17:00:20 #startmeeting RELENG (2018-05-03) 17:00:20 Meeting started Thu May 3 17:00:20 2018 UTC. 17:00:20 This meeting is logged and archived in a public location. 17:00:20 The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:20 Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:00:20 The meeting name has been set to 'releng_(2018-05-03)' 17:00:20 #meetingname releng 17:00:20 #chair dgilmore nirik tyll sharkcz masta pbrobinson pingou puiterwijk maxamillion mboddu Kellin dustymabe 17:00:20 The meeting name has been set to 'releng' 17:00:20 Current chairs: Kellin dgilmore dustymabe masta maxamillion mboddu nirik pbrobinson pingou puiterwijk sharkcz tyll 17:00:21 #topic init process 17:00:41 Hello Everyone :) 17:00:45 morning 17:00:49 Howdy 17:00:57 * sharkcz is here 17:01:25 sharkcz: Hey, long time no see, how are you doing? 17:01:46 mboddu: fine and busy as usual :-) 17:02:19 sharkcz: Good to see you :), but you need to put up with s390x for 1 more month, thats all 17:03:02 :-) 17:03:08 Since sharkcz is here and probably we might not need to do this after 1 month, lets start with alt arch updates 17:03:17 #topic Alternative Architectures updates 17:03:33 sharkcz: ^ anything you want to add? 17:03:53 I have a good news :-) there is new lorax build that's supposed to fix ppc64 and ppc64le composes 17:04:31 otherwise no problems afaict 17:04:50 or rather no new problem 17:04:51 sidetrack: sharkcz: have you seen: https://pagure.io/releng/issue/7481 17:04:53 sharkcz: Awesome, thanks for the update 17:05:21 nirik: ah, I noticed that, but haven't look yet 17:05:33 it's a weird one 17:05:58 yup 17:07:30 #info There is a new lorax build thats supposed to fix ppc64 and ppc64le composes 17:07:36 although I was able to run mock builds against rawhide/ppc64le few days ago ... 17:08:00 sharkcz: So, when f26 goes eol, you dont need to look at s390x koji instance and pushing updates, right? 17:08:07 * dustymabe lurks - in another meeting 17:08:13 mboddu: right 17:08:35 hopefully also we could get some of the secondary resources added to our primary... more builders 17:08:43 and no more dead secondary sigul 17:08:44 nirik: ^ Just FYI, you might need to do some work during that time 17:08:52 sharkcz: haha, true :) 17:09:22 nirik: +1 on moving the secondary resources to primary 17:09:27 sharkcz: I don't miss secondary bits at all ;-) 17:09:45 I think we can retire... 5 machines when that day comes. ;) 17:09:46 pbrobinson: I don't believe you ;-) 17:10:28 sharkcz: I have more than enough replacement tire fires! 17:10:39 :-) 17:11:30 Okay, moving on 17:11:38 * nirik is reminded he was meaning to talk to pbrobinson about moonshot/fedora fun. 17:11:46 have another! :) 17:12:06 nirik: ping me when ever, probably #fedora-arm might be best 17:12:17 can do 17:12:42 #topic #7482 bodhi - unable to request this update for stabilization 17:12:48 #link https://pagure.io/releng/issue/7482 17:13:01 ah, this is mine :-) 17:13:05 So, this is a crazy one 17:13:07 sharkcz: Yes 17:13:25 * mboddu thinking where to start 17:14:02 This is caused by setting post_beta.critpath.num_admin_approvals to 1 as per the docs in https://infrastructure.fedoraproject.org/infra/docs/docs/sysadmin-guide/sops/bodhi.rst 17:14:26 Which we can fix by changing to 0, its not a problem 17:15:01 Now, there is another issue with bodhi that is being looked by bowlofeggs - https://pagure.io/releng/issue/7482#comment-510248 17:15:24 This issue picked up a bug in bodhi 17:15:30 yes, this should be 0 as far as I know. we aren't using proventesters. 17:15:53 shall I make that change and push it out/ 17:15:54 ? 17:15:54 So, my question here is, why is it only set for one release in production.ini file in bodhi? 17:16:11 nirik: I can make change and you can push it out 17:16:19 * mboddu dont have permissions to run the playbook 17:16:32 nirik, mboddu +1 on changing that to 0 17:16:47 also the doc needs work. it's out of date. 17:17:09 well, if we branch f29 and have that, we want different rules for it vs f28/f27 17:17:10 nirik: That, I dont know where the sources are located 17:17:22 pagure.io/infra-docs/ PR's welcome 17:17:29 nirik: I can do that as well 17:17:38 Once I understood the process here 17:17:49 this should be what we implement: https://fedoraproject.org/wiki/Updates_Policy 17:18:13 nirik: So, the overwrites is only for branched release? 17:18:35 nirik: The problem is, bodhi has some default values in its configs 17:18:35 right. pre_beta (but does that make sense anymore when we enable at beta?) and post_beta 17:18:52 nirik: Yes, which is what I am trying to understand 17:19:09 we want the defaults for all stable releases, and only override when there is a branched. 17:19:25 at least thats my hope/understanding 17:19:50 makes sense 17:19:59 nirik: Okay, thats what I am thinking, then should I comment all the lines in the production.ini file or just change that admin approvals to 0? 17:20:15 * nirik looks at whats there now 17:20:38 By all I mean related to the policy 17:20:50 nirik: https://infrastructure.fedoraproject.org/cgit/ansible.git/tree/roles/bodhi2/base/templates/production.ini.j2#n523 17:21:14 I would say: change all those f28's to f29 and comment them all. 17:21:28 then when we enable bodhi for f29 we can comment in the ones we need 17:21:58 oh and change that one back to 0 17:22:44 nirik: Okay, during bodhi enablement to GA of a branched release we will use them 17:23:03 I will check if anything is needed for pre_beta or post_beta based on the policy in wiki 17:23:12 sounds good. 17:26:26 nirik: https://infrastructure.fedoraproject.org/cgit/ansible.git/commit/roles/bodhi2/base/templates/production.ini.j2?id=edc34b52ee796fe08caab6f9cc708d575d88710f 17:26:55 looks good 17:27:24 #info mboddu updated the config and nirik will run the playbook to fix the issue. 17:28:08 #action mboddu will go through the policy and make necessary changes to config and also will update the docs 17:29:03 thanks nirik 17:30:22 nirik: I still have to go through https://pagure.io/releng/issue/7445, if you have already have, I can talk about it or skip it for next week 17:31:33 I looked at it a bit, but again didn't read it in detail. 17:31:45 I think the approach is wrong. I think we should try and make the numbers work, not the name 17:32:05 Okay, I will skip it and update the ticket saying that it will be discussed in the next meeting 17:32:41 Okay, I dont have anything more, lets go to open floor 17:32:46 #topic Open Floor 17:32:53 Anybody got anything? 17:34:07 puiterwijk: Got a min? Just wanted to check on couple things 17:34:14 Did you see the SRPMS vs sources bug? 17:34:40 https://bugzilla.redhat.com/show_bug.cgi?id=1574211 17:35:06 I wish we had some kind of test suite on fedora-repos where it checks all the repo urls and confirms they are valid. 17:35:07 nirik: Yes, which is what I am trying to talk to Patrick 17:35:54 nirik: +1 on test suite 17:36:35 yeah, that would be cool. 17:37:37 of course someone would need to write it. ;) 17:38:20 nirik: thats the tricky part :) 17:38:29 my local mirror of fedora seems to disagree with the issue stated in that bz. weird 17:38:45 I mirror enchilada 17:39:28 masta: That is odd, can you update the ticket with your findings 17:39:34 * mboddu will try it as well 17:39:38 mboddu: maybe after the meeting? 17:39:44 yeah, I'm taking a closer look now 17:39:53 puiterwijk: Sure, thanks :) 17:47:29 masta: Well, I am hitting the same problem with the mirrors but when I try to download the source, its errored with "Error: Failed to synchronize cache for repo 'fedora-source'" 17:48:37 * mboddu did dnf clean a;; 17:48:38 mboddu, yeah, I'm looking at my rsync script, and meanwhile trying to remember where to find the master mirror shares in infra. 17:48:39 all 17:49:17 * nirik needs more ☕ 17:51:40 rsync rsync://mirror.math.princeton.edu/pub/fedora/linux/releases/28/Everything/source/tree/Packages/s/ <-- I see packages 17:52:14 so that is the mirror I go with, but perhaps they don't do --delete in their rsync script. 17:53:58 at one point I was going to put my IP in the master mirror ACL to test this kind of thing. 17:54:14 * bowlofeggs sends nirik a liter of espresso 17:54:45 stat! 17:55:33 🚨 17:55:52 masta: I guess the problem is with updates repo - https://mirror.math.princeton.edu/pub/fedora/linux/updates/28/Everything/ 17:56:03 looks like the reported problem statement is true, the updates are not populating sources 17:56:14 masta: Yes 17:56:18 but the releases are populated correctly 17:56:22 Yup 17:57:00 it's using SRPMS apparently 18:00:40 I seem to remember in a meeting recently there was some discussion about restructuring the mirror layout 18:02:09 There's a request to rename "Everything" to something else... 18:02:31 and yeah, there has been talk about rearranging more things... but as far as I know there's no plan for it yet 18:02:48 okay, good to know. 18:04:49 anyways, I just looking in pub/fedora to confirm , but that was by now redundant. 18:05:33 is this a bodhi config? I'm trying to think where best to investigate? 18:05:48 masta: /pub/fedora has source and SRPMS, but SRPMS has the content 18:06:16 mboddu, that is what I observed, yeah 18:06:18 well, could be: pungi, new-updates-sync or mirrormanager depending on what the fix is. 18:07:01 masta: I think its bodhi, because thats the only one which touches that repo(afaik) 18:07:21 Well, updates that repo 18:07:52 Oh, it went over board, we can take this to #fedora-releng 18:07:58 Sorry guys 18:08:15 #endmeeting