18:02:14 #startmeeting Fedora Infrastructure Ops Daily Standup Meeting 18:02:14 Meeting started Fri Apr 3 18:02:14 2020 UTC. 18:02:14 This meeting is logged and archived in a public location. 18:02:14 The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:02:14 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:02:14 The meeting name has been set to 'fedora_infrastructure_ops_daily_standup_meeting' 18:02:14 #chair cverna mboddu nirik smooge 18:02:14 #info meeting is 30 minutes MAX. At the end of 30, its stops 18:02:14 Current chairs: cverna mboddu nirik smooge 18:02:15 #info agenda is at https://board.net/p/fedora-infra-daily 18:02:15 #topic Tickets needing review 18:02:16 #info https://pagure.io/fedora-infrastructure/issues?status=Open&priority=1 18:02:27 ha, no needs review? 18:02:37 * mboddu is here 18:02:40 And I have one 18:03:04 sure, fire away 18:03:21 Well, its already closed, but wanted to check anyway 18:03:26 .releng 9380 18:03:27 mboddu: Issue #9380: Clean up side-tags in koji.stg - releng - Pagure.io - https://pagure.io/releng/issue/9380 18:03:32 hello 18:03:45 Why isn't the clean up script not running in stg? 18:03:51 pagure.issue.comment.added -- kevin commented on ticket fedora-infrastructure#8809: "add carlwgeorge to infra-sig" https://pagure.io/fedora-infrastructure/issue/8809#comment-640017 18:04:11 I couldn't get time to look at it, I wthought of looking later today 18:04:17 there isn't one? 18:04:35 or there wasn't one when we deployed it, there might be one now upstream in koji... 18:05:10 I thought there was one 18:05:12 * mboddu checks 18:05:20 not that I know of. 18:05:41 yeah I don't think there was one :( 18:05:48 but we need one for sure 18:06:06 well, we need to figure out critera, but yeah, we could. 18:06:14 https://pagure.io/koji/issue/2005 18:06:36 Well, https://pagure.io/koji/pull-request/2006 is the fix and I think we deployed it in prod 18:06:54 ha nice 18:07:27 I don't think it is... the plugin merge was not in the 1.20.1 stable relase... only going to land in 1.21 18:08:22 argle bargle. could I get help on https://pagure.io/fedora-infrastructure/issue/8810 my fix is not working 18:08:36 Oh, never mind then, sorry, I thought it went in to prod 18:08:43 But we still to discuss the criteria 18:09:04 yeah... so one thing we could do now is set it to delete them when empty... 18:09:07 I am guessing "--empty-delay 7" should do it 18:09:45 sure, we could do that. 18:09:47 nirik: ^ yeah, this checks if the side tags are empty for last 7 days and deletes them, we could reduce the time 18:09:55 But 7 days seems good enough 18:10:21 well, there's also a remove_empty option to the sidetag plugin, but I am not sure how it works fully 18:11:20 Oh, I didn't know that 18:11:42 Well, anyway, thats all I got 18:11:46 * mboddu updates the ticket 18:11:46 it seems like it could nuke a side tag before builds land in it... 18:12:25 but yeah, we should nuke empty ones... I don't think we should nuke old ones initially at least. Perhaps a report to shame the people who keep them around a long time? 18:13:04 I have some progress to report on the backups volume... 18:13:41 ntap-phx2-c01-fedora01-nfs.storage.phx2.redhat.com:/fedora_backups 30T 28T 1.6T 95% /fedora_backups 18:13:41 ntap-phx2-c01-fedora01-nfs.storage.phx2.redhat.com:/fedora_backups 30T 18T 12T 62% /fedora_backups 18:13:51 Well, if we go with more than 7 days option, people wouldn't mind and they can always create a new side if they want to 18:13:59 and I am still rming. 18:14:22 mboddu: sure, but if we delete ones that have builds in them people might mind... 18:14:35 Well, they are empty, right 18:14:57 They are empty for 7 days 18:15:16 "delete empty tags older than DAYS" 18:15:27 If I understand it correctly 18:15:56 yeah, sorry, I wasn't clear there... when I said " I don't think we should nuke old ones initially at least. Perhaps a report to shame the people who keep them around a long time?" I meant deleting sidetags with builds after N days... 18:16:16 (which the cleanup script can also optionally do) 18:16:33 Oh, yeah, we could that as well 18:16:54 But maybe, we set it for 14 days, just in case 18:16:58 they should really be short lived, so if people are keeping them around a long time we need to ask them why. 18:17:20 I know KDE, GNOME stuff takes more than 14 days 18:17:53 it shouldn't I wouldn't think... 18:18:14 but I could be wrong. we can ask before we push anything. 18:18:45 Back in the days when they used to request side tags, they used to take more than 14 days, and even like python, golang and other languages related side tags take more than 14 days 18:19:20 as for the GNOME side tags, I've been requesting named side tags for GNOME builds to more easily work as a team on the update 18:19:23 sure, something large like that might. 18:19:36 so that people wouldn't have to remember a gibberish name, but instead just know they can use f32-gnome 18:19:40 those should be the exception tho 18:19:48 kalev: sure. makes sense. 18:19:58 if you don't automatically clean up the manually created ones, I think the self-service tag 14 day garbage collection should be fine 18:20:13 kalev: Sure, I am happy to do it, but lot of other people opted for the side tag plugin rather than asking releng 18:20:19 * kalev nods. 18:20:47 there's a exception list if we want on the cleanup script. 18:21:01 and it shouldn't mess with non side* tags anyhow 18:21:59 anyhow, any other tickets to discuss, reviews pending, etc? 18:22:01 maybe send and email to the owner after a week, and then delete a week after? or look at history, if nothing has happened over a week in the tag, only then delete? 18:22:28 nirik: Right, but the problem is identifying them, since the tags are created using fxx-build-sie-blah 18:24:00 Okay, I think everyone agrees on "delete the tags that are empty for more than 7 days" 18:24:17 mboddu: koji knows. ;) 18:24:26 + tags = session.listSideTags() 18:24:55 +1 (as soon as we have the script in 1.21.0 we can look at deploying) 18:25:23 Tags that are created for more than X days, but still not empty, we can think look at how to handle them when we get there 18:25:30 sure 18:26:09 nirik: Sure, koji knows, but it doesn't know what side tag it created for huge python rebuild (which is what we are trying to skip) 18:26:18 ok, if nothing else will close out then... thanks for coming 18:26:30 But, yeah, we can look back at it 18:26:32 mboddu: sure. 18:26:54 * mboddu has nothing else 18:27:08 #endmeeting