21:01:24 #startmeeting swift 21:01:25 Meeting started Wed Apr 11 21:01:24 2018 UTC and is due to finish in 60 minutes. The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:01:26 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:01:29 The meeting name has been set to 'swift' 21:01:33 lol, thanks 21:01:34 o/ 21:01:48 #link https://wiki.openstack.org/wiki/Meetings/Swift 21:01:58 hi o/ 21:02:02 oh dear, wait... now *i'm* the chair... i don't know that i'm ready for the responsibility! 21:02:03 agenda this week is just a bit of catch-up of ongoing work 21:02:49 timburke: I think it just means you're the only one who can endmeeting. maybe topics, too. but I don't think we need those this week 21:03:10 first up, https://bugs.launchpad.net/swift/+bug/1755554 is still open and has an unreviewed patch at patch 555245 21:03:11 Launchpad bug 1755554 in OpenStack Object Storage (swift) "Percent signs in object names cause trouble for versioned_writes" [High,In progress] - Assigned to Kota Tsuyuzaki (tsuyuzaki-kota) 21:03:11 https://review.openstack.org/#/c/555245/ - swift - Fix versioned writes error with url-encoded object... 21:03:40 there's not a particular question associated with that right now. just something that needs to be landed by the time we do a release 21:03:47 if you have some time to do a review, that's a good one to do 21:04:07 kota_: how's feature/s3api looking? seems to me that all the patches have landed! 21:04:17 notmyname: true 21:04:25 that's great news 21:04:28 i just push +A on the last patch yesterday 21:04:45 so, from your perspective, it's ready to be merged to master? 21:05:10 yes. if i don't forget anything needed. 21:05:15 :P 21:05:15 heh, ok :-) 21:05:27 here's what I want to do about that... 21:05:41 for now, kota_ please do not make the merge commit proposal. I will do that 21:05:55 I talked to tdasilva about it earlier today (he's not available during the meeting today) 21:06:03 and I talked with you, kota_, about it some last week 21:06:35 I want to talk to several other people first, about how we can land this and when we should, given the oncoming feature/deep merge plans 21:07:20 kota_: I'll do all this before the end of this week, and I'll let you know what everyone thinks 21:07:21 ok 21:07:38 ok 21:07:45 then we'll make the merge commit (or not) and have the appropriate people lined up to review 21:08:41 so for everyone who's not kota_, I'll likely be asking you about the feature/s3api merge sometime in the next couple of days :-) 21:09:13 next up... torgomatic has been doing some good work improving the performance of the reconstructor and replicators 21:09:27 his work is a patch tree rooted at patch 555563 21:09:27 https://review.openstack.org/#/c/555563/ - swift - Multiprocess object replicator 21:09:50 after talking with acoles and timburke this week, turns out that this work could really benefit the feature/deep work 21:10:05 if not for pure performance reasons, at least to reduct merge conflicts 21:10:16 and I believe clayg has reviewed some 21:10:35 please help torgomatic with some reviews there. and please start at the beginning of the patch chain instead of somewhere in the middle :-) 21:10:41 So have i 21:10:41 torgomatic: anything to add there? 21:10:59 just that there's a follow-up commit to clean up a few things, so also look at ... 21:11:11 https://review.openstack.org/#/c/559872 21:11:11 patch 559872 - swift - Cleanups for multiprocess object replicator 21:11:29 mattoliverau: thanks! I didn't mean to forget you :-) 21:11:56 Nah that's ok. There some good improvements. 21:11:57 I'm fine with combining the patches or merging both. There's nothing badly broken in the first one, just some minor issues. 21:12:21 ok 21:12:25 thanks 21:12:42 acoles: timburke: what's the good word with feature/deep this week? 21:13:31 it takes a while to work with large containers :-/ 21:13:40 :( 21:13:52 :/ 21:13:55 I think acoles is stubling over the word "good" in my question :-) 21:14:03 heh 21:14:10 i had a 145M row db i was identifying shard ranges in. with no swift processes running, it took ~12 minutes 21:14:21 notmyname: lots of progress, lots of patches 21:14:23 Faster then the last poc :p 21:14:31 with container server and all background processes running, it was ~90 minutes 21:15:45 ...and then swift-manage-shard-ranges failed to insert them (or even create a shard ranges table) because the db was locked. so i need to get something going for that 21:16:09 ok 21:16:49 notmyname: the stated plan is to propose to master soon - I assume we'll follow previous pattern of using a final review branch? 21:16:55 originally, we'd planned to propose a merge commit next week 21:16:59 If we can skip the offset some more it might speed things up... But that isn't easy. Or at least obvious for me. 21:17:02 acoles: yes, same pattern 21:18:49 I'll handle the branch as needed 21:19:16 I don't want to press you too much on a "is it ready" question, but how are you feeling about it? 21:20:33 your not-so-rapid response worries me :-) 21:20:35 it's not ready yet. I've been wittling down the high priority items on backlog, there's a few still there, plus whatever we learn from timburke's expereinces 21:20:52 ok 21:21:43 my perspective is that while we have dates imposed on us at times by our employers (for this or other features), we should not release something that isn't ready. if it's not ready, that's ok, and we'll make sure it's ready before we release it 21:21:47 I'm trying to focus on stuff that is best to be settled and stable before proposing 21:22:09 it's better to know that sooner than later, and I can handle any fallout that happens because we missed a date we ourselves set for ourselves 21:22:26 vs 'minor' stuff that we can add in during review 21:23:11 acoles: and this week I'll try to balance the two sides of "is it done yet?" and "leave me alone john, or it will never get done" :-) 21:23:39 notmyname: the goal was that April 16th would be the *earliest* and I'm confident we'll satisfy that goal :) 21:23:48 great! good news, everyone! 21:23:50 :-) 21:24:17 rledisez_: I saw a golang patch from alex today 21:24:26 rledisez_: anything to share on LOSF? 21:25:04 this patch, it's not useful on its own, but the python code should follow in one or two weeks. alex want to do some clean up before pushing python 21:25:16 #link https://review.openstack.org/560480 21:25:17 patch 560480 - swift - Initial LOSF RPC commit 21:25:39 ok 21:25:41 we are converting right now a complete cluster. should be done in about 20 days 21:25:47 wow 21:25:49 so we will have real prod result on 3 weeks :) 21:25:49 yeah, I took a look at the LOSF one, and it'll be a lot more interesting with some Python code 21:25:49 cool 21:26:04 right now it's like "here's a bunch of RPC methods you could call if you want to" 21:26:19 torgomatic: exactly. not much fun ;) 21:26:40 :) 21:27:15 I know how much every one here loves meetings, so I'm sorry to say I've got to stop in 3 minutes. but is there anything else to bring up from anyone? 21:27:42 ah ha! but it's timburke who's the chair and has to do the endmeeting thing! so he can go All. Day. Long. 21:27:59 mwahahaha 21:29:08 The chair times out after some time then anyone can end it.. just can't remember the time limit... 30mins? An hour? 21:29:20 nice failsafe 21:29:43 Learnt that from all the spam start meetings a few months ago 21:29:59 sounds like that's about it. thank you all for your work on swift 21:30:11 #endmeeting 21:30:17 nope. not me. not yet 21:30:33 No not 30mins then :) 21:30:52 timburke: would you do the honours? 21:31:02 #endmeeting