14:31:29 <mboddu> #startmeeting RELENG (2017-06-12)
14:31:29 <zodbot> Meeting started Mon Jun 12 14:31:29 2017 UTC.  The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:31:29 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:31:29 <zodbot> The meeting name has been set to 'releng_(2017-06-12)'
14:31:29 <mboddu> #meetingname releng
14:31:29 <zodbot> The meeting name has been set to 'releng'
14:31:29 <mboddu> #chair dgilmore nirik tyll sharkcz masta pbrobinson pingou puiterwijk maxamillion mboddu Kellin
14:31:29 <zodbot> Current chairs: Kellin dgilmore masta maxamillion mboddu nirik pbrobinson pingou puiterwijk sharkcz tyll
14:31:29 <mboddu> #topic init process
14:31:55 * nirik is sort of here, sort of still catching up on things.
14:32:45 * threebean waves
14:32:47 * masta is in another meeting, will be half here
14:32:58 <contyk> .hello psabata
14:32:59 <zodbot> contyk: psabata 'Petr Šabata' <psabata@redhat.com>
14:34:03 <dgilmore> hi
14:34:48 <mboddu> Okay, lets start
14:35:10 <mboddu> #topic #6799 [koji] Add configuration for module build service content generator
14:35:17 <mboddu> #link https://pagure.io/releng/issue/6799
14:36:19 <dgilmore> there is nothing that really needs discussed
14:37:00 <dgilmore> we probably need a sop/docs on how and when to do content generation changes in koji
14:37:52 <mboddu> dgilmore: but I heard that content generators require a service running on some other machine which can handle new build types, its not part of koji? Am I right?
14:38:05 <dgilmore> mboddu: yes
14:38:11 <dgilmore> it needs documented
14:38:30 <mboddu> dgilmore: so, who takes care of that stuff? Is it releng or modularity?
14:38:35 <dgilmore> content generators are a way to build content and sput the result into koji
14:38:44 <dgilmore> mboddu: what stuff?
14:38:55 <IgorGnatenko> .hello ignatenkobrain
14:38:56 <zodbot> IgorGnatenko: ignatenkobrain 'Igor Gnatenko' <ignatenko@redhat.com>
14:39:17 <mboddu> dgilmore: the service for creating new builds and the infra related to it
14:39:22 * threebean nods
14:39:28 <threebean> in this case, that is the MBS.
14:39:28 <dgilmore> mboddu: really depends
14:39:40 <dgilmore> mboddu: releng has to make sure it does the right things
14:40:21 <dgilmore> mboddu: we need docs around when and how to setup content generators to try cover the bases
14:41:23 <mboddu> threebean, dgilmore : okay, thanks
14:41:51 <threebean> are there any questions I can help answer here?
14:43:01 <dgilmore> threebean: no
14:43:03 <mboddu> #info mboddu and dgilmore will work on documenting how and when to do content generation changes in koji.
14:43:08 <threebean> +1
14:43:17 <dgilmore> threebean: the main issue I think is lack of process and nohow
14:43:33 <threebean> I get it now :)  was confused at first.
14:43:39 <dgilmore> threebean: its something new that we need to get policy etc on
14:43:42 * threebean nods
14:43:54 <threebean> yeah, and new CGs are rare and special which makes documenting it tricky.
14:43:59 <threebean> ok, thanks.
14:44:14 <maxamillion> .hello maxamillion
14:44:15 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
14:44:18 <maxamillion> sorry I'm late
14:44:46 <dgilmore> bad maxamillion
14:44:48 <dgilmore> :D
14:45:10 <mboddu> #info MBS will take of the new service for new module build type and infra needed for it.
14:45:15 <mboddu> Okay, lets move on
14:45:24 <mboddu> #info #6739 F26 Alpha image paths and CHECKSUM files changed partially from underscore to dash
14:45:31 <mboddu> #link https://pagure.io/releng/issue/6739
14:45:49 <mboddu> We need a better way to handle this
14:45:58 <dgilmore> mboddu: this does not need to be discussed in the meeting
14:46:24 <dgilmore> though we screwed up the fix in Beta
14:46:28 <dgilmore> and made it even worse
14:46:50 <dgilmore> so we need to figure out a way to get what we have to have in the output without needing to do manual work
14:47:03 <mboddu> dgilmore: okay, it was tagged as meeting long back and we recently screwed up again, I thought we can discuss about it now
14:47:10 <dgilmore> mboddu: right
14:47:20 <dgilmore> we should have discussed it 2 months ago
14:47:25 <dgilmore> when it was tagged
14:49:15 <mboddu> dgilmore: I think we did, but afaik we just said that we will make sure we get them right
14:49:18 <mboddu> anyway, lets move on
14:50:03 <mboddu> #topic #6822 using include statements in pungi-fedora
14:50:10 <mboddu> #link https://pagure.io/releng/issue/6822
14:50:12 <contyk> again? :)
14:50:26 <mboddu> contyk: just wanted to know the update, thats it
14:50:41 <mboddu> but seems like lsedler is not around :(
14:51:44 <mboddu> masta: Last time we discussed about it, you said you will update the ticket with how it can be used, please update it when you get a chance, thanks
14:51:49 <mboddu> Lets move on
14:52:29 <dgilmore> mboddu: we talked about this last week and I do not think there is anything more to add currently
14:53:33 <mboddu> dgilmore: yes, I just wanted to check if Lubos has got any update on docs, but he is not around, we can check with later on releng channel
14:53:38 <mboddu> #topic #341 ability to test spin-kickstars by doing throw away image builds in koji
14:53:46 <mboddu> #link https://pagure.io/releng/issue/341
14:54:28 * mboddu reading about it
14:55:09 <dgilmore> mboddu: you mean lubomir?
14:55:57 <dgilmore> mboddu: lsedler's first name is lubomir, its just pronounced lubos
14:56:00 <mboddu> dgilmore: I meant lsedler
14:56:24 <contyk> Lubomír can be shortened to Luboš
14:56:25 <mboddu> dgilmore: okay, thanks
14:56:35 <dgilmore> anyway
14:56:54 <dgilmore> contyk: given we have lkocman and lsedler its best to be clearer :D
14:57:05 <dgilmore> on to the ticket
14:57:14 <contyk> dgilmore: I understand :)
14:57:49 <dgilmore> the main reason historically we have not allowed people to scratch build images is that fedimg would upload all scratch built images to ec2
14:57:54 <dgilmore> it no longer does that
14:58:39 <dgilmore> I think we could allow some limited acces sto the image build permission
14:58:59 <dgilmore> the only concern I have is that people can still do real builds
14:59:16 <dgilmore> I also think we need to document much better how people can test locally as well
15:00:33 <mboddu> dgilmore: If I am not wrong, currently only releng user can create image builds in koji, so cant we just filter them out based on user? Like other than releng every other can only do scratch image builds?
15:01:09 <maxamillion> mboddu: I think there's more users than just that one who can create image builds
15:01:16 <maxamillion> iirc, I can
15:02:09 <dgilmore> mboddu: you are not correct in your understanding
15:02:22 <dgilmore> mboddu: admin users can do image builds
15:02:31 <dgilmore> mboddu: the releng user does most of them
15:02:43 <dgilmore> because they are done as part of the compose process
15:03:29 <dgilmore> $ koji list-history --permission=image
15:03:30 <dgilmore> Tue May  6 17:24:42 2014 permission image granted to masher by ausil [still active]
15:03:32 <dgilmore> Tue May  6 17:25:05 2014 permission image granted to mattdm by ausil
15:03:35 <dgilmore> Tue May 20 13:55:41 2014 permission image granted to imcleod by ausil [still active]
15:03:38 <dgilmore> Wed Feb 11 05:28:59 2015 permission image granted to vpavlin by ausil
15:03:40 <dgilmore> Fri Feb 27 15:04:28 2015 permission image granted to mmcgrath by ausil
15:03:43 <dgilmore> Mon Jan 30 05:58:00 2017 permission image revoked for mmcgrath by ausil
15:03:46 <dgilmore> Mon Jan 30 05:58:22 2017 permission image revoked for mattdm by ausil
15:03:47 <dgilmore> imcleod and masher can do image builds also
15:03:50 <dgilmore> Mon Jan 30 05:58:44 2017 permission image revoked for vpavlin by ausil
15:04:05 <dgilmore> releng does it via admin perms
15:04:45 <dgilmore> mboddu: afaik regular users can not do a scratch build without admin perms
15:04:49 <dgilmore> gashh
15:04:53 <dgilmore> without image perms
15:06:48 <mboddu> dgilmore: okay, understood, thanks
15:07:44 <mboddu> dgilmore: so, how do you think we should handle users creating real image builds? Do you have anything in mind?
15:08:31 <dgilmore> mboddu: I think that we should only give it to people actively involved in the atomic/cloud wg
15:08:45 <dgilmore> at the moment I think only dustymabe would qualify
15:09:05 <maxamillion> +1
15:09:45 <mboddu> dgilmore: okay
15:10:41 * nirik is fine with adding dustymabe there.
15:11:33 <mboddu> #info Currently fedimg supports removing scratch builds based uploading
15:12:54 <dgilmore> #undo
15:12:54 <zodbot> Removing item from minutes: INFO by mboddu at 15:11:33 : Currently fedimg supports removing scratch builds based uploading
15:13:03 <dgilmore> mboddu: thats not correct
15:13:19 <dgilmore> mboddu: fedimg no longer uploads scratch builds
15:13:39 <dgilmore> #info fedimg no longer uploads scratch built images
15:13:41 <mboddu> dgilmore: ah right, thanks
15:14:12 <mboddu> #info We should be careful about who can do the image builds and at the moment it should be people who are actively involved in atomic/cloud wg and that would be dustymabe
15:14:19 <dgilmore> #info The big cost concern over allowing people to scratch built images has been removed.
15:15:02 <dgilmore> #action we need to have some documentation written up on criteria for granting image perms
15:15:21 <dgilmore> people doing spins are restricted also
15:15:38 <dgilmore> but they have never that I can recollect asked for permission to do scratch builds
15:15:59 <mboddu> dgilmore: okay
15:16:27 * mboddu while pagure gets back online, we should cover alternate arches and open floor
15:16:49 <mboddu> #topic Alternative Architectures updates
15:17:04 <mboddu> sharkcz: any updates you want to share?
15:17:29 <sharkcz> yup
15:17:56 <sharkcz> f26 beta fo rs390 is ready, will do one more respin due issues found in the config by dgilmore
15:20:20 <mboddu> sharkcz: thanks for the update, anything else?
15:20:38 <sharkcz> nope, everything else looks good :-)
15:20:57 <sharkcz> except the s390 builder capacity, where we need to ping sysops again
15:21:37 <mboddu> sharkcz: thanks for the update
15:21:38 <nirik> the person we have dealt with on that is on paternity leave... but I exchanged some emails with him and he was going to look at it as he could
15:21:52 <mboddu> thanks nirik
15:22:00 <sharkcz> nirik: thx
15:22:26 <dgilmore> we need a lot more build capacity for armv7 aarch64, ppc64 ppc64le and s390x
15:22:33 <mboddu> #info f26 beta for s390 is ready and sharkcz will do one more respin due to issues found in the config
15:22:51 <dgilmore> sharkcz: how many vms are in the s390 hub?
15:22:51 <mboddu> dgilmore: +1
15:23:13 <sharkcz> dgilmore: s390 koji has 7 vms
15:23:24 <dgilmore> sharkcz: can we cut that to 2?
15:23:45 <dgilmore> and move 5 to primary koji?
15:24:20 <sharkcz> dgilmore: that's the plan, but it's different LPAR so we can't move them easily
15:24:34 <dgilmore> okay
15:24:58 <nirik> We have 5 more aarch64 carts... so we could add 5 more aarch64 / armv7 vm's
15:25:25 <nirik> well, perhaps not... some of those might be defined as buildhw's
15:25:27 <nirik> not sure
15:26:11 * contyk has an open floor question
15:26:19 <dgilmore> nirik: the mass rebuild for gcc showed a big change in behaviour
15:26:23 <dgilmore> and not for the better
15:26:24 <maxamillion> nirik: I read that as "aarch64 certs" and was very confused for a moment
15:27:05 <nirik> dgilmore: yeah, not sure what all we can do without more hw tho...
15:27:27 <dgilmore> nirik: not sure
15:27:38 <nirik> I would really like to retire the calxeda stuff
15:27:42 <dgilmore> I expect the next full mass rebuild will take about twice as long
15:28:05 <dgilmore> nirik: we need to get the buildvm's able to run lorax okay
15:28:18 <nirik> yep
15:28:55 <mboddu> Okay, lets move to Open Floor
15:29:06 <mboddu> #topic Open Floor
15:29:10 <contyk> yay, one minute :)
15:29:12 <mboddu> contyk: go ahead
15:29:30 <contyk> I wanted to ask about process for changes and whether there's something I need to do to get feedback for mine
15:29:46 <contyk> or whether I can just push my change to wrangler even without a formal ack from you guys
15:30:20 <nirik> contyk: which one is this? ticket url?
15:30:25 <contyk> nirik: https://pagure.io/releng/issue/6815
15:31:01 <dgilmore> contyk: last I looked the change was not a change nor close to ready
15:31:13 <contyk> dgilmore: what was missing?
15:31:24 <dgilmore> contyk: you said its incomplete
15:31:31 <dgilmore> contyk: its also in the wrong spot
15:31:41 <contyk> it's formally incomplete but the only thing I'm waiting for is releng ack
15:31:44 <dgilmore> contyk: all changes have to bve under the /Change/ namespace in the wiki
15:31:55 <contyk> ah, wrong url
15:31:58 <dgilmore> contyk: you should be clear about that then
15:32:13 <contyk> the change url is https://fedoraproject.org/wiki/Changes/Host_and_Platform
15:32:19 <dgilmore> I know I diod not pay it much attention other than a quick glace because you said it was incomplete
15:32:22 <contyk> okay; comments like that would have helped in the ticket
15:32:42 * contyk fixes the url
15:32:44 <dgilmore> and I assumed it was still being developed and was waiting on an update saying that it was ready for review
15:33:07 <dgilmore> contyk: sure. I assumeed that it was all part of the development
15:33:21 <dgilmore> and you would move it when it was ready to go in
15:33:21 <contyk> right; well, thanks for clearing that up
15:33:30 <contyk> had no idea why nothing was happening
15:33:41 <dgilmore> because we were waiting on you :D
15:33:49 <contyk> dgilmore: but nobody said so :)
15:33:49 <dgilmore> we should have been clearer also
15:34:06 <dgilmore> contyk: did not feel we had to
15:34:15 <dgilmore> but point taken
15:34:47 <dgilmore> contyk: List of deliverables: Not affected.
15:34:49 <nirik> so, this is just the modules, would there be images made from them or just the modules?
15:34:59 <dgilmore> contyk: i find that extremely difficult to believe
15:35:06 <dgilmore> as its a change to how we deliver things
15:35:26 <contyk> dgilmore: this one is about content, not delivery; similar to how we had Base Runtime change vs. Modular Compose
15:35:34 <dgilmore> contyk: at a glance there is a ton of missing information in the change proposal
15:35:42 <dgilmore> like what you are actually proposing
15:35:53 <dgilmore> what it is that we should get out at the end
15:36:12 <contyk> nirik: this is about some modules merging and some other splitting
15:36:18 <contyk> nirik: it's not about delivery
15:36:51 <nirik> ok, this is just the module part...
15:37:09 <dgilmore> contyk: I think we should have an out of band chat
15:37:20 <contyk> dgilmore: sure
15:37:57 <contyk> I kinda think everything this change is about is in there but I may try making it more clear or something
15:38:06 <dgilmore> contyk: because I do not know what you are currentl;y proposing in the change request
15:38:56 <dgilmore> contyk: I disagree with your statement
15:40:01 <contyk> ok
15:40:27 <mboddu> Okay, we went past the scheduled time, so lets take it to #fedora-releng
15:40:31 <contyk> +1
15:41:08 <mboddu> Thank you all for joining
15:41:11 <mboddu> #endmeeting