19:00:48 #startmeeting swift 19:00:49 Meeting started Wed Jul 9 19:00:48 2014 UTC and is due to finish in 60 minutes. The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:50 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:52 The meeting name has been set to 'swift' 19:00:59 welcome everyone. who's here? 19:01:02 o/ 19:01:02 o/ 19:01:05 o/ 19:01:07 no, Who's on first 19:01:17 torgomatic: indeed he is 19:01:28 here 19:01:46 hey 19:01:46 mattoliverau: I'm impressed you've started getting up early for us :-) 19:02:08 ok, let's get started 19:02:09 yeah, and am mostly awake :P 19:02:12 #link https://wiki.openstack.org/wiki/Meetings/Swift 19:02:25 a few things to go over this week 19:02:38 portante is alive! ;-) 19:02:43 :) 19:02:53 poison ivy hasn't killed me yet 19:02:59 howdy 19:03:05 #topic old reviews and auto-abandons [torgomatic] 19:03:16 torgomatic: you were looking in to this. can you give a summary and update? 19:03:22 so we used to have a bot that would kill these things off for us 19:03:43 do we know what happened to that bot? 19:03:46 after N weeks with a negative score (either Jenkins or people), it'd get abandoned... N was either 1 or 2, I don't recall 19:03:58 after the Gerrit upgrade, that bot was deliberately killed by the infra folks 19:04:00 * portante remembers 2 19:04:13 torgomatic: ah, interesting. did they say why? 19:04:19 now, core team members can abandon/WIP any patch... before, it was only the patch author 19:04:25 oh 19:04:26 Cores have more powers now and can abandon things 19:04:27 so now that humans can do it, the bot is gone 19:04:38 oh 19:04:43 so humans are replacing robots in this brave new future? 19:04:49 that seems like technology going the wrong way :) 19:04:54 :) 19:04:58 * torgomatic liked the bot 19:04:58 * portante is now a bot 19:05:05 lol 19:05:07 * peluse_ would like to be abandoned 19:05:11 * portante or a cylon 19:05:47 dfg! 19:05:57 torgomatic: do you want to reinstate a bot for swift reviews? 19:05:58 ok, so what do we do then? add filters to our view of gerrit (I know how to do that)? 19:06:33 I think we should go back to automatic abandons of old things, whether that's done by a bot or by a person armed with copy+paste 19:06:36 accept that cores need to click more buttons to keep the gerrit views cleaner? 19:06:52 but it should be a simple, time-based rule that's applied fairly, like the bot used to 19:07:33 I'm actually heading over to Germany this weekend to goto the infra mid-cycle, did you want me to bring this up and complain? I can be pretty whiny when need to be :) 19:07:34 torgomatic: ie "I'm abandoning this patch because it hasn't had any activity in the two weeks since the last negative review"? 19:08:07 mattoliverau: sure 19:08:11 notmyname: exactly that sort of thing 19:08:30 mattoliverau: the concern is keeping more patches around that have been essentially abandoned by their owners 19:08:48 it does sond stupid! it would be better if projects could opt out of the bit 19:08:54 sound even 19:09:17 bot as well.. did I mention I'm 1/2 alseep :P 19:09:33 are there any concerns with, until mattoliverau get's to GER, letting cores abandon stuff? 19:09:45 none here 19:09:48 portante: dfg: clayg: peluse_: torgomatic: acoles? 19:09:48 nope... how about a 2-week (14-day) threshold? 19:09:56 I think 2 weeks is good 19:10:01 not problem by me, a 2 week would be good 19:10:07 sounds reasonable 19:10:07 great 19:10:22 Let me know if there is anything else I can fight for for swift while I'm there :) 19:10:23 * portante thinks it was 1 originally, now that he thought about it, but would rather 2 19:11:16 #agreed mattoliverau will ask at the -infra meetup about getting the bot again, and in the meantime cores will manually abandon patches 2 weeks after a negative review 19:11:30 ok, moving on 19:11:36 it used to be that WIP would not be abandoned, I think, even if Jenkins had -1 19:11:36 #topic python-swiftclient release 19:11:55 acoles: ya, IMO WIP is just "there" 19:12:09 yeah. move on :) 19:12:18 I'd like to make another release of python-swiftclient after we get the storage policies support added 19:12:29 open patches are https://review.openstack.org/#/q/status:open+project:openstack/python-swiftclient,n,z 19:12:48 I think https://review.openstack.org/#/c/73920/ is the only patch that needs to land 19:13:00 are there others that should land before a swiftclient release? 19:13:34 acoles: I really want the keystone v3 support before the juno release, but that can come after a release supporting SP (IMO) 19:13:58 wait, why hasn't https://review.openstack.org/#/c/102510/ landed? 19:14:02 notmyname: was about to say...would make me hapy to see that land soon 19:14:19 acoles: ya, me too 19:14:23 dunno... it seems to have the requisite number of checkmarks 19:14:32 there 19:14:52 thanks 19:14:53 notmyname: acoles: +1 for the keystone v3 patch 19:15:05 just checking if there were other concerns on it. I didn't see anything 19:15:25 ok, anything else needed for swiftclient before a release? 19:15:32 beyond being able to set storage policies? 19:15:50 cschwede: thanks for the review btw, if you get chance could you look at the latest version (small changes) 19:16:44 FWIW, I like the tempurl support one (the concept) and torgomatic sold me on the concept of the SwiftService one (but that one is _huge_) 19:17:04 acoles: np, you’re welcome - just noticed the small changes, was wondering where my +2 has gone :) 19:17:20 yeah, the SwiftService one is nice, but the API is still a little under discussion... I'd like that one to land, but not for the next release 19:17:29 that way there'll be a little time to tweak the API before it goes out to the world 19:17:30 but I'll take silence for other patches before a release to mean that jsut getting the one to set a storage policy is all that's needed ;-) 19:17:36 torgomatic: that swiftclient help chagne did add a weird regression with 'swift info --help' vs. 'swift capabilities --help' 19:17:37 torgomatic: ack 19:18:21 clayg: yeah, i wasn't sure about that 19:18:23 hmm mark isn't here. maybe we can bug him to fix that, clayg 19:18:39 clayg: you want me to not approve that one, then? 19:18:58 torgomatic: well, clayg did give a +2 with that comment ;-) 19:19:16 acoles: can you ask mark to submit another patch to fix it? 19:19:29 (unless clayg wants to stop the merge train right now) 19:19:30 notmyname: yeah, or i will push over 19:19:34 acoles: thansk 19:19:35 no no no, it's fine 19:19:37 :-) 19:19:44 but that's one more patch that should go in before the release or whatever 19:19:44 ok, I'll do nothing. easier done than said! 19:20:12 actually what would folks think of swift help rather than swift --help to avoid regression 19:20:16 acctually I did want to ask why we can't use argparse - either as py27 or a depends if we still care about py26 on that repo (i'm guessing yes?) 19:20:22 (Small regresssion mind you) 19:20:30 acoles: very small 19:20:31 ok, I think that wraps it up for swiftclient patches for the next release. everyone go review https://review.openstack.org/#/c/73920/ 19:20:49 clayg: yep, we still care about py26 19:20:49 notmyname: well and the other patch we're discussing to fix the bug torgomatic just merged :P 19:21:06 clayg: "the" bug? that seems overly optimistic! ;) 19:21:07 right :-) 19:21:07 torgomatic: so do we care about an argparse depends for py26? 19:21:13 clayg: I don't 19:22:34 I'm ok with moving to using argparse 19:22:43 ok, well that would be betterish 19:22:54 thus adding the dependency to py26 19:23:24 #topic priority reviews 19:23:33 #link https://wiki.openstack.org/wiki/Swift/PriorityReviews 19:23:34 i think it would just be a dependency, like even if your're running py27 you still install it or something... idk 19:23:48 anyway, that thing would be better with argparse - so maybe we do that at somepoint 19:24:01 I wanted to go over the sections of the priority review page 19:24:05 acoles: if you have an idea for the help fix please catch me offline or w/e 19:24:09 so from the top.... 19:24:20 how's the review dashboard working for you? 19:24:35 torgomatic: I think you had some concern that "not everything" is shown? 19:24:45 clayg: ack 19:24:58 torgomatic: have you looked in to what's not shown and why it's not? 19:25:22 notmyname: nope, haven't figured out why... I just star things that I care about and find them that way 19:25:28 ok 19:25:50 should we add a starred section? 19:26:31 (either you're perfectly happy or left speechless at the horribleness of it) 19:26:32 * torgomatic is indifferent 19:27:31 I like it, but it would be nice if it showed changes you haven't reviewed yet and changes what have new patchsets (so again haven't reviewed) 19:27:54 mattoliverau: I think I saw some filters for that somewhere... 19:28:10 looking at what's open, I'd like to see some feedback on the specs. we've got 2 submitted, but no comments. and peluse_ will be submitting one soon for the EC work 19:28:31 I've have some filters that kind of do it.. I should just polish em up a bit... hopefully 19:28:43 ie it's good to give feedback to people planning some bigger stuff, and if we can't do it for active devs, we wont' do it well for new contributors 19:30:13 ok, looking at the other stuff on the priority review page... 19:30:55 the case changing behavior in eventlet. important for the swift3 middleware and makes use of a feature added to a new version of eventlet https://review.openstack.org/#/c/93780/ 19:31:27 needs another +2 to land, but are there any major concerns about this patch? 19:31:43 please say yes, no, or don't know 19:31:48 no 19:32:06 no 19:32:08 I just want to quickly go throught the patches listed to see if there is anything major outstanding on them 19:32:23 torgomatic: mattoliverau: well you two have actually reviewed it and said lgtm ;-) 19:33:43 ... 19:34:09 I assume others are in the process of reviewing :) 19:34:42 perhaps. I'm really just curious what people think about it high level now. saying "I don't know" is ok 19:35:46 dont know :-) 19:35:57 elambert: :-) thanks 19:36:08 notmyname: no major concern from my side, lgtm 19:36:26 just ttok a quick look and seems reasonable 19:36:30 took 19:36:38 ok, thanks 19:37:06 we've already looked at the next 2 sections. now for the performance section of the priority reviews page 19:37:28 https://review.openstack.org/#/c/75941/ needs to be rebased (adds mutlithreds to object updater) 19:37:50 anyone know Alex's IRC nick? 19:37:59 Alex_Gaynor? 19:38:04 that's the one 19:38:09 not Alex Pecoraro 19:38:17 s/not/no/ 19:38:28 author of that patch 19:38:38 apecdown? 19:38:56 Not sure what exactly the test tests in 93780 19:39:08 Huh I'm late to the party as usual. 19:39:31 zaitcev: thanks for looking :-) 19:39:50 I'l try to get in touch with Alex to rebase. or I'll rebase it and push it up 19:40:04 torgomatic: anything on the zero copy patch? 19:40:11 or any questions on the zero copy patch? 19:40:51 the GET patch is ready; the PUT patch still needs some benchmarks, though now that I've beaten $internal_cluster into submission I can see about getting those 19:41:00 cool 19:41:11 or if anyone else wants to try it out on their hardware, that would be great :) 19:41:42 torgomatic: both are in my best laid plans ... :( 19:41:53 ok 19:42:07 clayg: anything needed to be brought up about the reconciler patches? https://review.openstack.org/#/q/status:open+project:openstack/swift+branch:master+topic:multi-reconciler,n,z 19:43:22 clayg: other than "go review them" ? 19:44:20 no it's fine 19:44:25 ok 19:44:36 mattoliverau: had a couple of comments on one of the patches that i'm still milling over 19:44:41 ok 19:44:46 that's all I wanted to cover about what's on the priority reviews page 19:45:07 also, just to experiment with this meeting to see if a walkthrough of patches would be usefule 19:45:20 might be nice to have a quick glance at them on 103779 19:45:21 clayg I just noticed you had responded to some of my questions (this morning), thanks.. will read them when my brain it a little more awake :) 19:45:35 mattoliverau: no worries, thatnks for looking at it 19:45:49 notmyname: seemd ok to me! 19:46:00 #topic open discussion 19:46:11 anything else to bring up in the meeting 19:46:17 I would like to ask cores to review those: https://review.openstack.org/#/c/90016/, https://review.openstack.org/#/c/90066/. It resolves two bugs and stuck with reviews for a long long time. 19:46:50 gvernik: looking 19:47:17 thanks 19:47:20 gvernik: both of those titles sound reasonable 19:47:30 gvernik: can you check that they still rebase properly? 19:47:35 gvernik: but they both have -1's on them currently? 19:48:17 yes, i got some remarks, responded to them immedeately and it's not reviewed anymore. I can rebase them again 19:48:35 gvernik: on the first one, you've got a -1 from cschwede. cschwede can you respond by either removing the -1 or stating that you still have a -1 there? 19:48:36 clayg: gvernik: i think these -1 need some discussions, would be good to have some more eyes on them 19:48:47 mattoliverau: same with you for the second 19:49:14 Sure I'll make sure I revisit it today :) 19:49:14 cschwede: werd, i'm looking through the comments now 19:49:23 mattoliverau: thanks 19:49:23 notmyname: yep, will review both of them tomorrow morning 19:49:27 cschwede: thanks 19:49:53 that way it's clear (to the author and others) that it's still an issue and not just a "reviewer abandonded" thing 19:49:54 i'll try to take alook tomorrow 19:50:00 thanks acoles 19:50:45 gvernik: as you've experienced (unfortunately), we struggle with giving timely feedback. I hope we're getting better, but thanks for bringing those up 19:50:51 maybe a starred section of the dash board is a good idea :) 19:50:58 I'm hoping that the dashboard capability can help 19:51:27 gvernik: also 90066 doesn't have any tests 19:51:34 and I've still got the "core sponsors" concept on my todo list that we talked about in atlanta 19:52:16 cool 19:52:32 wait, max_container_per_account is a config option? 19:52:32 clayg: i will rebase it and add tests. no problem, thought it doesn't need tests 19:52:33 anything else to bring up? 19:52:37 notmyname: cool if you can get like mercedes to sponsor us all :) 19:52:48 portante: is there such a thing as a change that doesn't need tests? 19:52:51 acoles: wait what? 19:53:00 no 19:53:03 :) 19:53:17 I wanted to say, If anyone thinks of anything they want me to bring up / fight for at the infra midcycle, let me know :) 19:53:30 mattoliverau: thanks. that's next week? 19:53:34 notmyname: 'core sponsors' - sorry, clearly lame joke on my part 19:53:46 acoles: ah. I like that a _lot_ better! 19:53:51 and.... we digress 19:53:52 acoles: i like the idea ;) 19:53:53 yup, so have a long flight ahead of me, and more jetlag!! 19:54:15 mattoliverau: in january there was talk about limiting the duplicate test runs in jenkins. it would be nice to ask about that 19:54:46 notmyname: ok noted :) 19:54:49 thanks 19:54:55 I think we're good then 19:55:02 thanks everyone for attending 19:55:06 #endmeeting