21:00:57 <notmyname> #startmeeting swift
21:00:58 <openstack> Meeting started Wed Aug  8 21:00:57 2018 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:59 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:01:01 <openstack> The meeting name has been set to 'swift'
21:01:04 <notmyname> who's here for the swift team meeting
21:01:05 <timburke> o/
21:01:05 <notmyname> ?
21:01:31 <kota_> hi
21:01:46 <tdasilva> hello
21:02:22 <notmyname> hmm... anyone else coming this week?
21:03:08 <notmyname> hello zaitcev
21:03:10 <notmyname> rledisez: ping?
21:04:05 <notmyname> ok, we can get started. just a few things to go over, then we should have open discussion for anything that needs brought up, then we'll be done
21:04:14 <notmyname> #topic updates, release, priorities
21:04:30 <notmyname> I added a bunch of stuff to the priority reviews page...
21:04:31 <notmyname> #link https://wiki.openstack.org/wiki/Swift/PriorityReviews
21:04:42 <notmyname> this is my thinking...
21:05:04 <notmyname> for the openstack release cycle, we need to tag a release during the week of August 20 (two weeks from now)
21:05:15 <notmyname> sooner is better, of course, in case any issues come up
21:05:28 <notmyname> however, looking at what's outstanding, I updated the page
21:06:04 <notmyname> swiftstack's current short-term priority is to get the multi-key KMIP stuff landed and the s3api patches landed. for the whole community, I would like to see PUT+POST land, too
21:06:27 <notmyname> also, I know https://review.openstack.org/#/c/447129 has been open for a long time, and I think it would be nice to see it included
21:06:28 <patchbot> patch 447129 - swift - Configure diskfile per storage policy
21:06:42 <notmyname> however, just that right there is a big list to get done in the next 1.5 weeks
21:07:22 <notmyname> those of us at swiftstack are focussing on the s3api patches and the kmip ones. and I hope we can help with progress on put+post, but I also hope others can help there too
21:08:06 <notmyname> later this week, I'll work on the authors/changelog patch, so that it's near-done and can be landed quickly
21:08:29 <notmyname> I'm taking some time off next week (before my kids go back to school), so I want to have that part nearly done before I leave
21:08:44 <notmyname> (this also implies skipping next week's meeting, unless someone else volunteers to run it)
21:08:58 <notmyname> any questions or comments on any of that?
21:09:32 <zaitcev> Nope, all is clear.
21:09:39 <notmyname> cool
21:09:52 <notmyname> kota_: is all that clear to you too?
21:09:52 <kota_> thinking of 447129, i think it can be included in this cycle
21:10:20 <kota_> if it's ok to work with some minor fixes on the patch
21:10:23 <kota_> p 447129
21:10:24 <patchbot> https://review.openstack.org/#/c/447129/ - swift - Configure diskfile per storage policy
21:10:24 <notmyname> kota_: I think that would be great, but it's probably only going to get included if you review it and help out
21:10:26 <notmyname> ya
21:10:38 <notmyname> or maybe with zaitcev too
21:10:56 <zaitcev> Well, Kota added -1, I need to check why.
21:10:59 <kota_> i finished the review already and have a follow up, waiting rledisez... but he seems busy?
21:11:17 <notmyname> yeah, I haven't heard from rledisez. not sure what's going ok
21:11:31 <kota_> the -1 is soft one, regression from older patch set and i have an idea how it imporove it.
21:12:18 <notmyname> oh, I'd also meant to link https://review.openstack.org/#/c/507808/ because the authors have been very patient and it's largely an independent piece of work
21:12:19 <patchbot> patch 507808 - swift - Add ability to undelete an account.
21:12:41 <kota_> I'll try to poke rledisez to get it merged sooner.
21:12:56 <notmyname> thanks
21:13:09 <kota_> speaking of the others, probably i should go over s3api's :P
21:13:14 <notmyname> :-)
21:13:43 <timburke> fwiw, a lot of them were ports from swift3 patches
21:13:58 <notmyname> I think tdasilva is looking at the s3api patches, too
21:14:08 <timburke> but i'm sure they'd benefit from some review and critical thinking
21:15:16 <notmyname> also, FWIW, I'll set up an etherpad to start planning PTG discussion topics. I should have that ready later this week
21:16:05 <kota_> +1
21:16:15 <notmyname> #topic open discussion
21:16:26 <notmyname> ok, what else need to be brought up this week during this meeting?
21:16:36 <notmyname> tdasilva: I believe you wanted to bring something up
21:16:47 <tdasilva> yeah!!!
21:17:06 <tdasilva> hey everyone, just wanted to announce that I've joined SwiftStack, started here on Monday
21:17:21 <kota_> wow
21:17:51 <zaitcev> well
21:17:57 <zaitcev> that's only 1 for 2
21:18:03 <kota_> big surprise!!
21:18:11 <kota_> oh what?
21:18:15 <zaitcev> A good news but not enough to keep us ticking.
21:18:47 <notmyname> zaitcev: meaning?
21:19:21 <zaitcev> We need 2 more core reviewers with Sam and Alistair gone, right?
21:19:48 <notmyname> yeah, we're still net down 2 people from where we were
21:19:57 <timburke> we could always use more core reviewers
21:20:18 <notmyname> it's definitely a topic we should discuss in person at the PTG
21:20:43 <notmyname> tdasilva: being an employee of swiftstack myself, I'm super excited you've joined
21:21:08 <zaitcev> tdasilva: Will you relocate to SF?
21:21:28 <notmyname> overall, as a whole community, we've got a lot to do and think about when it comes to attracting new people, including core maintainers
21:21:32 <tdasilva> notmyname: thanks
21:21:36 <tdasilva> zaitcev: nope
21:24:29 <notmyname> kota_: zaitcev: one thing on that topic that I've been thinking about a lot lately is to recognize not on the contraction of the community but the concentration of dev work at swiftstack. and I would love to hear from you (email or privately on irc) your thoughts on how we can address issues and potential issues
21:25:23 <notmyname> we've got quite a few good things going for us as a community, but we've also got some challenges. i want to honestly look at where we are and make sure we aren't unintentionally excluding people
21:25:43 <notmyname> and I think we all have ideas on how to grow the community overall
21:26:04 <notmyname> and this too is something we should talk about at the PTG
21:27:04 <zaitcev> notmyname: I noticed some of it, but I thought that the fault was with us 3 at Red Hat for not driving the agenda more actively. Like, that PUT+POST spent much too long gestating. Or, the Python 3 was our thing for a while and it went nowhere until Tim in particular started attacking it seriously.
21:27:07 <notmyname> we are a team, regardless of who pays us or where we live, and together we need to figure out the best ways to go into the future
21:27:44 <notmyname> zaitcev: I definitely do not want to do any blaming. if there are any failings, we need to address them as a group (and solve them as a group)
21:28:13 <notmyname> but I think you're right about some patches like put/post taking too long to gestate
21:28:39 <zaitcev> Some people even ask me what's going to happen to Swift now that OpenStack itself ran out its hype cycle (according to the number of Google Searches for the term, e.g.). My stock answer is that people still need to store data, and Ceph is not doing everything needed.
21:29:04 <zaitcev> I suppose we could always divorce from OpenStack if it starts to collapse in a measurable way.
21:29:05 <notmyname> "people still need to store data" +100
21:29:17 <zaitcev> Well, there's S3, you know
21:29:28 <kota_> +1
21:30:05 <notmyname> I'm not ready to give up and say the whole world should "just use s3" :-)
21:30:47 <zaitcev> Yeah. Remember how they lost power in Virginia datacenter, which is the default region. Whoo, boy.
21:30:47 <timburke> (zaitcev: speaking of py3, you might be interested in some of where my patches are going -- i've got some horribly-disorganized commits up at https://github.com/tipabu/swift/tree/moar-py3 that helped me shake out some of the bugs in https://review.openstack.org/#/c/548727/)
21:30:48 <patchbot> patch 548727 - swift - py3: port bufferedhttp (hopefully)
21:30:52 <kota_> I believe Swift will be needed too.
21:31:17 <zaitcev> timburke: thanks
21:31:34 <notmyname> timburke: you just reminded me of something with that comment
21:32:01 <notmyname> it *is* important to keep in mind where we're going. eg what's the point of a certain patch and how it fits in
21:32:26 <notmyname> right now, there are a few high-level things that current and planned patches are leading to
21:32:30 <timburke> tl;dr: i've got some py3 func tests running! just gotta make sure that data written with py2 is readable on py3 and vice-versa, and that a mix of py2 and py3 servers can all communicate, and...
21:32:50 <zaitcev> nice nice
21:32:56 <zaitcev> upgrade path, too, I presume
21:33:00 <notmyname> first, with s3, after the current s3api patches land, we'll be able to work more freely on some new s3 compatibility. we'll likely be doign s3 versioning next
21:33:30 <notmyname> also, since the best we can do with s3 is to follow amazon, we'll always have more work to do there
21:34:03 <notmyname> second, OVH is working on their LOSF work, and it seems like an improvement that will significantly help a lot of clusters
21:34:55 <notmyname> one day, I hope we're able to have swift dynamically choose how data is stored on disk (LOSF slabs or not) and what durability scheme it uses (EC, replication) just based on the data
21:35:20 <notmyname> eventually, we may even be able to do stuff like change that over time (similar to policy migration work)
21:35:58 <notmyname> third, we've still got the "big picture" ideas about replacing rsync with our own code ("tsync") and rewriting the storage servers in golang
21:36:21 <notmyname> those three things above will take ... oh, idk ... another 3 years at least :-)
21:36:30 <notmyname> so we've got a lot of work to do still
21:37:06 <notmyname> but at a high level, those are the big things I see us working towards
21:37:42 <kota_> I know we have the big picture, always I answer the story when I asked from my co-workers (or boss) ;-)
21:37:58 <notmyname> kota_: nice :-)
21:38:17 <notmyname> anyway, that's all I've got for this week. anything else to bring up?
21:38:30 <tdasilva> kota_: are you going to the PTG?
21:38:43 <zaitcev> I think SwiftStack hoovers all the work like this because it's basically impossible to do anything unless one's company derives most of its revenue from Swift. I have to deal with stupidest thing like clustering all the things and suddenly you can't find your logs.
21:38:50 <kota_> tdasilva: sure, I booked whole week on the PTG
21:39:47 <kota_> tdasilva: how about you?
21:40:07 <tdasilva> yep
21:40:16 <kota_> :-)
21:42:07 <notmyname> anything else?
21:42:49 <notmyname> no meeting next week (unless someone else volunteers)
21:43:02 <notmyname> thanks for coming today, and thanks for working on swift
21:43:06 <notmyname> #endmeeting