21:05:51 <notmyname> #startmeeting swift
21:05:52 <openstack> Meeting started Wed May  9 21:05:51 2018 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:05:53 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:05:56 <openstack> The meeting name has been set to 'swift'
21:05:56 <notmyname> whoa!!
21:05:59 <clayg> !!!???
21:06:00 <openstack> clayg: Error: "!!???" is not a valid command.
21:06:05 <clayg> sorry
21:06:08 <acoles> we can meet here, it'll be like dublin airbnb all over again. swifters will prevail.
21:06:12 <notmyname> ok, meeting time is now and in here
21:06:39 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:06:51 <notmyname> the agenda this week is simply to talk about feature/deep-review
21:06:55 <corvus> notmyname: i can remove that
21:06:57 <notmyname> #link https://review.openstack.org/#/q/status:open+project:openstack/swift+branch:feature/deep-review
21:07:16 <notmyname> corvus: thank you
21:07:56 <notmyname> acoles: I've seen lots of new patch sets for the review banch this week. how is it going?
21:08:44 <acoles> notmyname: it's ok. thanks to those who have reviewed.
21:08:58 <notmyname> there have not been a lot of reviews
21:09:05 <notmyname> (granted, it's a huge topic)
21:09:08 <acoles> so far this week I have turned around review comments and pushed new patchsets daily.
21:09:20 <notmyname> thank you for that
21:09:26 <acoles> notmyname: but, yes, we're not making much headway up the chain of patches yet
21:10:12 <notmyname> given who we have available for reviews, I wrote a wall of text here in IRC earlier this week :-)
21:10:17 <acoles> thanks mattoliverau for helping with the docs
21:10:46 <fungi> notmyname: reviewing the akick entries, i don't see any ban for your ip address. looks like one was added to #openstack-meeting for the "patchbot" nick specifically but not by address mask. also i see your notmyname nick present in #openstack-meeting
21:11:51 <notmyname> acoles: where should we be focusing on for now?
21:12:07 <notmyname> should we start with the first refactoring patch?
21:12:49 <mattoliverau> I got sidetracked with the diagrams/doc.  But will do a good review over the next 2 days.  Tho have looked at the code alot over the last while,  and probably a little code blind,  except for acoles better code that is :p
21:12:51 <acoles> notmyname: that's had quite a lot of review from timburke and kota, so no, skip that if you're starting
21:12:53 <fungi> aha, not in the akick list, it's a +b on the channel for your address mask
21:12:57 <fungi> i'll fix that
21:13:27 <fungi> nevermind, someone else already has
21:13:44 <notmyname> also, reminder to everyone: when leaving a comment, please include a diff with improvements
21:13:47 <notmyname> fungi: thanks
21:13:59 <acoles> I liked notmyname's suggestion that some start at top and some start at bottom of chain, but not sure how to coordinate that
21:14:49 <notmyname> I started with the docs and got as far as the "beyond here it needs to be updated" marker
21:14:51 <acoles> if you want to dive in, the final patch (docs) now has sufficient content to guide you through dpeloying and sharding
21:15:07 <notmyname> yeah, the docs look great
21:15:43 <notmyname> acoles: you mentioned something earlier this week about dropping some stuff from the docs. has that happened already, or is that something someone else can do?
21:16:41 <acoles> notmyname: no I was hoping to work on that tomorrow - everything should stay up to the obvious 'below here is outdated' heading , but beyond that I think we can trim down the content.
21:16:55 <notmyname> ok
21:17:12 <acoles> and I'm happy if anyone else wants to do that.
21:17:17 <notmyname> :-)
21:17:51 <notmyname> I would love to be able to do that during my tomorrow
21:18:17 <acoles> it's great to have plenty of explanation but not if it slows us down by needing to review lots of text - we can always add to it on master
21:18:37 <notmyname> right
21:19:02 <notmyname> clayg: can you give a quick update of what we've seen with this feature branch on the really big DB in our test clusteR?
21:19:08 <acoles> plus, I have relatively large amount of comment in code
21:19:19 <clayg> notmyname: we're getting there!
21:19:42 <clayg> I'm playing with some of the knobs trying to figure out the right tuning
21:19:57 <notmyname> a week or two ago, I said something about it being pretty slow, but it looks like it's gotten a lot better since then
21:19:59 <mattoliverau> An explaination on the shard range states woukd be good. But yeah trimmming down the other under hood makes sense
21:20:04 <clayg> we're learning with a 140M row db on a cluster with only HDD - totally under water with out of sync db's and async pileups
21:20:10 <clayg> so this thing was a MESS - and it's WORKING!!!
21:20:21 <acoles> mattoliverau: yes, there is some new content I want to add too
21:20:29 <clayg> we've dug outself out of the whole and watching things just sort of settle down and clean up
21:20:34 <acoles> mattoliverau: states being one
21:20:46 <notmyname> clayg: that's really good news
21:20:48 <clayg> async's have made their way into the new shard db's and everything looks good
21:21:11 <clayg> still waiting for the final unlinks to happen on a couple of replicas - then we're moving on to more testing on some new/different datasets and different hardware configs
21:21:18 <clayg> this stuff is AMAZING
21:21:22 <clayg> total game changer for swift
21:21:49 <clayg> mattoliverau & acoles & timburke definately get some mad props here
21:21:51 <notmyname> kota_: m_kazuhiro: have either of you had a chance to run this feature branch yet?
21:21:59 <notmyname> clayg: yes, definitely!
21:22:35 <kota_> notmyname: not yet playing. just run unit tests in my local at the refactoring patch.
21:22:46 <notmyname> kota_: that's good
21:22:56 <kota_> i'll make it maybe in this week or the next.
21:23:05 <notmyname> thanks
21:23:16 <acoles> It would be great to start getting some review on https://review.openstack.org/565744
21:23:17 <patchbot> patch 565744 - swift (feature/deep-review) - Add support for sharding in ContainerBroker
21:23:21 <mattoliverau> clayg \o/
21:23:28 <kota_> at least in the review process, I should do something to check the behaivor and performance
21:23:36 <notmyname> I'm still hoping that we can have the patch chain land at the end of next week
21:23:52 <notmyname> (remember that may 17 is swift's birthday)
21:24:08 <clayg> notmyname: I say if it's NOT landed we just go ahead and do it anyway - I trust Al
21:24:12 <clayg> this stuff is too awesome not to have
21:24:14 <notmyname> :-)
21:24:21 <acoles> clayg: do not trust anyone
21:24:28 <clayg> I mean you can glance at it or whatever - it works exactly how mattoliverau told you it did 3 years ago ;)
21:24:34 <clayg> well.. *mostly* :P
21:24:45 <m_kazuhiro> notmyname: not yet. but I will do.
21:25:13 <mattoliverau> Lol
21:25:23 <clayg> mattoliverau: 4?  :P
21:25:48 <mattoliverau> With alot of polish from acoles and timburke
21:25:48 <notmyname> I'm not super worried yet about not having a lot of visible reviews on the patch chain, if everything is actually mostly ok. but I'd like to see some by friday or monday
21:26:03 <acoles> if any core wants to discuss patches with me in real time (voice/video) then I'll do my best to make that happen
21:26:33 <notmyname> seeing some reviews (positive and negative) will help us all get it landed sooner
21:27:14 <mattoliverau> clayg too early to math :p
21:27:27 <notmyname> I know for a fact we can still make improvements to container sharding. there's still a lot of opportunity to do more (eg automatic sharding is something we're leaving to do later)
21:28:20 <notmyname> so this comes down to (1) does it make users' lives better? (2) is it maintainable? (3) is it written so we can keep making it better?
21:29:56 <notmyname> is there anything else we need to talk about on this topic?
21:30:07 <notmyname> acoles: clayg: kota_: mattoliverau: you good?
21:30:22 <mattoliverau> Yup
21:30:30 <kota_> yes
21:30:35 <acoles> notmyname: yes
21:30:51 <notmyname> ok, let's look forward just a little bit...
21:31:13 <clayg> i gotta bounce, I'll catch up on logs!
21:31:35 <notmyname> I'd like to tag a swift release after this feature branch lands
21:31:51 <mattoliverau> clayg: o/
21:32:07 <notmyname> so if it lands next week, that means a new swift release will be the end of next week or, more likely, the beginning of the week of the 21st
21:32:15 <notmyname> that week is the week of the openstack summit in vancouver
21:32:41 <kota_> ¥o/
21:33:01 <notmyname> there's a lot of great stuff that's in swift that we'll be able to talk about there :-)
21:33:07 <notmyname> s3api and container sharding
21:33:23 <mattoliverau> That would be  huge release
21:33:27 <notmyname> yep
21:33:48 <notmyname> there's also an onboarding sessions that a few people have signed up for. I'll be leading that (and I hope with a lot of help from kota_ and mattoliverau :-)
21:34:08 <mattoliverau> Yup I'll be there :)
21:34:38 <kota_> oh, i didn't notice the onboarding session. I'll add it to my schedule
21:34:39 <notmyname> speaking of openstack events, please be sure to sign up for the PTG happening in september in Denver
21:34:52 <notmyname> kota_: it's immediately after the swift project update on thursday
21:35:24 <kota_> (that means i didn't check the schedule in detail yet :P)
21:35:27 <notmyname> the PTG is $200 right now and the prices will go up at the end of next week
21:35:36 <notmyname> #link https://www.openstack.org/ptg
21:36:10 <notmyname> that's all I had for this week. feature/deep-review update. next swift release. sing up for the ptg
21:36:17 <notmyname> does anyone else have anything for the meeting this week?
21:36:18 <mattoliverau> Also to be a little annoying, ill be in Fiji next week for a family holiday,  so will review as much as i can before i leave,  and try among keep up with what ever internet i end up getting.
21:36:23 <notmyname> hmmm
21:36:39 <notmyname> tdasilva should be back next week, too
21:37:09 <kota_> ah, RedHad summit in Boston this week.
21:37:24 <notmyname> nah, he's on vacation this week
21:37:34 <notmyname> but there is a red hat summit. but in san francisco
21:37:36 <kota_> oh, really
21:37:42 <mattoliverau> Damn phone auto complete, * try to keep up..
21:37:50 <kota_> ok, not in Boston
21:38:19 <kota_> i saw takashi's activities. he joined it.
21:38:47 <notmyname> thank you for your work on swift.
21:38:50 <notmyname> #endmeeting