21:00:01 <timburke> #startmeeting swift
21:00:02 <openstack> Meeting started Wed Jul  3 21:00:01 2019 UTC and is due to finish in 60 minutes.  The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:05 <openstack> The meeting name has been set to 'swift'
21:00:31 <timburke> who's here for the swift meeting?
21:00:36 <kota_> o/
21:01:07 <mattoliverau> o/
21:01:51 <clayg> more like 🥳 time!!!
21:02:24 <timburke> agenda's at
21:02:25 <timburke> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:02:33 <timburke> #topic py3
21:02:44 <timburke> we're soooo clooose!
21:02:57 <kota_> sounds awesome.
21:03:04 <timburke> mattoliverau's started looking at https://review.opendev.org/#/c/662546/ already
21:03:05 <patchbot> patch 662546 - swift - py3: Be able to read and write non-ASCII headers - 5 patch sets
21:03:15 <timburke> which is good, as it's the one i'm more nervous about
21:03:34 <timburke> https://review.opendev.org/#/c/653548/ can be looked at in parallel
21:03:35 <patchbot> patch 653548 - swift - py3: add swift-dsvm-functional-py3 job - 14 patch sets
21:04:03 <timburke> between those two, you can get py2 tests running against py3 services
21:04:17 <timburke> which we start gating on in https://review.opendev.org/#/c/666942/
21:04:18 <patchbot> patch 666942 - swift - Run all func tests under py2 against services runn... - 1 patch set
21:04:25 <alecuyer85> o/
21:05:32 <timburke> i've already put py3 in my release notes - https://review.opendev.org/#/c/668990/ - so let's get those last few landed!
21:05:33 <patchbot> patch 668990 - swift - Authors/changelog for 2.22.0 - 1 patch set
21:05:35 <tdasilva> hello!
21:05:43 <mattoliverau> +1
21:05:49 <kota_> o/
21:05:54 <timburke> who can commit to reviewing patches?
21:07:11 <clayg> i review stuff sometimes
21:07:30 <kota_> 1 patch, I might be able to review in the next week
21:07:47 <timburke> clayg or kota_, want to take on https://review.opendev.org/#/c/653548/ ?
21:07:48 <patchbot> patch 653548 - swift - py3: add swift-dsvm-functional-py3 job - 14 patch sets
21:08:08 <kota_> ok
21:08:14 <timburke> i'm guessing mattoliverau will keep working on https://review.opendev.org/#/c/662546/ ;-)
21:08:15 <patchbot> patch 662546 - swift - py3: Be able to read and write non-ASCII headers - 5 patch sets
21:08:27 <mattoliverau> yup
21:09:06 <timburke> i think i'm ok with moving forward on the release after *just those two*
21:09:55 <timburke> that's about it for py3
21:10:00 <timburke> #topic losf
21:10:39 <alecuyer85> I wrote more tests,
21:10:43 <alecuyer85> https://review.opendev.org/#/c/666378/
21:10:44 <patchbot> patch 666378 - swift (feature/losf) - Add tests for vfile.py - 3 patch sets
21:10:47 <timburke> yay!
21:10:49 <kota_> alecuyer85: nice!!!
21:11:04 <alecuyer85> I think there are a few corner cases I want to handle or test better (a few comments about it in the test code)
21:11:34 <timburke> also, i think our dsvm job isn't happy on the branch any more :-(
21:11:48 <timburke> i've been trying to sort out why
21:11:51 <alecuyer85> oh
21:11:59 <kota_> +1414, -373 :D
21:12:18 <timburke> those RETRY_LIMITs seem pretty consistent
21:12:32 <alecuyer85> Ill look at it also
21:12:52 <timburke> bumped into it while trying to get https://review.opendev.org/#/c/664424/ and https://review.opendev.org/#/c/668007/ in
21:12:53 <patchbot> patch 664424 - swift (feature/losf) - Get new unit tests running under py3 - 2 patch sets
21:12:54 <patchbot> patch 668007 - swift (feature/losf) - Merge remote-tracking branch 'gerrit/master' into ... - 1 patch set
21:12:55 <kota_> i see. did we change something on dsvm?
21:13:18 <kota_> i see
21:13:56 <timburke> i think it's something external to us, but i'm not sure what. that first patch was happy Jun 17, but failed Jun 26
21:14:58 <kota_> ok. let's clarify what causes the problem.
21:16:41 <timburke> http://logs.openstack.org/24/664424/2/gate/swift-dsvm-functional/5f4859d/controller/logs/screen-s-proxy.txt.gz seems to be the cause
21:17:58 <kota_> looks like distribution error on urllib3 which is from... requests module maybe?
21:18:27 <timburke> but the version of requests (which i *think* is the only thing that would bring in urllib3) that gets installed doesn't have that version constraint: https://github.com/kennethreitz/requests/blob/v2.22.0/setup.py#L47
21:18:33 <kota_> i saw the problem recently on requests... but not sure.
21:19:49 <timburke> if anyone has ideas on how to further diagnose that, i'd love to hear them... but maybe after the meeting, in #openstack-swift
21:20:07 <timburke> alecuyer85, is there anything else we should be aware of or helping with?
21:20:21 <clarkb> https://opendev.org/openstack/requirements/src/branch/master/upper-constraints.txt#L190 ?
21:20:57 <alecuyer85> timburke, taking a look at the patch would be great, see if that's going OK even if I know I still need a few changes
21:21:00 <timburke> clarkb, yeah, but requests (https://opendev.org/openstack/requirements/src/branch/master/upper-constraints.txt#L249) should be happy with that
21:21:11 <clarkb> timburke: ya looks like both versions are fine
21:21:18 <clarkb> or rather they don't conflict with each other
21:21:25 <timburke> right
21:22:07 <timburke> alecuyer85, can do
21:22:40 <timburke> #topic autosharding
21:22:52 <timburke> mattoliverau, looks like you've been busy :-)
21:23:10 <mattoliverau> So last week was hackweek at SUSE, so I spent some time playing with a new implementation of auto sharding
21:23:38 <mattoliverau> It's still a WIP but have even written a rambly doc on it (and it's on the ideas page)
21:23:44 <mattoliverau> #link https://docs.google.com/document/d/15yG9zOatwPyXCA3YLNLdNIue9qDR4wLyMPIbypcbALc/edit?usp=sharing
21:24:05 <mattoliverau> Just wanted to bring it to peoples attention.
21:24:38 <mattoliverau> If your interested take a look at the current iplementation. It's just a first parse at the version + index 0 attempt
21:24:39 <alecuyer85> nice doc! thanks
21:24:48 <mattoliverau> ta
21:25:31 <mattoliverau> There is a possible edge case ( 1of many I'm sure) I talk about at the end, and possible ways to eliviate it.
21:26:09 <mattoliverau> not 100% it is a problem, that just comes from a white board session I had had trying to brain storm issues.
21:26:55 <mattoliverau> I need to go and refamilarise myself with the sharding replication changes to really confirm, or better make a probe test :)
21:27:54 <mattoliverau> Anyway, if your interest take a look. I made some interesting asumptions and moves in the patch, which may or may not be right (like adding an ELECT verb to the container-server so they can vote and keep the code simple
21:28:14 <mattoliverau> That's all I got for now :)
21:28:25 <timburke> i like new verbs for new behaviors
21:28:52 <timburke> but then, you may have noticed that from https://review.opendev.org/#/c/659162/ ;-)
21:28:53 <patchbot> patch 659162 - swift - Add operator tool to async-delete some or all obje... (MERGED) - 7 patch sets
21:29:12 <mattoliverau> I did ;)
21:30:21 <timburke> i'll at least read through the doc by next meeting, try to form some opinions
21:30:36 <mattoliverau> ta, can't ask for any more then that ;)
21:30:46 <timburke> mattoliverau, anything else we should know or do?
21:31:13 <mattoliverau> nope, the doc points to the bottom of the patch set (currently of 2 patches)
21:31:37 <mattoliverau> if it interests you in reviewing.. but also feel free to leave comments in the doc.
21:32:05 <timburke> all right
21:32:08 <timburke> #topic open discussion
21:32:14 <timburke> anything else to bring up?
21:34:10 <timburke> so has anyone else ever seen container replicator lockups since we did sharding?
21:34:36 <timburke> i noticed a missing timeout that prompted https://review.opendev.org/#/c/668963/ last night/this morning...
21:34:37 <patchbot> patch 668963 - swift - container-replicator: Add a timeout for get_shard_... - 1 patch set
21:37:26 <timburke> huh. just us, i guess... seems suspicious :-/
21:38:10 <timburke> how does everyone feel about the py3 support? do we think this is going to go well, or terribly, or somewhere in between?
21:38:36 <mattoliverau> in between :P
21:38:43 <alecuyer85> +1 :)
21:39:04 <mattoliverau> but experimental label for a release would make be feel better.
21:39:35 <kota_> going to go well, maybe?
21:40:25 <timburke> i hope it genuinely results in better outcomes :-) i kinda worry that people will see experimental, not really use it yet, and then we won't have the extra testing we wanted
21:40:40 <kota_> the py3 gate prevent me to push incompatible patch. it may make slow down but the compatibility will be required anyway?
21:41:46 <timburke> yeah, then there's that aspect. and it's going to be a bit until we can drop py2
21:42:02 <clayg> Drop it!!!
21:42:30 <timburke> clayg, not *yet*! we have to *at least* have py3 support first!
21:42:41 <mattoliverau> well I hope when we have it running we all could run in SAIO or at least 1 SAIO and iron out any obivous shortcomings :)
21:42:47 <clayg> 😂
21:42:50 <alecuyer85> I think clayg wanted to use tauthon
21:43:06 <clayg> Or that yes.
21:43:41 <clayg> This is fine. When we merge LOSF we’ll have 3 languages.
21:43:44 <alecuyer85> I think we can switch some test clusters to use py3 and hopefully catch some things (if there is anything to catch!)
21:44:27 <timburke> hmm... i should look at dropping the devstack exclusion to make sure swift runs on py2 even when everything else is on py3...
21:44:34 <timburke> alecuyer85, that'll be great! thanks
21:44:56 <clayg> I think we’re going to have a pretty big lift to get py3 going. We might do it on NEW distros only e.g. rhel8
21:45:17 <timburke> i'm sure there's some bugs lurking in the non-client-path
21:46:00 <clayg> But we’ll probably be importing Swift code in py3 as ecosystem stuff starts to add more py3 support. I’ll probably default vsaio to py3
21:46:33 <clayg> This’ll be great.
21:46:38 <timburke> 👍
21:46:43 <kota_> +1
21:47:06 <timburke> all right, does anyone have anything else to bring up?
21:48:09 <timburke> i think i'm going to call it then
21:48:25 <timburke> thank you all for coming, and thank you for working on swift!
21:48:31 <timburke> #endmeeting