21:00:25 #startmeeting swift 21:00:26 Meeting started Wed Oct 12 21:00:25 2016 UTC and is due to finish in 60 minutes. The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:27 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:29 The meeting name has been set to 'swift' 21:00:40 who's here for the swift team meeting? 21:00:46 o/ 21:00:46 o/ 21:00:47 o/ 21:00:49 o/ 21:00:53 o/ 21:00:54 hi 21:00:57 o/ 21:01:20 o/ 21:01:34 o/ 21:01:39 o/ 21:01:39 oh hi mrda 21:01:46 hey notmyname! 21:01:53 mrda \o/ 21:02:01 mattoliverau: :) 21:02:06 hi 21:02:18 hello 21:02:18 mrda: thanks for joining us today 21:02:44 notmyname: Just having a look... 21:02:45 welcome everyone 21:02:58 agenda for this week is... 21:03:00 #link https://wiki.openstack.org/wiki/Meetings/Swift 21:03:26 #topic summit topics and prep 21:03:35 the summit starts in just a week and a half 21:03:58 so, if you're like me, it's time to schedule all your tourist things :-) 21:04:03 here 21:04:16 and good job if you've already done that 21:04:35 please continue to fill out the topics etherpad 21:04:36 #link https://etherpad.openstack.org/p/ocata_swift_summit_topics 21:04:40 I have my ticket and accom does that count :P 21:05:07 I want to start doing scheduling at the end of this week. there's still plenty of room for other topics, so please add whatever's on your mind 21:05:16 mattoliverau: good job 21:05:19 a week and a half - that's nuts 21:05:23 cschwede: hi! 21:05:41 so much to do, so little time 21:05:48 and if you see something that's already on the list, but not very detailed, feel free to add some notes abotu specific stuff you'd like to go over 21:06:03 clayg: \o/ 21:06:18 eg (taking the first one on the etherpad) if there's more you want to know/study or are concerned about with container sharding, add that to the etherpad 21:07:32 * mattoliverau does have some more info to write up (re:sharding), which I hope to do well before the start of summit... but don't have a great track record of it :P 21:07:35 are there any questions about the summit or topics? 21:07:52 anythign you want to go over that hasn't been brought up yet? 21:08:15 I assume next meeting is too close to go though the group prioritising, so end of this week you say? 21:08:37 correct. I'll take what's on the etherpad and build a schedule from that 21:08:37 nadeem: i have this weird feeling we won't see you dfg or redbo in spain? 21:08:47 clayg: only mattoliverau from rax, AFAIK 21:08:51 pdardeau: ? 21:08:51 that is true 21:08:56 :( 21:09:11 clayg: i won't be there 21:09:30 $%^&*( 21:09:52 budget is the issue... 21:09:53 clayg: maybe they can stay home and do a lot of reviews instead? ;-) 21:10:08 bright side? :) 21:10:45 we should try and figure out some conference call thingy for the humming bird guys, cause it's going to be an important topic. 21:10:48 oh, if you've got a link to some info about a summit topic that's on the etherpad, please add it to the etherpad 21:11:37 mattoliverau: I'll let you be the one to suggest to redbo to be awake and on the phone at 4am his time ;-) 21:11:37 if you're on that pad can you add a "[not ]going" next to your name 21:12:05 oh, yeah, going or not going is good info 21:12:09 I live in another continent so I should be safe.. you however :P 21:12:18 at mattoliverau points out - it's important for planning 21:12:34 no reason to wait till we get there and start being like "anyone see jrichli?" 21:13:06 jrichli: your coming right? 21:13:11 sorry I'm late peeps, I must have picked up clayg time in CA 21:13:13 :-) yes. 21:13:18 \o/ 21:13:28 lol 21:13:40 joeljwright: are you going to barca? 21:13:51 yes, I'll be there all week 21:14:00 yay (add your name and that fact to the etherpad) 21:14:04 b7a? 21:14:15 *ugh* 21:14:23 on it 21:14:27 gah, thanks obarca 21:15:05 ok, anythign else about the summit? jrichli acoles kota_ hosanai cschwede anyone? 21:15:34 i'm trying to decide if I am coming or going ;) 21:16:03 notmyname: i want my sleep back! running out of time for preparing everything… ;) 21:16:05 please keep updating the etherpad. if you're working on something or thinking about something, write it down 21:16:13 cschwede: you can sleep after the summit ;-) 21:16:57 also note that there's a section at the bottom for schedule conflicts, if you're giving a talk or something 21:17:05 i'd be interested in talking about that touristy stuff later 21:17:29 +1 21:17:47 ok, moving on to the next topic 21:18:18 #topic faster/more/different reviews or review process. or something. not sure what to call this 21:18:35 we've talked a lot in the past about how to handle the review load 21:19:03 and how to make progress on long-lived patches. well, all patches, but especially those that are embarrassingly old 21:19:29 oh yay. netsplit 21:19:36 nice timing 21:19:39 :( 21:19:49 i'm still on this side 21:19:51 o/ 21:19:55 me too 21:19:55 me too 21:19:58 o/ 21:20:02 doesn't look too bad for people active in this meeting 21:20:04 so lets make all the decisions 21:20:06 ok, carrying on... 21:20:41 we've tried a few things (stars, wiki pages, dashboards). all of which have had varying degrees of success, and IMO none of which are bad 21:20:58 but the reality is that we still have a lot of old reviews and slow review times 21:20:59 it looks like frequently disconnected 21:21:23 and I've realized that the reason isn't because people aren't doing stuff or interested in it 21:21:29 clayg: raise a hand, ok. o/ 21:21:34 it's because there's *so much* going on at once 21:21:38 kota_: hi. I see you 21:21:55 hi notmyname 21:22:12 it's the fact that we've got ~30-40 people actively working on 40-50 things 21:22:34 so it's hard to mentally shift from global EC to container sharding to part power increase to EC bugs fixes to golang to whatever 21:23:10 if you're really good, you'll be able to do like 2 reviews a day, when accounting for all the "load this entirely new thing into your head" time 21:23:11 oh gosh don't list them out like that it's just exahusting reading it 21:23:14 lol 21:24:01 I've had crazy ideas before (give everyone commit access!) and not so crazy ideas (not really) 21:24:18 I was just going to say the same thing.. a review can take a while (at least for me and I could be slow) because of the context switching in my brain. 21:24:39 Tho I'd admit I could try and do more and probably should 21:24:47 lol, no 21:24:59 context switching hurts 21:25:12 the fact that you're able to do all the context switching and still make progress is great. don't feel bad about that at all 21:25:24 after talking to a few other people about this, here's something that I think we should try 21:25:58 our current social agreement is that we have 2 cores add a +2 before somethign lands. and it should be people that aren't involved in writing the patch 21:26:32 this is silly. the more people that are involved means that the less people who can actually land stuff. that's crazy when you think about stuff like crypto, ec, etc. 21:26:48 all that big stuff is *better* because more people are involved, not "needs more checking" 21:27:06 agreed 21:27:37 I was reminded (by clayg and timburke) that the point of the 2 +2 rule is so that there is more info sharing about what's going on 21:28:03 if we have 1 +2 or 2 or 3 or whatever, we will still land bugs. more reviews does not tend towards zero bugs in the code 21:28:18 the point of reviews is to find obvious things, have different perspectives, and share info abotu what's going on 21:29:02 so with those things as the goal (and not being legalistically bound to our own rules which we created) here's the change I propose 21:29:16 (baby steps first before we try something more different) 21:29:42 if you're collaborating with someone and you push up a change (thus making you a co-author), it's ok to +2 or +A that patch 21:30:07 that will help a lot 21:30:15 that does get us stuck alot 21:31:03 is it just the 3 of us left?! 21:31:05 it's still good to get a 2nd +2 21:31:05 hmm, what happened to the rule „patch and +2/+A should not come from the same company“? does that still apply? 21:31:24 cschwede: I trust you to land code when I'm sleeping. do what you think is appropriate 21:31:27 joeljwright: nope 21:31:49 cschwede: I think we've always been sorta lax about that 21:31:57 my irc client keeps going crazy with logout/join 21:32:17 joeljwright: mine too 21:32:19 yeah, it will be interesting to see what the meetbot logs are for this 21:32:32 cschwede: the deversity thing - i'm not even sure that was a rule as much as general ackowledgement that just as much as an individual can get blinders on it can happen to small team that works together closely as well 21:32:38 I already +2 when I have co-authored *a little*, but not when I have added a lot. e.g. I would not +2 this https://review.openstack.org/210099, so I think it is a matter of judgement 21:32:54 I'm more likely to +2 if I only added to the tests for example 21:33:17 notmyname: clayg: maybe we should still apply the rule if it’s a new feature at least. to get a community ack that this is wanted by more than one company 21:33:24 acoles: +1 21:33:34 acoles: you wouldn't +2 it? or you wouldn't +A it without some sanity checking? 21:33:43 notmyname: clayg: bugfixes, improvements etc are fine otherwise imo 21:34:01 I *want* you to review your own contribution to that patch with a critical eye and say "look, in all honestly, I know I worked on this, but this change is ready to land IMHO" 21:34:15 clayg: if I contributed a lot, to real code, I'd rather two others reviewed it 21:34:18 i can consder the source and make the choice to +A or "just" add the second +2 21:34:30 cschwede: +1 21:34:37 but sure we should also leave our own opinions 21:34:47 acoles: i can understand that - if you don't think you can give it an honest +2 you shouldn't 21:35:07 I know I will feel that way about 95% of the code I feel was my own fault 21:35:39 ... but at the same time ... i need a signal ON THE PATCH when you think you can't make it any better 21:35:54 clayg: currently in that case I would +1 it and say why I am not +2, what worries me is that someone else sees my +2 and thinks I have independently reviewed it and add a +2 21:36:14 and a +A and *my* code never got critical review 21:36:55 acoles: didn't the person that +A'd offer critical review? 21:37:44 clayg: what worries me is that the presence of one +2 might influence the second review. but maybe I worry too much. 21:38:03 acoles: I think that's fair 21:38:14 cschwede: none of this is written down - i'm not sure how careful we have to be about putting in place a policy that outlives us and ends up doing the wrong thing for the next generation 21:38:40 i hear what you are saying about diversity being a goal - i don't want to circumvent that - i also want to operate with trust and freedom for sucess 21:38:40 like I said, I have +2/A's stuff I have co-authored, I am just saying there are times when I would be hesitant. 21:38:44 clayg: to be fair, I'm pretty sure some of this is written somewhere in some openstack rules wiki 21:39:02 acpahcpahp - w/e 21:39:04 ;) 21:39:05 acoles: yes! 21:39:30 acoles: I think notmyname is asking us to step out of our comfort zone for the greater good 21:39:35 I think the point is not to make people +A their own stuff. it's to make sure people feel free to use their judgement 21:40:01 acoles: if that means you have to loop back to a patch after a day with it - and re-examine your own code with a critical eye before voice an opinon - that's good 21:40:05 instead of following a process that's slowing down reviews and not substantially adding lots of value 21:40:39 clayg: sure, I get that 21:41:07 clayg: I've also looked at enough of my own code after a month of it being n master and shuddered ;) 21:41:11 cschwede: and YES definately MORE on bug fixes and cleanup than on new features 21:41:19 for new stuff it *has* to take awhile 21:41:28 clayg: yep, exactly 21:41:28 not just for lots of people to chime in and consider options 21:41:37 but for all of us as a community to understand what we're doing 21:41:47 jrichli: kota_: mattoliverau: hosanai_: what do you think? 21:42:06 sounds like a good next step 21:42:19 not too crazy 21:42:40 sounds nice. let's try :-) 21:42:46 I agree, its about people using their judgement. But only +2 if your happy 21:42:54 torgomatic: any thoughts? 21:43:13 sounds ok to progress but I'm feeling I am similar with acoles, sometimes I'd like to more eyes for critical review. 21:43:32 notmyname: seems reasonable 21:43:40 if its something that makes the code better (bug fix, cleanup, docs) and it's better then it was before, then thats good. Bigger new things.. that's when it gets hard. 21:44:00 kota_: I agree. if you aren't ready to +2 something, that's ok 21:44:12 kota_: acoles: maybe this is ust a reminder that the second +2 is not cursory or a formality 21:44:42 ... but esp so when an collaborator had to step out of their comfort zone and examine the patch as a whole - including their contribution with a critical eye 21:45:13 i think comments to each other with our scores is the best thing here - communication is keye 21:45:22 yes 21:45:45 "I helped out with tests; but I think the tests are good - no crazy new infrastructure or weird racy stuff; just standard fare - LGTM" 21:46:00 clayg: agree, communicate to other reviewers. 21:46:26 maybe we should sometimes just ask if anyone else *don’t* sees a need to add a second +2 (after giving it a very quick and brief view), and then just +A with a single +2. 21:46:32 clayg, acoles: yeah, i think gerrit is a tool to collabolate with someone. 21:46:35 "I redid some of the bits in the error handling to make me feel better; otherwise patch looks good; I'm pretty sure the error handling is sane - but would love you some close eyes on that - LGTM" 21:46:54 ok, let's leave this topic here for now. maybe bring it up next week again, and definitely talk more in barcelona. 21:46:54 so we can call someone like 'i want one more +2 for this' 21:47:01 if it's really "I'm exahusted; I can't possibly offer this an unbaised review - I'll leave others to score" 21:47:03 kota_: cschwede: +2 21:47:04 ... so be it 21:47:15 notmyname: (playing devil's advocate) isn't the bigger problem re. context switching that there is a lot of new feature development going on *simultaneously*? and that might distract from bug fixes and older patches? 21:47:23 acoles: yep 21:47:41 notmyname: that discussion could wait for barcelona too ;) 21:47:48 acoles: I'm trying to adjust our tools/mindset/process to fit that reality instead of changing reality ;-) 21:47:54 seems simpler 21:48:11 ok, last topic. top bugs patches 21:48:16 #topic top bugs/patches 21:48:17 well ok, i'm not sure we mad any decisions then 21:48:21 good talk about it more i suppose 21:48:25 clayg: good talk :-) 21:48:46 https://review.openstack.org/#/c/99598/ 21:48:52 clayg: hopefully we encouraged each other to feel free-er to add out own +2/A as co-authors without fearing reprimand 21:48:56 Ensure update of the container by object-updater 21:49:20 that one is from june of 2014. fun! 21:49:38 ah, yeah, that's takashi's one 21:50:21 oh, only june 2014? - so it's not like we going on three years or anything 21:50:25 and it's 5 digit, too old! 21:50:31 no activity in quote a while on this one, and it fixes https://bugs.launchpad.net/swift/+bug/1328735 21:50:31 Launchpad bug 1328735 in OpenStack Object Storage (swift) "Object-updater gives up updating container with no success if all containers are placed at handoff" [High,In progress] - Assigned to Takashi Kajinami (kajinamit) 21:50:44 notmyname: is the patch done? 21:51:05 yes, from what I can tell 21:51:53 I also want to call out https://bugs.launchpad.net/swift/+bug/1624088 -- EC missing durable can prevent reconstruction 21:51:53 Launchpad bug 1624088 in OpenStack Object Storage (swift) "EC missing durable can prevent reconstruction" [High,Confirmed] 21:52:02 patches at https://review.openstack.org/#/c/376630/ 21:52:26 tdasilva left a comment yesterday, but there aren't any review scores 21:52:58 this one matters for failure handing in EC 21:53:09 I need to revisit that tomorrow and reply to any comments 21:53:13 notmyname: i've been reviewing that and the remove .durables one, so it's been back and forth 21:53:26 tdasilva: thanks. and I was just about to mention the other :-) 21:53:42 https://review.openstack.org/#/c/355958/ eliminates .durables and has a +2 from mattoliverau and tdasilva is looking at it 21:53:49 mattoliverau: tdasilva: thank you for looking there 21:54:05 and acoles for writing it, of course :-) 21:54:53 I've seen some other talk in IRC about fixing some of the memory stuff related to pyeclib. what's the status there? 21:54:55 tdasilva's potentiall found something there, so I'll loop back around and have another look. tl;dr I'll keep my eye on it 21:54:57 acoles: kota_: ? 21:55:25 notmyname: not a memory stuff actually 21:55:35 ah, my mistake. what's up then? 21:55:39 kota and I have been working on this newer ec bug which is horrible https://bugs.launchpad.net/swift/+bug/1631144 21:55:39 Launchpad bug 1631144 in OpenStack Object Storage (swift) "Swift Erasure Coding : Error when decoding fragment" [Critical,Confirmed] 21:55:54 acoles: just trying to link, thanks ;-) 21:55:59 ack, thanks https://review.openstack.org/#/c/385609/ 21:56:04 mattoliverau: I figured out what I found is actually fixed by patch 376630 21:56:31 acoles: yes, that has the right importance tag 21:56:44 tdasilva: then maybe I should look at that patch too then ;) 21:56:51 yeah, my mistake for not seeing that earlier 21:57:10 I would put it ahead of the others for priority TBH 21:57:24 yes 21:57:26 yes 21:57:33 me too 21:57:52 also looks like something that would warrant a release and backport when it lands 21:58:36 i'll try that one and then others like eliminate .durable if they didn't land yet. 21:58:45 kota_: thanks 21:58:56 I'll track the critical one 21:59:08 anything else to bring up during the last minute of the meeting? 21:59:27 what happens if you put root beer in a square glass? 21:59:37 torgomatic: what happens? 21:59:43 us talking about priority of patch 1631144 22:00:05 https://media.giphy.com/media/4aFQ6F8s4wgz6/giphy.gif 22:00:45 thanks for coming, everyone. thanks for working on swift 22:00:55 #endmeeting