanks!
18:20:17 * adamw still doesn't really understand why this needed an entire new endpoint as opposed to just being an additional property of releases, or something, but hey.
18:21:00 <smooge> still doesn't really understand pdc, endpoints or properties... so you are ahead of me
18:21:02 <nirik> apparently upstream didn't want to have another state there.
18:21:08 <adamw> smooge: endpoint as in an API endpoint.
18:21:11 <nirik> only development/released
18:21:24 <adamw> oh well, just means we get to do more work when updating, i guess.
18:21:28 <pingou> adamw: I believe because this is going to be a Fedora-only endpoint
18:21:54 <adamw> i don't understand why no-one would want this information about RHEL releases, but hey.
18:22:07 * nirik either, but oh well indeed.
18:23:48 <smooge> So my one thing is I would like to put an end -date for pkgdb
18:24:13 <smooge> can we look at say F28 or F29 release?
18:24:42 <pingou> for F28 w/ gnome-software it may be a little short
18:24:47 <pingou> for F29 sounds doable
18:25:10 <pingou> one thing we could do: dump the JSON of that endpoint, place it on our proxies, redirect the URL to it
18:25:42 <smooge> #info We are looking at pkgdb being turned off after F29.
18:27:45 <adamw> pingou: right, that file does not need to be *served by pkgdb*
18:27:52 <nirik> yeah, we need another place to update after a release
18:27:58 <adamw> it just needs to exist at that URL in that form with correct data, until PDC has sufficient data and significant things have been ported to it.
18:28:12 <pingou> nirik: same as currently :s
18:28:47 <nirik> pingou: yeah. It's titanically sad how many places we change release numbes in per branch event. ;(
18:29:23 <pingou> pdc will fix it all (once we figured out how to update it ^^)
18:29:45 <smooge> ... :)
18:30:34 <smooge> ok anything more here other than it is all a sad sad world :)
18:30:37 <nirik> I think releng and infra could also do better... have a file with releases thats included in various places, etc... instead of committing to 20 scripts the same number
18:31:13 <pingou> nirik: virtual fad topic?
18:31:21 <nirik> perhaps
18:32:14 <adamw> the funny thing is, the bit of fedfind that uses this is meant to avoid *exactly that problem*
18:32:31 <adamw> i carefully set it up so stuff like the openqa scheduler roll over from release to release without me manually going in and changing numbers
18:32:54 <adamw> ...only now we have this problem, i have to instead go and poke other people to manually update the API fedfind queries.
18:32:57 <adamw> oh, irony
18:33:08 * nirik nods.
18:33:17 <pingou> adamw: don't be mean, you didn't have to poke anyone for the F28 branching :)
18:33:31 <adamw> pingou: no, but i did have to poke someone to change f25 to eol :P
18:33:37 <pingou> ^^
18:33:55 <adamw> (which matters because the script that builds disk images for openqa builds some images for 'all current stable' releases, so it was trying to build f25 images that were not needed at all, and failing...)
18:34:39 <smooge> so we need a yaml file everything pulls from which says:
18:34:41 <pingou> anything else on this topic?
18:34:47 <smooge> yeah never mind
18:34:51 <smooge> I will fix this in a ticket
18:34:57 <pingou> thanks
18:35:03 <smooge> #topic Darkserver and tagger retirements
18:35:04 <smooge> #info next steps/action items?
18:35:17 <nirik> yeah, where are we here with these?
18:35:47 <pingou> the discussion on the devel list made me unsure about darkserver
18:35:54 <smooge> sudo -i .../goodnight-sweet-prince.yml -l tagger01:darkserver01
18:35:56 <pingou> there seems to be a usecase for it
18:36:09 <smooge> well these is a usecase but no one to work on it for that usecase
18:36:11 <pingou> but not many people (but one) interested
18:36:25 <smooge> they wanted it, but weren't looking to fix it to do that
18:36:29 * pingou called afk, brb
18:36:30 <smooge> unless I misread it
18:36:34 <pingou> yup
18:36:39 <pingou> and that person has moved team
18:36:47 <smooge> see you later
18:36:54 <nirik> yeah, I think we just set a sunset date and retire it. If someone wants to pick it up, great, they can do so...
18:36:56 <smooge> so I was going to say, we sunset with F28
18:37:30 <nirik> smooge: how about a week after f28? we don't want it to interfere with release stuff/freeze...
18:37:56 <smooge> ok we sunset when the F28 freeze is over
18:38:16 <smooge> my brain was shortcutting to mean that
18:38:20 * pingou back
18:38:33 <pingou> sounds good to me
18:38:42 <pingou> for tagger, I need to reach out to commops
18:38:50 <pingou> we could also retire during the fad no?
18:39:06 <smooge> #info tagger and darkserver will be turned off after the F28 freeze is over.
18:39:41 <smooge> That is in one month with people doing lots of things to get the beta out the door. We could find out commops has their F28 releases somehow tied to tagger
18:39:42 <pingou> tagger is still on the hook
18:39:55 <pingou> it's quite a nice entry doors for new contributors (badges & all)
18:40:07 <smooge> still on the hook?
18:40:16 <pingou> for being fixed vs shut
18:40:30 <cverna> tagger is used by packages during indexing, so that you can search using the tags. But it does not always work :S
18:40:42 <pingou> cverna: good to know
18:41:08 <pingou> so if we turn it off, we should adjust packages as well
18:41:09 <cverna> I think we receive a lot of 404 from tagger during the indexing
18:41:30 <smooge> ok if tagger is not fixed/taken over by F28 end date, it is to be made deprecated
18:41:38 <pingou> sounds good
18:42:48 * pingou eof
18:43:05 <smooge> ok next up
18:43:19 <smooge> #topic Removing old sysadmins
18:43:19 <smooge> #info dgilmore removed himself and mmcgrath from various groups
18:43:19 <smooge> #info we should do some more
18:43:41 <nirik> sure, cleaning up things is good.
18:43:59 <nirik> reminds me I need to clean up the apprentice group again, been busy and haven't in a bit.
18:44:10 <smooge> so I was going to go through the various sysadmin groups and see who had logged into 'things' in the last N months.
18:44:31 <mizdebsk> shall we remove sysadmin-build group?
18:45:04 <smooge> we don't remove groups. We can 'close' it for some definition of 'close'
18:45:22 <nirik> is it no longer used? we didn't use it for a while, but I thought we brought it back...
18:45:55 <mizdebsk> sysadmin-build is used only for pkgs02
18:46:06 <mizdebsk> not for buildsystem, which is confusing
18:46:28 <nirik> I think we used to use it for builders...
18:46:32 <mizdebsk> and most of people in sysadmin-build have access to pkgs02 from other groups, others are not active afaict
18:46:50 <nirik> but now apparently we use sysadmin-releng
18:48:09 <smooge> ok I will add that to the list
18:49:25 <smooge> so I will open tickets on both those items, take the tickets and put in the info. we can clean it up at the FAD or some similar time
18:49:37 <relrod> +1
18:50:05 <dgilmore> smooge: I also removed skvidal :(
18:50:19 <smooge> yeah. it needed to be done
18:50:24 <dgilmore> indeed
18:50:25 <pingou> :(
18:50:27 <dgilmore> still sad
18:50:28 <smooge> but I will have a sad
18:50:34 <nirik> 😢
18:50:56 <smooge> so now that I am sad and depressed..I think I have all I can say on this
18:50:58 <smooge> thanks dgilmore
18:51:05 <dgilmore> nirik: also now owns a ton more groups
18:51:19 <smooge> well I mean really thanks because it needed to be done
18:51:58 <dgilmore> right
18:52:05 <dgilmore> I barely touched the surface
18:52:18 * pingou called afk, for good this time
18:52:23 <pingou> havea nice day everyone!
18:52:25 <nirik> yeah, theres a ton of groups.
18:52:25 <smooge> night
18:52:31 <smooge> #topic Apprentice Open office minutes
18:52:31 <nirik> night pingou thanks for coming
18:52:31 <smooge> #info A time where apprentices may ask for help or look at problems.
18:52:46 <smooge> Any apprentice questins on this?
18:55:32 <smooge> ok.
18:55:36 <smooge> #topic Open Floor
18:55:48 <relrod> So, re zodbot/pkgdb:
18:55:51 <relrod> .what firefox
18:55:51 <zodbot> relrod: firefox: Mozilla Firefox Web browser
18:55:55 <relrod> how do I replace that?
18:56:27 <nirik> query to src.fedoraproject.org perhaps? not sure.
18:56:41 <relrod> nirik: well, pagure's repo summaries don't seem to have the pkg summary
18:57:23 <nirik> query to packages app ?
18:57:35 <cverna> relrod: mdapi has that
18:57:37 <nirik> I don't know that it has any api tho
18:57:38 <relrod> hm, possibly
18:57:44 <nirik> oh yeah, mdapi is a good choice
18:57:49 <cverna> packages gets it from mdapi
18:58:05 <relrod> okay, will look at those
18:58:22 <relrod> packages I think has an api because pkgwat uses it. But I'll compare the two options
18:58:28 <nirik> I guess packages has to have an api since pkgwat uses it.
18:58:32 <nirik> jinx.
18:58:32 <relrod> jinx!
18:58:34 <relrod> hah!
18:58:53 <smooge> ok we are coming up to the hour.
18:58:58 <smooge> anything else for the floor?
18:59:00 * nirik has nothing more.
18:59:07 * relrod neither
18:59:08 <cverna> relrod: pkgwat scrap packages html :\
18:59:16 <relrod> cverna: oh. ew. :(
18:59:21 <smooge> #endmeeting