16:38:24 <dgilmore> #startmeeting
16:38:24 <zodbot> Meeting started Mon Jan 20 16:38:24 2014 UTC.  The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:38:24 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:38:30 <dgilmore> #meetingname releng
16:38:30 <zodbot> The meeting name has been set to 'releng'
16:38:39 <dgilmore> #topic whos here
16:38:46 <janeznemanic> hi guys
16:38:49 <dwa> ack
16:38:49 <tyll> Hi
16:39:00 * sharkcz here
16:39:07 <nirik> I'm here, but as usual, monday mornings are a horrible time for me to try and be in a meeting. :)
16:39:52 <dgilmore> #topic tickets
16:39:54 <dgilmore> https://fedorahosted.org/rel-eng/report/10
16:40:05 <dgilmore> there is no tickets for the meeting this week
16:40:18 <dgilmore> is there any that people want to bring up?
16:40:21 <sharkcz> I have https://fedorahosted.org/rel-eng/ticket/5835
16:40:33 <dgilmore> sharkcz: you need to add the meeting keyword
16:40:50 <sharkcz> ah, yes, was going to ask
16:40:57 <dgilmore> sharkcz: thats due to me accidently removing the f18 tags
16:41:02 <dgilmore> i ran teh wrong command
16:41:10 <dgilmore> koji shouldnt have let me do it
16:42:01 <sharkcz> can we somehow sync the pkg list back to s390 and ppc?
16:42:31 <dgilmore> its already been done
16:42:40 <sharkcz> in primary?
16:42:57 <dgilmore> no
16:43:03 <dgilmore> on the secondaries
16:43:11 <dgilmore> it was only the secondaries effected
16:43:55 <sharkcz> but I still see no packages under f18 or f18-updates
16:44:23 <dgilmore> ill look at it later
16:44:26 <sharkcz> ok
16:44:42 <dgilmore> I ran a manual sync of packages to f18
16:45:06 <dgilmore> since the automated syncing was moved to f19
16:46:20 <tyll> I have a question that might be of interest for everyone: Why don't we assign tickets in releng to persons?
16:46:47 <dgilmore> tyll: we dont assign them to anyone because they dont have to be done by one person
16:47:12 <dgilmore> as soon as you assign a ticket to a person they no longer go to the list
16:47:23 <dgilmore> I may do the start of the ticket
16:47:34 <dgilmore> but anyone else can finish it
16:47:54 <dgilmore> assigning removes all visability to the ticket
16:48:01 <tyll> hm, but this way there is also no indicaton about who is working on something
16:48:14 <tyll> or which tickets are not worked on
16:48:22 <nirik> we could actually fix that.
16:48:34 <nirik> we could make it always bcc or cc the list on all tickets.
16:48:51 <dgilmore> take https://fedorahosted.org/rel-eng/ticket/5831 i setup the tags/target but when its done anyone can remove the target and move the builds into rawhide
16:49:33 <dgilmore> tyll: I personally do not think most tickets need assigned
16:50:09 <dgilmore> i dont see that assigning a ticket gives us anything.
16:50:12 <tyll> dgilmore: I agree that tasks do not need to be assigned, but most of the open tickets are more like feature request or long-term changes
16:50:27 <dgilmore> happy to change things if i can be convinced there is value in doing so
16:51:34 <tyll> dgilmore: imho it might help for people to find a task to work on, because it is clear which are currently not worked on
16:54:59 <dgilmore> tyll: how though. I realise you need to look at the ticket to see what if anythings been done on it today. In the past people took tickets and never followed up and finished things
16:55:13 <dgilmore> there was a lot of tickets in limbo
16:55:33 <dgilmore> not assigning means its more likely to be caught. at least thats my opionion
16:55:52 <dgilmore> I know when a ticket had been assigned id not bothered to look at
16:56:20 <tyll> I see, I will think about this
16:56:48 <tyll> btw. there are some old tickets, that probably can be closed
16:56:57 <dgilmore> i agree if its something specifc assigning can be a good thing
16:57:11 <dgilmore> but most tickets really are not that way
16:57:24 <dgilmore> probably
16:57:52 <tyll> It might be true for most tickets that are opened but less for tickets that are still open
16:58:00 <dgilmore> we need to do better at dealing with tickets
16:58:40 <tyll> Shall we look at some tickets now as we did two weeks ago?
16:58:45 <dgilmore> a lot of the open tickets dont really have a clear way to deal with them
16:58:49 <dgilmore> sure
16:59:02 <dgilmore> https://fedorahosted.org/rel-eng/report/1
16:59:06 <dgilmore> is teh open tickets
16:59:16 <tyll> https://fedorahosted.org/rel-eng/ticket/5808 - seems to be done as far as I understand
17:01:00 <dgilmore> yeah
17:01:06 <dgilmore> i fogot to go and close it
17:01:13 <dgilmore> but anyone could have done so
17:01:54 <dgilmore> https://fedorahosted.org/rel-eng/ticket/1501
17:02:16 <dgilmore> probably needs reassigned. but there is not a clear way to resolve it
17:03:19 * masta is here
17:03:47 <masta> sry, was heads-down in other things and in the zone... now I'm here
17:04:21 <dgilmore> https://fedorahosted.org/rel-eng/ticket/5362 needs analysis to see if it would break anything
17:05:22 <tyll> https://fedorahosted.org/rel-eng/ticket/5757 https://fedorahosted.org/rel-eng/ticket/5761 are requests for f20 tags for gnome and kde - I assume the tickets can be closed as well
17:05:33 <dgilmore> https://fedorahosted.org/rel-eng/ticket/3903 needs feedback from abadger1999
17:05:44 <dgilmore> tyll: you cant assume taht
17:05:47 * abadger1999 clicks
17:06:25 <dgilmore> tyll: we need to ping them, see if they are needed still. if not then remove the targets and tag builds into updates-pending
17:06:58 <dgilmore> tyll: KDE tends to always have side tags/targets because it simplifies there constant version bumps
17:07:05 <abadger1999> Ugh yeah --
17:07:50 <abadger1999> making the query that will do that is complex and I keep forgetting what I've already figured out before getting to the end of it.
17:07:55 <tyll> dgilmore: I see
17:08:34 <dgilmore> abadger1999: :( yeah. it looks messy
17:08:49 <dgilmore> tyll: likely gnome can be cleaned up
17:09:08 <dgilmore> rdieter: tends to manage the KDE side of things, he can do so himself
17:09:09 <abadger1999> The email requirement is where it gets messy: https://fedorahosted.org/rel-eng/ticket/3903#comment:10
17:09:20 <dgilmore> though he doesnt always do so with a ticket
17:10:12 <dgilmore> abadger1999: :( thats to notify that the owner has changed?
17:11:04 <abadger1999> dgilmore: Yeah.  The SQL to perform the actual change isn't nearly as hard as figuring out the list of people to notify.
17:11:27 <dgilmore> and we want one email not 4
17:13:16 <dgilmore> some of the git tickets could do with someone to take ownership over.
17:13:31 <tyll> dgilmore: at least the builds in the f20-gnome tag seem to be already in f20/f20-updates-testing/f20-updates - so I guess nothing needs to be merged back, only the tag removed
17:13:46 <dgilmore> there are some things people would like to do in git, that we should try to enable
17:14:09 <dgilmore> we need to look at updating gitolite and generally be more active in its workings
17:15:01 <dgilmore> tyll: they may have done so themselves and not updated the ticket to have the target removed. to close the ticket we need to ping them and remove the target then we can close the ticket
17:15:51 <tyll> dgilmore: in the gnome ticket kalev wrote it can be removed
17:16:01 <tyll> dgilmore: But I do not know how and do not find a SOP for it
17:16:53 <dgilmore> https://fedoraproject.org/wiki/Adding_Side_Build_Targets_SOP i guess we should add something there
17:17:09 <dgilmore> koji remove-target f20-gnome
17:19:59 <tyll> dgilmore: does the tag need to be removed as well?
17:21:21 <dgilmore> tyll: no
17:21:25 <dgilmore> we do not remove tags
17:23:52 <dgilmore> tyll: https://fedoraproject.org/wiki/Adding_Side_Build_Targets_SOP#cleanup
17:24:16 <dgilmore> any other tickets people want to talk about?
17:24:44 <tyll> dgilmore: thanks
17:25:07 <tyll> dgilmore: about the repo md signing, is the problem that this would be a manual step after pushing?
17:25:25 <dgilmore> tyll: yes
17:26:06 <dgilmore> tyll: one the push was done and prepped, someone would need to sign the metadata then tigger the push to the mirrors
17:27:32 <dgilmore> it could introduce an extended delay
17:27:36 <tyll> dgilmore: ok
17:28:38 <dgilmore> we really would need a good way to notify that the metadata is ready for signing
17:28:57 <dgilmore> id not want to keep the creditnals stored in memory for the duration of the push
17:29:36 * nirik wishes we could get some developer time on sigul, to make it more flexable on signing.
17:29:59 <dgilmore> sigul really needs a lot of work
17:30:10 <dgilmore> everything needs a lot of work
17:30:37 * nirik nods.
17:30:38 <tyll> dgilmore: it can be signed as soon as it appears
17:31:16 <dgilmore> tyll: yes, but it appears hours after the push is started
17:31:30 <dgilmore> usually about 3 or 4 hours later
17:31:41 <dwa> actually after doing a few fedora updates pushes my wish is that bodhi would just store the list of updates we fetch the first time running the command and re-use that list after signing, so we don't have to dance around people adding new updates while we sign them
17:32:07 <dgilmore> dwa: bodhi is full of bugs on the backend and clunky to work with
17:32:33 <dgilmore> dwa: lmacken really dooesnt get enough time to work on it
17:33:18 <lmacken> dgilmore: what bugs are you speaking of? I'll try and prioritize them if you file/tell me about them
17:33:47 <lmacken> dwa: that sounds like a reasonable RFE that could be fixed pretty easily
17:34:08 <dgilmore> lmacken: tagging issues when multiple builds go stable at once for instance
17:34:32 <dgilmore> lmacken: bodhi really needs to be aware of signing processes
17:34:42 <dwa> lmacken: I'll write up a bz for it :)
17:34:51 <dgilmore> lmacken: queueing changes when a push is in progress
17:35:00 <dgilmore> so you dont hit snags if a mash fails
17:35:02 <lmacken> dwa: thanks
17:35:09 <dgilmore> things like builds going stable mid push
17:35:24 <lmacken> dgilmore: well, it should be tagging multiple builds for the same package in the same push in the correct order...
17:35:59 <dgilmore> lmacken: there is a bunch of issues we hit regullary, and they have been happening for years so im kinda used to them but really need fixed
17:36:13 <dgilmore> im sure they have all been reported at some point in time
17:36:29 <dgilmore> lmacken: bodhi always tags them in the wrong order
17:36:49 <dgilmore> the oldest build always ends up latest
17:36:59 <lmacken> dgilmore: the ordering issue should have been resolved a little while ago, but I would definitely appreciate bugs instead of just ignoring/complaining about them
17:36:59 <dgilmore> at least it happens a lot
17:37:07 <dgilmore> lmacken: its not been
17:37:54 <dgilmore> lmacken: I know you have said a bunch of them are fixed in bodhi 2
17:37:58 <lmacken> well, I have cycles for bodhi now, so please put important issues on my radar.
17:38:35 <dgilmore> okay
17:39:02 <dgilmore> so we are over an hour now
17:39:07 <dgilmore> anything else?
17:39:22 <dgilmore> #topic followup from last week
17:39:59 * dgilmore wanted to follow up from last weeks discussion on moving the git processing into releng
17:41:39 <dgilmore> i contacted the people in cvsadmin and the response was positive. limb, tibbs and peterson are now releng members and the script for proccesing the git queue along with its history has been merged into the releng git repo
17:42:09 <dgilmore> limb will be attending meetings from next week
17:42:38 <dgilmore> its one of those strange public holidays in the US today that only some people get off
17:42:45 <dgilmore> eom
17:42:56 <dgilmore> #topic open floor
17:43:11 <dgilmore> does anyone have anything to bring up?
17:43:33 * dgilmore knows he needs ot get an agenda out for the meetings
17:44:20 <dgilmore> if nothing ill close up in a minute
17:44:25 <dwa> Nothing to discuss here. FWIW Fedora for Power updated koji/all of our build systems last week.
17:44:39 <sharkcz> FYI I rebuilt the s390 builders last week
17:44:48 <tyll> There were two annoncements for the meeting on the list at the same time
17:44:48 <dgilmore> dwa: sharkcz: cool
17:44:58 <dwa> popular time for maintenance :)
17:45:07 <dgilmore> tyll: i dont know what fedocal did there
17:47:13 <tyll> dgilmore: ah, ok, I thought maybe the lists address was added to the meeting twice
17:47:21 <dgilmore> tyll: its supposed to send a reminder 12 hours before
17:48:05 <dgilmore> ill have to file a bug about it sending it twice
17:48:35 <dwa> lmacken: https://bugzilla.redhat.com/show_bug.cgi?id=1055663
17:48:45 <dwa> (rfe)
17:49:01 <dgilmore> okay, well if we have nothing else.
17:49:52 <dgilmore> #endmeeting