21:00:08 <timburke> #startmeeting swift
21:00:09 <openstack> Meeting started Wed Feb 10 21:00:08 2021 UTC and is due to finish in 60 minutes.  The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:10 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:22 <openstack> The meeting name has been set to 'swift'
21:00:24 <timburke> who's here for the swift meeting?
21:00:51 <kota_> hi
21:00:58 <zaitcev> o/
21:01:00 <acoles> o/
21:01:10 <rledisez> o/
21:01:28 <mattoliverau> o/
21:02:11 <timburke> as usual, the agenda's at https://wiki.openstack.org/wiki/Meetings/Swift
21:02:20 <timburke> first up
21:02:28 <timburke> #topic vagrant swift all in one
21:02:37 <timburke> just a quick follow-up from last week
21:02:40 <timburke> it lives!
21:02:46 <timburke> #link https://github.com/NVIDIA/vagrant-swift-all-in-one
21:03:35 <timburke> not really much more to say about it, though i liked the discussion it sparked last week about what we want dev environments looking like in the future
21:03:51 <timburke> #topic sharding patches for train
21:04:15 <timburke> so zaitcev proposed a bunch of patches (great!)
21:04:34 <timburke> but apparently the gate's broken (not so great)
21:04:59 <zaitcev> So... should I mark a few with "WIP" so that Zuul isn't trying to test them?
21:05:00 <timburke> i've proposed a backport to fix it -- https://review.opendev.org/c/openstack/swift/+/774483
21:05:28 <timburke> but that's slowed down by general gate flakiness
21:06:06 <timburke> zaitcev, i wouldn't worry about it. we can recheck the front of the chain once the gate's fixed and just work our way down
21:06:54 <zaitcev> In theory I pre-tested them on my laptop. But in reality I only ran unit tests consistently through evey one in the bunch.
21:07:14 <zaitcev> I'll be checking what Zuul finds in case it's read.
21:07:36 <timburke> in light of the volume of patches, i think i should probably backport some of the retry-tests patches, specifically https://review.opendev.org/c/openstack/swift/+/765589 and https://review.opendev.org/c/openstack/swift/+/766104
21:09:02 <timburke> one lingering question i've got: should we ensure these have been backported to ussuri/victoria? my gut says yes
21:09:21 <zaitcev> Yeah, you don't want gaps.
21:09:31 <zaitcev> That said, I don't know if these retries are needed.
21:10:36 <timburke> all right. next up
21:10:54 <timburke> #topic eventlet, ssl, and dnspython
21:11:36 <timburke> again, a follow-up from last week. i proposed a patch to eventlet at https://github.com/eventlet/eventlet/pull/684
21:12:44 <timburke> it's pretty simple (and doesn't provide full dnspython>=2.0 support) but it seemed to at least unblock zigo
21:13:51 <timburke> bonus: temoto seems on board, so it's likely to merge in the not-too-distant future
21:14:24 <mattoliverau> nice work timburke
21:14:44 <zaitcev> I cannot make heads from tails in that eventlet thing.
21:14:55 <zaitcev> Magic, more magic.
21:16:29 <timburke> the trick was that dnspython started importing ssl (and requests, which would import urllib3, which would import ssl) with v2. since greendns imports dns, it needed a greened ssl
21:16:58 <timburke> #topic system-level read-only role
21:17:31 <timburke> zaitcev recently added this for keystone -- https://review.opendev.org/c/openstack/swift/+/771158
21:17:55 <zaitcev> It took me by surprise how quickly you approved of my implementation. I'm more used to the timeframe of autditor watchers and 40+ revisions over a year or two.
21:17:57 <timburke> it seems pretty useful, so i figured i ought to add it for tempauth -- https://review.opendev.org/c/openstack/swift/+/774539
21:18:03 <zaitcev> Yes
21:18:19 <timburke> it was a remarkably simple change! i liked it
21:19:34 <zaitcev> I already reviewed the tempauth addition.
21:19:42 <timburke> i must admit, i hadn't actually gotten a keystone env up to functionally test it, though. i tried to get the tempauth change to mirror the keystone one pretty well, though, and between the two i feel like we could probably start writing some functional tests with a new test user
21:20:34 <zaitcev> So, how do you guys do it at present? You just give audit account the reseller_admin role?
21:21:11 <zaitcev> I see no other way, and that's what some of my customers were doing.
21:21:14 <timburke> yeah -- it's not great
21:21:54 <timburke> this seems way better
21:21:59 <zaitcev> Right... I heard they get the audit team and cluster ops in the same room and babysit the scripts, then sign it off for a year.
21:22:54 <timburke> acoles, clayg, mattoliverau we should probably add something like this in *our* auth ;-)
21:23:04 <zaitcev> Okay. I hope this does not regress. My review focus was not to affect something when reader is _not_ configured (like today).
21:23:27 <zaitcev> Maybe swauth too :-)
21:24:04 <timburke> https://opendev.org/x/swauth seems dead
21:24:35 <zaitcev> Speaking of bikeshedding, I made my peace with intersection() because we already use it. Interestingly enough it is a method of sets, but it can take a list as argument.
21:25:27 <timburke> i did ultimately port it to py3 at https://github.com/tipabu/swauth -- but i wouldn't claim to be maintaining it
21:25:50 <timburke> #topic py39 support
21:26:16 <timburke> i think we can claim it! so i proposed https://review.opendev.org/c/openstack/swift/+/774749 to do so
21:26:52 <zaitcev> If it's any consolation Go programmers fight the platform shift all the time.
21:27:21 <timburke> only the test-release-openstack job is failing. i think https://review.opendev.org/c/openstack/project-config/+/774906 fixed it
21:27:48 <tosky> timburke: yes, you can recheck
21:28:11 <timburke> 👍
21:28:53 <timburke> #topic shrinking
21:29:02 <timburke> acoles, mattoliverau how's it going?
21:29:38 <acoles> the new swift-manage-shard-ranges compact command has merged, which facilitates manual shrinking of small shards
21:29:57 <mattoliverau> \o/
21:30:08 <acoles> mattoliverau is working on adding recon stats so we know about small shards
21:30:38 <acoles> and we're going to try it out soon on some real life small shards
21:30:55 <mattoliverau> that's https://review.opendev.org/c/openstack/swift/+/772624
21:31:38 <timburke> sounds great
21:31:59 <acoles> the next proposed addition to swift-manage-shard-ranges is 'repair' to fix overlaps https://review.opendev.org/c/openstack/swift/+/765624
21:32:27 <acoles> which takes advantage of the improved shrinking infrastructure we now have
21:33:43 <mattoliverau> with the combination of compact, root driven final collapse and collecting small shards we have the peices of manual shrinking. Repair is an audit when we do the wrong thing. Things are looking great.
21:34:56 <timburke> anything we ought to discuss about the patches?
21:35:37 <mattoliverau> if people want to take a look at the shrink recon patch, the one I linked to, that'll be good
21:35:49 <mattoliverau> we're stil discussion what stats are best.
21:37:13 <mattoliverau> otherwise, climb the tower of patches and review away :)
21:37:36 <timburke> all right
21:37:42 <timburke> #topic relinker
21:38:02 <zaitcev> Oh, brother. I forgot about the relinker.
21:38:40 <timburke> things are progressing. the first three patches in my chain have merged; thanks for all the review acoles and mattoliverau!
21:39:23 <mattoliverau> nps
21:39:27 <timburke> that's about all i had to say about it, though
21:39:34 <timburke> #topic open discussion
21:39:44 <timburke> anything else we ought to discuss this week?
21:41:33 <zaitcev> I don't have anything.
21:42:48 <timburke> all right, we'll let kota_ and mattoliverau get their day started, then
21:43:01 <kota_> thx
21:43:02 <timburke> thank you all for coming, and thank you for working on swift!
21:43:08 <timburke> #endmeeting