21:00:14 #startmeeting swift 21:00:15 Meeting started Wed May 27 21:00:14 2020 UTC and is due to finish in 60 minutes. The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:16 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:18 The meeting name has been set to 'swift' 21:00:20 who's here for the swift meeting? 21:00:33 hello! 21:00:35 o/ 21:00:43 o/ 21:00:45 hi 21:01:15 hi 21:01:52 o/ 21:02:10 as usual, the meeting agenda's at https://wiki.openstack.org/wiki/Meetings/Swift 21:02:14 #topic PTG 21:02:19 it's next week! 21:02:41 🤯 21:03:24 we've got time booked for a zoom room, though it looks like we could also use https://meetpad.opendev.org/swift-ptg-victoria whenever without needing to worry about scheduling conflicts 21:04:15 Cool 21:04:38 it there are any other topics we ought to discuss, write them down on the etherpad 21:04:41 #link https://etherpad.openstack.org/p/swift-ptg-victoria 21:05:14 i'm looking forward to seeing everyone, even remotely :-) 21:05:39 yeah :) 21:05:43 :D 21:06:23 #topic more tests in swift-dsvm jobs 21:06:52 so i've been playing with our gate tests a bit lately 21:07:44 and i've managed to get the number of skips down from like 350 to 20 21:08:24 timburke: nice work! 21:08:59 but i've got a question: does anyone actually want to bother reviewing that? or should i just go ahead and self-approve as things get to passing? 21:09:47 (still using good judgment, of course; if something seems to be getting out of hand, i'll raise a flag and ask for review) 21:09:51 🤔 21:12:17 land that shit! 21:12:19 i'm ok with self-approve, but also feel free to ping me if you want a double-check of a patch 21:12:40 It's testing code and your making it better, I say for now land it :) 21:13:15 If it starts making the gate fail then we can rethink 21:13:33 fwiw, the long-term goal is for us to be able to drop some of the func test variations -- it seems silly to run a whole other job just to check that staticweb works, for example, when we could do it all at once *and* verify that it works with keystone 21:14:41 the main variations we'd want to keep (i think) would be more backend-focused: replica vs ec, encrypted vs not, py2 vs py3 21:15:52 ok, sounds like people are generally on board with landing them and making our testing better :-) 21:16:04 #topic py2 and dsvm 21:16:12 speaking of the dsvm job... 21:16:36 i just realized that our py2/py3 dsvm variations stopped being different 21:17:02 i proposed https://review.opendev.org/#/c/731318/ to drop it down to just ipv6 vs not 21:17:02 patch 731318 - swift - Remove swift-dsvm-functional-py3 job - 1 patch set 21:17:52 devstack dropped support for installing packages on py2 back in https://github.com/openstack/devstack/commit/6b6bdc711 21:18:04 (so, october) 21:19:02 wow 21:19:04 nice catch 21:19:11 the question is: do we care? do we want to invest in maintaining a gate job that runs swift under py2 and keystone under py3? 21:20:20 ouch - that's a tough one 21:20:45 to be clear we're talking about doing work to get back something we lost in October and haven't noticed till now 21:20:51 did we notice because something broke? 21:20:51 or do we say, "to the best of our knowledge, this still works -- we'll fix bugs you report if it doesn't" ? 21:21:00 swift on py2 should certainly still get tested, of course -- but the interaction with keystone hasn't been exercised for a while now 21:21:09 aka "keep doing what's already happening; but on purpose" 21:21:31 oic, the dsvm job is specific to keystone - that's significant 21:21:38 and we ARE testing py3 swift + keystone 🤔 21:21:48 clayg, no, i noticed it (or rather, takashi pointed it out) while i was trying to get a comparable configuration going for storlets 21:22:02 and shops that are deploying modern keystone are already running some kind of py3? like... does keystone still test with py2 (like we do; just not in dsvm jobs) 21:22:19 clayg: no I don't think it does 21:23:02 and of course, i'd expect it to *not work on py2* in the very near future 21:23:05 timburke: and we wouldn't be able to test a "swift + keystone on py2" setup - it'd be "py3 and keystone; but somehow still swift on py2" 21:23:16 yup 21:23:32 similar to what we used to be able to do with devstack 21:23:45 ok, i'm pretty sure I'm advocating for not trying to 21:24:13 tho... maybe we could do everybody on py2 if we took keystone from a stable branch... 🤔 21:24:45 that might be a more reasonable scenario to test - but also possibly still low value 21:25:18 shops that are having a hard time getting off py2 (like us!) are probably already pretty leary of upgrades as projects are dropping py2 support 21:29:11 kota_, alecuyer, mattoliverau, rledisez what are your thoughts? i feel like clayg and i will be very easily convinced to just drop it ;-) 21:29:47 i hate coming down on the side of "do the work" unless i'm going to do the work - i'm not going to do the work in this case 21:30:04 timburke: I think that sounds reasonable 21:30:37 alecuyer: fwiw, aren't you guys on py2 + keystone? 21:30:44 we are 21:30:44 reasonable 21:31:12 But I think the plan is to switch to py3 for both in a short window 21:31:21 I'm leaning on the side of dropping as every distro tends to be dropping py2 anyway. ie focus on py3 and make sure that's working. But can understand legecy users. 21:31:58 happy to have it to support OVH and swiftstack/nvidia users tho. 21:32:13 all right. sounds like we'll let it go, then 21:32:42 alecuyer, that reminds me, we still need to address https://bugs.launchpad.net/swift/+bug/1872553 :-( 21:32:42 Launchpad bug 1872553 in OpenStack Object Storage (swift) "fallocate_reserve is still broken on py3" [Undecided,New] 21:33:18 ohh right the fallocate % :/ 21:34:00 something we can hack on next week ;-) 21:34:11 yes :) 21:34:14 that's all i had 21:34:19 #topic lots of small files 21:34:31 alecuyer, i saw another patchset for the new key format! 21:35:28 Yes it's going slowly :/ So I have no real news, but that slow progress is for lack of time, and we will keep working on it, we will need that for future hardware 21:36:50 got it, no worries 21:36:58 #topic open discussion 21:37:08 what else should we talk about today? 21:37:26 no normal meeting next week, for obvious reasons 21:40:14 all right, let's let kota_, mattoliverau, and seongsoocho start their morning ;-) 21:40:25 thank you all for coming, and thank you for working on swift! 21:40:28 thx 21:40:29 thanks tim 21:40:30 see u next week! 21:40:36 \o/ 21:40:38 #endmeeting