21:00:01 <notmyname> #startmeeting swift
21:00:02 <openstack> Meeting started Wed Jun 27 21:00:01 2018 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:05 <openstack> The meeting name has been set to 'swift'
21:00:08 <notmyname> who's here for the swift team meeting?
21:00:12 <timburke> o/
21:00:13 <m_kazuhiro> o/
21:00:27 <rledisez> hi
21:00:32 <mattoliverau> o/
21:00:52 <notmyname> tdasilva: just pinged me and said he's got something else going on but may be able to join in a little bit
21:01:15 <kota_> o/
21:01:29 <acoles> hi
21:01:54 <torgomatic> o/
21:02:06 <notmyname> welcome, everyone
21:02:39 <notmyname> agenda this week just has a few topics on it
21:02:41 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:03:18 <notmyname> #topic python-swiftclient
21:03:45 <notmyname> I was looking at the openstack release cycle dates, and we need to tag a release of python-swiftclient in about a month from now
21:03:56 <timburke> oh, that's probably a good idea. i should clean up https://review.openstack.org/#/c/577584/
21:03:56 <patchbot> patch 577584 - python-swiftclient - Stop leaking quite so many connections
21:03:59 <timburke> and maybe figure out how we can land https://review.openstack.org/#/c/573355/
21:03:59 <patchbot> patch 573355 - python-swiftclient - fix tox python3 overrides
21:04:11 <timburke> and look at some of timur's paches
21:04:12 <notmyname> yeah, getting together some patches we want to land will be useful
21:04:23 <notmyname> timburke: can you add those to the priority reviews page, plese?
21:04:46 <notmyname> right now, there's a few things that have landed that would be good to release, but AFAIK nothing critical
21:05:13 <notmyname> point is, keep swiftclient in mind, and expect a release in about 4 weeks
21:05:31 <notmyname> and it seems timburke (as always) has a pretty good idea of some good, open patches for swiftclient :-)
21:06:22 <notmyname> #topic https://review.openstack.org/#/c/337960/
21:06:23 <patchbot> patch 337960 - swift - Send correct SLO ETag for container updates
21:06:35 <notmyname> clayg: just said a good summary about this patch in -swift
21:07:08 <notmyname> this is an example of something that seems to be a good case for wanting some more eyes on the review, in part because there are some API concerns
21:07:33 <notmyname> from clayg: "A bunch of people think it makes the world better to merge as is. I’ve advocated more work should happen to make the world better but with only additive API changes. Contention seems to be broad support for use-case and weak support for more-work/legacy-api-warts. So we’re stuck. We need a value judgment."
21:08:54 <notmyname> I'll take some time this week to review this patch, but I'd definitely appreciate some other eyes on it. kota_ and m_kazuhiro and zaitcev, your opinions will be valued
21:09:09 <timburke> fwiw, there's a related change for s3api (https://review.openstack.org/#/c/575860/) that's also somewhat impacted, just because it needs to come up with *something* to send as the base etag for the container server
21:09:09 <patchbot> patch 575860 - swift - Include '-' in multipart ETags
21:10:19 <notmyname> timburke: ah, good to know
21:10:30 <kota_> I'll look at it. As well as s3api case, I'm interested in the change.
21:10:57 <notmyname> kota_: thanks
21:11:19 <m_kazuhiro> I'll take time to review it.
21:11:24 <notmyname> m_kazuhiro: thank you
21:11:32 <notmyname> #topic other topics
21:11:44 <notmyname> acoles: zaitcev: how's the put/post patch looking this week?
21:11:44 <timburke> fwiw, they both build on some nice x-object-sysmeta-container-update-override-* work that we'd done for crypto but haven't really used much since then...
21:12:05 <notmyname> timburke: it's really cool to see those mechanisms reused
21:12:36 <acoles> notmyname: I believe zaitcev pushed a new version with my changes squashed in but I have not had chance to look at that yet
21:12:42 <notmyname> ok
21:12:52 <timburke> iirc i came up with the slo patch in part as a means of testing the reusability of the changes for crypto :-)
21:13:18 <acoles> timburke: ah yes, I remember the-longest-header-name-swift-so-far :)
21:13:22 <zaitcev> notmyname: I squashed acoles' work in, and found that we don't have enough unit tests... We're well protected against regressions, at least. New code is not well covered. The functests actually test more of it than unit tests.
21:13:52 <acoles> zaitcev: did you squash in the changes to in process func test jobs too?
21:13:58 <notmyname> zaitcev: ok. are you working on improving unit test coverage now? or is there something else to do first?
21:15:22 <acoles> at last week's meeting I promised to write up options for object server version 'discovery' ... https://etherpad.openstack.org/p/swift-put-post
21:15:23 <zaitcev> acoles: everything, I didn't find much to disagree. There was one place where a comment became senseless because you optimized an assignment away. It was useless as code, but meaningful fore readability. So, only one very minor change from your code.
21:15:44 <acoles> zaitcev: ok, yeh I saw your comment there. that's fine
21:16:22 <zaitcev> notmyname: working... But it can go in too. Or at least other people can play with it safely.
21:16:29 <notmyname> acoles: thanks! I'll review that too
21:17:05 <notmyname> zaitcev: ok, thanks
21:17:14 <acoles> notmyname: IIRC clayg favoured option 2 or 3, maybe he went for 3 in the end, over - -swift a hile back
21:17:27 <acoles> while*
21:17:48 <acoles> people could express preference on the etherpad https://etherpad.openstack.org/p/swift-put-post
21:18:07 <notmyname> that's what I was just about to ask!
21:18:16 <notmyname> is there a link on the review to that etherpad?
21:18:27 <acoles> notmyname: yes, at the top
21:18:47 <notmyname> acoles: no, other way around.
21:18:53 <acoles> oh, sorry, no, other way round
21:18:54 <zaitcev> I saw the PUT+POST, but I don't know what to say. I like the path thing well enough. Some of these options seem like too much trouble for the same result.
21:19:06 <zaitcev> er, saw P+P etherpad
21:19:34 <zaitcev> But I'm not wedded to anything.
21:19:46 <notmyname> let's review it together and see what we like. rledisez, if you've got some time, I'd especially like your perspective on it
21:21:19 <notmyname> meeting summary so far: review PUT/POST etherpad, check out that patch. look at swiftclient reviews. check out the etag api change
21:21:38 <notmyname> which brings me to a topic that isn't on the meeting agenda
21:21:48 <mattoliverau> Review all the things!
21:21:56 <rledisez> sure, I already looked few weeks ago. I should check it again for the last modification. I was a bit disturbed by the file descriptor cache and the /./ trick. the first one seems inevitable. the second one, I don't know. I'd like alecuyer to he here because he looked into it to check how it will interact with losf
21:22:44 <notmyname> rledisez: ok. but especially the migration stuff
21:22:51 <notmyname> unfortunately (for us), both torgomatic and acoles are leaving us to work on other things
21:23:12 <zaitcev> ouch
21:23:13 <mattoliverau> :(
21:23:18 <kota_> oh
21:23:41 <timburke> they will be sorely missed :(
21:23:42 <notmyname> torgomatic: you'll be around for the rest of this week, correct? and acoles, through the end of next?
21:23:49 <torgomatic> that is correct for me
21:24:12 <acoles> notmyname: yes
21:24:53 <notmyname> torgomatic: acoles: I'd like to take this opportunity to publicly, on the record, say that it's been a privilege and honor to work with you for as long as you've been contributing to swift. you have made the project, the community, and myself better. I'm grateful for the time we've worked together, and I hope we can again someday.
21:25:00 <notmyname> and I wish you the best for what's net
21:25:02 <notmyname> *next
21:25:38 <torgomatic> Well, thank you. Much like the word "plethora", that means a lot.
21:25:40 <zaitcev> That fast-post thing was amazine.
21:25:57 <zaitcev> orz
21:26:11 <mattoliverau> I think I speak for us all when I say, it sucks to see you go, but all the best, you both are awesome Devs and people so go thrive and be awesome! And also don't be a stranger and lurk or come say hi at times :)
21:26:57 <zaitcev> I think I learned a lot about the proper way to think about distributed systems just from that one patch.
21:27:23 <acoles> notmyname: thank you. it has likewise been a privilege and huge pleasure to be part of the swift community. I hope not to be a stranger.
21:27:31 <mattoliverau> You know if either of you find yourself in Oz you should pop in ;)
21:27:33 <kota_> +1 mattoliverau. You both works awesome thing. I have learned a lot of things from you.
21:27:37 <acoles> zaitcev: me too :D
21:28:37 <kota_> looks like zaitcev said my thought just short before than me.
21:29:16 <notmyname> it feels a bit inappropriate at this point to just go start talking about patches again for the rest of the hour, so I'll close the meeting now. but we can keep chatting in here or over in -swift
21:29:21 <notmyname> #endmeeting