21:00:13 <notmyname> #startmeeting swift
21:00:14 <openstack> Meeting started Wed Mar 14 21:00:13 2018 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:15 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:17 <openstack> The meeting name has been set to 'swift'
21:00:22 <notmyname> who's here for the swift team meeting?
21:00:24 <timburke> o/
21:00:33 <m_kazuhiro> o/
21:00:37 <kei-ichi> o/
21:00:44 <kota_> hello
21:00:56 <torgomatic> o
21:01:11 <rledisez> hi o/
21:01:29 <zaitcev> o/
21:01:39 <notmyname> tdasilva: clayg: ping
21:01:47 <clayg> yup, sorry - i'm here
21:01:53 <clayg> thans
21:01:55 <mathiasb> o/
21:02:15 <notmyname> good morning, afternoon, and evening to everyone
21:02:23 <notmyname> agenda this week is at
21:02:25 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:02:39 <notmyname> there's a few things that have been added, so let's get going
21:02:44 <acoles> hello
21:02:51 <notmyname> #topic s3api
21:02:56 <kota_> my turn
21:03:09 <notmyname> kota_: yep. what's going on with this feature branch
21:03:11 <tdasilva> hello
21:03:40 <kota_> in this week, I've been started to work around cleanup/docs updates to get ready to propose it to the master
21:03:45 <notmyname> #link https://trello.com/b/ZloaZ23t/s3api
21:03:49 <kota_> yes
21:04:02 <notmyname> nice!
21:04:12 <notmyname> kota_: when do you expect to have the merge proposal ready?
21:04:18 <kota_> I categorize the tasks to backlog and future works
21:04:47 <kota_> notmyname: i'm still trying to get my effort in a month since the ptg.
21:05:22 <notmyname> no worries. I understand
21:05:24 <notmyname> in dublin, we agreed that the feature/s3api merge would be to only bring in the existing swift3 codebase. all tests will pass and it will work if enabled, but no changes to any architecture or enabling anything by default
21:05:26 <kota_> talking on trello board
21:05:34 <notmyname> is that still the same plan you're working towards?
21:06:33 <kota_> i'm thinking, I put the items that we don't want to develop on feature/s3api immediately (i.e. will do after merged to the master) in the future work groups
21:06:42 <notmyname> good
21:06:51 <kota_> so please checkout if my collection is correct (or not)
21:06:59 <kota_> notmyname: yes
21:07:29 <kota_> so the remaining tasks in the backlog/in-progress are what I'd like to resolve for a monce.
21:07:33 <kota_> month.
21:07:53 <notmyname> ok. it doesn't look like a lot. (and that's good)
21:08:01 <kota_> that is a summary for the updates
21:08:10 <kota_> and then, one thing I'd like to get your help
21:08:34 <notmyname> we're getting closer to when feature/deep will land something in master, so the biggest concern will be to make sure those two merges don't happen at the same time
21:08:44 <kota_> I had 3 outstanding patches for the work now
21:09:04 <kota_> oic, nice to hear for feature/deep
21:09:08 <kota_> the patches are
21:09:11 <kota_> https://review.openstack.org/#/c/552853/
21:09:11 <patchbot> patch 552853 - swift (feature/s3api) - Update S3api Docs
21:09:26 <kota_> https://review.openstack.org/#/c/552854/
21:09:26 <patchbot> patch 552854 - swift (feature/s3api) - Cleanup swift/common/middleware/s3api/test dir
21:09:53 <kota_> https://review.openstack.org/#/c/551273/
21:09:53 <patchbot> patch 551273 - swift (feature/s3api) - Avoid global CONF instance
21:10:13 <kota_> in particular, I'd like to hear advice on the first one (docs)
21:10:21 <kota_> what's needed there.
21:10:31 <notmyname> ok
21:10:37 <notmyname> I'll take a look
21:10:56 <kota_> because probably I know everything other than you all about s3api so I may be missing something.
21:11:12 <notmyname> anything else for an update on this work?
21:11:33 <kota_> if you don't have any questions :)
21:11:48 <kota_> nothing
21:12:00 <tdasilva> kota_: thanks for driving this work!
21:12:10 <notmyname> +100!
21:12:29 <notmyname> #topic add force_auth_retry mode to python-swiftclient when got errors other than 401
21:12:36 <notmyname> https://review.openstack.org/#/c/551956/
21:12:37 <patchbot> patch 551956 - python-swiftclient - Add force auth retry mode in swiftclient
21:12:41 <notmyname> kota_: this is your topic too
21:13:00 <notmyname> kota_: what is the question we need to answer in this meeting?
21:13:06 <kota_> ah, short answer on the conflict with feature/deep, i don't think it's not so big matter because basically s3api touches on only middleware or docs.
21:13:10 <kota_> yeah
21:13:17 <kota_> the python-client thing
21:13:45 <kota_> I'd like to hear the feels it's better to have on swift community upstream
21:13:53 <kota_> let me describe short history of that.
21:14:02 <kota_> mainly described at the commit message.
21:14:52 <kota_> the problem I'd like to resolve is the case using sort of reverse proxy(e.g. nginx/Load balancer) between the python-swiftclient - swift-proxy-server.
21:15:44 <kota_> the reverse proxy can close the connected socket when it gets error response from swift-proxy.
21:16:34 <kota_> and it causes broken pipe (socket error, EPIPE) in the python-swiftclient, if python-swift client is still attempts to put the http body (e.g. put object)
21:17:15 <notmyname> so the proxy is hiding the 401 response by simply closing the socket, and swiftclient doesn't reauth because it never saw the 401
21:17:33 <kota_> the best way to resolve it, is supporting expext: 100-continue header in the client but it seems to be hard to get in immediately .
21:17:42 <kota_> notmyname: yes
21:17:48 <timburke> i think the 401 gets sent, *but the client hasn't read it*
21:18:04 <kota_> timburke: correct
21:18:32 <kota_> but i don't think we have the way to read it after the socket was closed... right?
21:18:59 <kota_> so the proposal on the patch is to add a mode to force the re-auth on any errors.
21:19:22 <kota_> i know, it's not straight forward to resolve the problem, just mitigation.
21:19:23 <timburke> not sure -- i haven't had a chance to play around with it...
21:19:24 <notmyname> ok, that seems reasonable
21:19:46 <notmyname> sounds like it was a fun issue to debug
21:19:52 <notmyname> "fun"
21:20:01 <notmyname> clayg already has a +2 on it
21:20:07 <kota_> so my point is that.
21:20:08 <clayg> heck yeah!
21:20:21 <kota_> thanks clayg!
21:20:58 <notmyname> matt left an earlier review, but he's not hear right now. perhaps he'll be able to pick it up again and give another review?
21:21:15 <kota_> maybe
21:21:48 <kota_> I'll poke him later again
21:21:51 <notmyname> I said something to him in -swift about it, too
21:21:55 <notmyname> kota_: thanks
21:22:21 <notmyname> ok, on the theme of bugs, I'm going to skip around just briefly... bear with me kei-ichi
21:22:40 <notmyname> #topic multiple or bad url quoting
21:22:49 <notmyname> https://bugs.launchpad.net/swift/+bug/1755554
21:22:50 <openstack> Launchpad bug 1755554 in OpenStack Object Storage (swift) "Percent signs in object names cause trouble for versioned_writes" [High,Confirmed]
21:22:50 <kei-ichi> np
21:22:57 <notmyname> we found this bug yesterday
21:23:22 <notmyname> timburke gave a good summary of the issue
21:23:43 <timburke> "we suck at URL quoting"
21:23:56 <clayg> 🤣
21:23:57 <notmyname> I wanted to bring it up in the meeting because (1) I'd love to find someone who will volunteer to work on a fix and (2) ask if it should be "critical" instead of "high"
21:24:03 <notmyname> timburke: perfect summary :-)
21:24:28 <clayg> idk, it's not a regression - it's just always been this way hasn't it?
21:24:55 <notmyname> I don't know. perhaps the rewrite to versioned_middleware broke it?
21:25:29 <timburke> i still have torgomatic's https://gist.github.com/smerritt/ca5a6701d46c55b43059e4d0e302c2f9 up in a tab... in case i ever find myself with a surplus of time...
21:25:56 <timburke> the big bad (imo) is that versioning flat out *doesn't work* for some objects
21:26:17 <timburke> you think you've got versioning enabled, go overwrite an object *and it's gone*
21:26:30 <clayg> not great
21:26:31 <kota_> too bad
21:26:32 <notmyname> yeah, that's where I get the question of "critical"
21:27:28 <notmyname> really, if we've got someone working on a patch, the difference between high and critical is holding a release for the critical patch to land. and, to be fair, we don't have a release scheduled soon (just lots of other stuff going on)
21:28:34 <notmyname> so the most important question is this: "can someone volunteer to work on a patch for this bug?"
21:29:07 <notmyname> (please, not everyone at once)
21:29:20 <clayg> 👃👈
21:29:29 <timburke> i might be able to? but almost certainly not this week
21:29:55 <kota_> me too, it might be able to.
21:30:09 <notmyname> ok. timburke and kota_, thank you
21:30:37 <notmyname> I'll leave it on the agenda for next week, and we'll see if you've had a chance to look at it
21:30:55 <notmyname> #topic slogging
21:31:00 <notmyname> last topic on the agenda
21:31:03 <kei-ichi> This is mine.
21:31:10 <kei-ichi> I would like to confirm that which channel should I use to inform slogging event.
21:31:18 <kei-ichi> Firstly I plan to create new channel named "#openstack-slogging" and sent patch to project-config repository. --> plan-1
21:31:19 <kei-ichi> Now infra-team give me a comment that is it really mandatory.(send to swift channel --> plan-2)
21:31:23 <notmyname> if you remember from dublin, we talked with kei-ichi about him moving slogging into an openstack namespace
21:31:33 <kei-ichi> yes
21:32:23 <kei-ichi> this is my patch to openstack-infra/project-config  https://review.openstack.org/552282
21:32:24 <patchbot> patch 552282 - openstack-infra/project-config - Add slogging projects
21:32:31 <notmyname> here's a question to everyone: would anyone object to having slogging patches reported in the #openstack-swift channel? this implies that discussion about those patches may also happen in the -swift channel
21:32:42 <acoles> no
21:32:53 <acoles> no objection I mean
21:33:03 <kei-ichi> I think, this time slogging will be created as separate repository from swift so I think IRC channel should also be separated. (means plan-1)
21:33:37 <kei-ichi> I would like to hear about this point to swift team prior to fixing patch.
21:33:48 <notmyname> kei-ichi: I believe the -infra team has concerns about the number of channels that they manage and that are supported by the IRC server and that's why the like to consolidate when possible
21:34:04 <notmyname> eg a bot can only be in a certain number of channels at a time
21:34:08 <kei-ichi> I understood.
21:34:25 <tdasilva> i think it makes sense to have it in the -swift channel
21:34:33 <timburke> it seems like the target audiences would be similar, as well. i know we've gotten a fair number of swift3 questions in #openstack-swift, despite there being a separate #swift3 channel
21:35:04 <notmyname> I agree that slogging should be separate from swift in governance (eg not relying on swift-core for patches or releases), but I would almost prefer to have slogging updates in -swift instead of separate
21:35:09 <kota_> timburke: +1
21:35:13 <notmyname> ...base on exactly what timburke just said
21:35:46 <kei-ichi> OK, so same as liberasurecode, can I use -swift channel to inform ?
21:35:48 <tdasilva> which means we should shutdown swift3 channel at some point??
21:35:55 <clayg> tdasilva: +1
21:36:02 <notmyname> tdasilva: we should do that anyway because of feature/s3api
21:36:08 <tdasilva> right
21:36:10 <kota_> tdasilva: after, we get merged feature/s3api
21:36:18 <notmyname> kei-ichi: are you ok with having your updates go to -swift?
21:36:21 * tdasilva has considered shutting down swiftonfile too
21:36:29 <notmyname> tdasilva: there's a swiftonfile? ;-)
21:36:30 <clayg> i think more traffic in openstack-swift is generally a good thing - I don't mind a variety of topics.
21:36:35 <notmyname> clayg: yes!!
21:36:38 <kei-ichi> yes, patch is really simple, so it is easy to fix.
21:36:51 <notmyname> I don't expect slogging to have enough activity to overwhelm us in -swift
21:37:28 <kei-ichi> I think so. So I'll fix patch so that information will be sent to  -swift channel
21:37:42 <kei-ichi> would that be okay?
21:37:46 <notmyname> kei-ichi: so what I'm hearing from everyone, not only are we happy to see you work on a project related to swift so that more people can use it, we want you to do it with us!
21:37:52 <notmyname> kei-ichi: yes!
21:38:07 <kei-ichi> Thank you notmyname !
21:38:16 <notmyname> #topic open discussion
21:38:21 <notmyname> anything else to cover this week?
21:39:03 <notmyname> I take that as a "no"
21:39:18 <timburke> i've said it repeatedly already, but just wanted to give another thank you to acoles for getting sharding probe tests going in zuul v3!
21:39:30 <notmyname> thank you everyone for your work on swift
21:39:35 <notmyname> #endmeeting