Wednesday, 2026-06-17

mattolivercschwede, acabista you both around in about 30 mins for an irc meeting? No stress, just checking if I need to be at my desk then :) 09:30
acabistamattoliver yes I am!09:32
mattoliver👍️09:33
acabistamattoliver cschwede I think I forgot how to join again :p 10:02
mattoliverwe'll do it here as text.. so nothing just be here :) 10:02
mattoliver#startmeeting swift10:02
opendevmeetMeeting started Wed Jun 17 10:02:44 2026 UTC and is due to finish in 60 minutes.  The chair is mattoliver. Information about MeetBot at http://wiki.debian.org/MeetBot.10:02
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.10:02
opendevmeetThe meeting name has been set to 'swift'10:02
mattoliverwhose here for the swift meeting?10:02
mattoliverThis way it's logged, but we can also do more meetpad meeting too10:03
mattolivercschwede: you around?10:03
mattolivernot too much on the agenda today. 10:03
mattoliverAgenda:10:03
mattoliver#link https://wiki.openstack.org/wiki/Meetings/Swift10:04
mattoliverReally just a touch base and talk about how removing eventlet is going.. so really need cschwede for that :) 10:04
mattoliver#topic eventlet removal10:04
acabistayeah I wanted to ask a feedback about my reviews, if they are being useful and stuff10:05
mattoliveracabista: have you looking into the patches at all?10:05
acabistayeah I've been reviewing a bunch of them10:05
mattolivernice, I love that! 10:05
mattoliverlet me ping cschwede a different way and see if he's around10:06
acabistaI intend to keep doing that, but also I need more people reviewing it, more like core reviewers10:06
mattoliverOK can't see him online so maybe he's nor around. 10:07
mattoliverPlease keep on doing it. Yeah your right, cores and everyone should help more.. I'll try and spend some time digging in, because it is really important for us10:08
mattoliverBecause it's on a feature branch we can just land after some time and not worry too much as it isn't going into a release yet. 10:08
acabistathank you!10:09
mattoliverFor a major change like this, we collaborate and iterate fater, change directions and it doesn't matter too much (though we want to be involved because it's were alot of design decisions are made10:09
acabistaI've been using a different approach setting the tone for the code reviews, saying when it's only a question, a suggestion, a requirement and stuff like that10:09
mattoliverso safe to land patches. Then when we think it's ready someone needs to take the branch and turn it into a bunch of reviewable patches and they get merged into master (ie into swift for reals).10:10
acabistaI think it's helpful since we don't know each other and I don't want anybody to feel like I'm demanding changes10:11
mattoliverso it's 2 phase and the feature branch phase allows alot of merging with out worries10:11
acabistayeah makes sense matt10:11
mattoliveryeah. great idea10:11
acabistaif we are done about eventlet-removal I wanted to bring up the hash location mismatch change as well10:12
mattoliverplease ask questions in the reviews, heck its also ok to leave a 0 and just ask questions. Asking questions is how we've all learnt10:12
mattoliverI still ask questions, swift is a big code base :) 10:12
mattoliveryeah, we can move on.. can always swing back if cschwede turns up.10:13
mattoliver#topic Add quarantine_when_hash_location_mismatch config option10:13
mattoliverHow's your patch coming along acabista ?10:13
mattoliver#link https://review.opendev.org/c/openstack/swift/+/98456210:13
patch-botpatch 984562 - swift - Add quarantine_when_hash_location_mismatch config ... - 10 patch sets10:13
acabistamattoliver do you use Claude? I wrote a different Claude skill that helps me with code reviews, I would love demoing for swift upstream another time because it can help increase the number of reviews, but it's not claude reviewing, it's human reviewing10:14
acabistaso the patch is good I think, I know cschwede is reviewing a lot and it would be great to have one more person reviewing it10:15
mattoliverI love that idea. 10:15
acabistathis patch is really important for us because we keep having customer issues10:15
mattoliverHow about next 1000 meeting we do meetpad?10:15
acabistaperfect10:16
mattoliverI sware I reviewed.. but I don't see it there.. so now I'm really confused and worndered if I dreamt it.. don't see draft comments either.. hmm10:17
acabistathis happened to me the other day too10:17
mattoliverWell hopefully it's still all in my brain anyway, I mean it looking good. I'll make sure I review it as top of my list when I'm back at my desk next10:18
acabistathank you very much!10:18
acabistado you think we are done for today? I don't have anything more to talk about10:20
mattoliverback to skills and agents. We've been playing with a bunch. I have a shared agent agnostic context where I have my skills defined because I use a mix of codex and claude, testing this new era out.. but I'm a little slow to the game. Maybe we should share swift related skills so we can all benefit and speed up swift work! 10:20
mattoliveroh #topic open discussion10:20
mattoliver#topic open discussion10:20
mattoliverI am working on a ring-manager in swift as a swift control-plane. It's coming along nicely and codex and claude are helping me. 10:21
acabistaso I a working with skills lately, trying to learn a lot about them and I created this 3 that are really helpful https://github.com/andressadotpy/claude-skills10:21
mattoliverNice! thanks for sharing! awesome!10:22
acabistabasically what they do is openstack-swift-engineer has all the policies and context for openstack, openstack-swift-engineer has all the basic knowledge for openstack swift and openstack-swift-code-review use both skills10:22
acabistaalso it uses the same framework I talked about, to set the tone for a code review10:23
mattoliversounds amazing! Do you mind if I test them out?10:23
acabistayou call it /openstack-swift-code-review <patch number> and it basically give the summary, curl the comments in the review and stuff like that10:23
acabistaplease test them!10:24
acabistaand if you have any ideas to make it better, feel free to improve them and open a PR10:24
mattoliverI will!10:24
acabistasince I'm just a beginner, I use the skill for openstack-swift a lot to help me understand the code base10:25
acabistait's been really helpful10:25
mattoliverWe'll I'm a beginner to AI agents, so I can definitely learn from you :) 10:25
mattoliverMy ring-manager POC (agenticly driven) is coming along and can be found in my github fork: https://github.com/matthewoliver/swift/tree/ring_manager 10:26
acabistaopen source community is awesome, we are all learning together!10:26
acabistathank you for sharing!10:26
mattoliverstill a work in progress, but the VSAIO repo has a branch that can be used to deploy it out.10:26
mattoliverbasically a new wsgi servvice for control plane that you can hit the api to define and build rings. An agent on each storage node that pulls the latest ring manifest and installs. Even got a builder (for building) and HA for the ring-manager service. And the cLI tool works well. 10:27
mattoliverthere is overivew and api docs in there and a howto_ring_manager as well. Some slides and a basic design doc too (https://github.com/matthewoliver/swift/blob/ring_manager/doc/ring-manager-design.md ) 10:29
acabistathat's awesome!10:29
acabistacan I review it?10:29
mattoliverplease! feel free to have a play and let me know. No doubt there is still rough parts. Still a WIP. But I hope to convince people to add something like it to upstream swift 10:30
mattoliverAlso throwing some tokens at auto-sharding to finally get that finished.. so hopeing we can work on some really cool stuff thanks to our AI friends :) 10:30
mattoliverThere is alot of code there so lots to review, but that's what AI will help with. There are some interesintg bits I haven't fixed yet.. but in my testing it seems to be working great (ring-manager) 10:32
mattoliverAnyway, that's all I have. 10:32
mattoliverI need to spend more time reviewing, but the POC has kind of nerdsniped me :P 10:32
mattoliveracabista: you got anything else?10:33
acabistahahahaha I understand10:33
acabistaI was taking a look in the doc for the ring manager and it looks really good, I'm very interested in it 10:34
acabistaI think after my demo next time I will ask a few questions about it as well, if its ok by you10:34
acabistaI have nothing else for now10:34
mattoliverHowto doc helps alot: https://github.com/matthewoliver/swift/blob/ring_manager/doc/source/howto_ring_manager.rst 10:35
mattoliverCall, then I'll call the meeting. acabista thanks for everything, especially your great ideas and willing to dig in! Love it!10:36
mattoliverAnd as JD always used to say, thanks for coming and thanks for working on Swift!10:36
acabistathank you mattoliver! see you next time, I'm very excited10:36
mattoliver#endmeeting10:36
opendevmeetMeeting ended Wed Jun 17 10:36:32 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)10:36
opendevmeetMinutes:        https://meetings.opendev.org/meetings/swift/2026/swift.2026-06-17-10.02.html10:36
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/swift/2026/swift.2026-06-17-10.02.txt10:36
opendevmeetLog:            https://meetings.opendev.org/meetings/swift/2026/swift.2026-06-17-10.02.log.html10:36
opendevreviewTim Burke proposed openstack/swift master: s3api: 501 unsupported bucket subresources at other levels  https://review.opendev.org/c/openstack/swift/+/99224622:12

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!