19:01:53 #startmeeting swift 19:01:53 Meeting started Wed Aug 13 19:01:53 2014 UTC and is due to finish in 60 minutes. The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:54 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:01:57 The meeting name has been set to 'swift' 19:02:10 hello, everyone. who's here for the Swift team meeting? 19:02:15 o/ 19:02:16 o/ 19:02:18 o/ 19:02:19 o/ 19:02:24 hello 19:02:26 hi 19:02:32 here 19:02:42 hello 19:02:53 hi 19:02:57 agenda for today is at https://wiki.openstack.org/wiki/Meetings/Swift 19:03:12 "normal" stuff :-) 19:03:24 first up, hackathon 19:03:35 * notmyname notes "hackathon", not "hackathong" 19:03:41 #topic hackathon 19:03:53 tdasilva: hello 19:04:01 no speedos please 19:04:05 hello 19:04:11 lol 19:04:21 the fall hackathon will be during the bridge week between september and october 19:04:27 before the paris summit 19:04:44 in general, there are a couple of primary reasons that we do these 19:05:29 first, it's a good chance to see each other in person and talk face-to-face. this becomes terribly useful, especially when discussing designs and future plans 19:05:42 it's also nice for doing in-person code reviews for big stuff 19:06:27 second, the summits are hard to get stuff done, and not everyone can get their company to pay to go to the summits, so the hackathons are a place where we can focus on the code, rather than the social part of openstack 19:06:59 I've had some people express concern that it seems like there are now requirements to travel around the world 4 times a year (2 summits, 2 hackathons) 19:07:35 although it's nice to go to all of them, certainly don't consider it a requirement. the hackathons are great for code progress. the summits are great for community progress 19:07:43 so, that all being said... 19:08:02 Red Hat is hosting the fall hackathon, and tdasilva is coordinating logistics form their side 19:08:07 thanks tdasilva! 19:08:22 tdasilva: I believe the invites are publicly open now? 19:08:27 ur welcome :) 19:08:29 yes 19:08:32 https://swift-hackathon.eventbrite.com/ 19:08:57 #link https://swift-hackathon.eventbrite.com/ 19:09:10 I tried to add info there regarding airports, hotels, etc 19:09:21 if people have questions feel free to reach out to me 19:09:39 Is easier to get to for some :P Seeing as it's so close to the summit (for traveeling 1/2 round the world) it may or maynot be possible to trick work into sending me to both the hackathon and summit (but I'll ask) ;P 19:10:43 mattoliverau: I'd love to see you there if you can make it 19:11:12 notmyname: how far in advance are hackathon dates scheduled? I ask cuz I will miss out on this one due to a conflict. I'd like to plan to make the next one. 19:11:26 so, to everyone, sign up ASAP if you are able to go. there are a limited number of places 19:11:48 cutforth: mostly it's "hey we want to sponsor, when do you want it?" in an email to me :-) 19:12:19 cutforth: that being said, I'd like to have the one after before the "L" summit in the spring. I've already got a volunteer to host it on the US West coast 19:12:46 ok, if there were a general time frame, that could help with planning 19:13:01 notmyname: is the location for the "L" summit already defined too? 19:13:11 and date? 19:13:24 cutforth: there isn't yet. I'd guess likely in the march(ish) timeframe 19:13:44 tdasilva: no, not that I know of. dates will be in May (it's always May and November) 19:13:58 ok, thanks 19:14:05 any questions around the fall hackathon? 19:14:13 notmyname: thank you, no more questions 19:14:51 ok, moving on 19:15:04 #topic swift gap analysis follow-up 19:15:10 now for the "fun" topic 19:15:59 as I mentioned last week, the TC has done a "gap analysis" of Swift wrt its integration with the rest of OpenStack (all currently-integrated projects have gone through this--we aren't singled out) 19:16:52 https://etherpad.openstack.org/p/swift_gap_analysis is a high-level summary, but the real juicy details are in the meeting logs from the TC meeting last week. if you're curious, then go read those 19:17:00 but, here's the summary (again) 19:17:08 there are 2 "gaps" identified 19:17:23 quotation marks are intentional, because these may not in fact be gaps 19:17:31 but they are points of investigation 19:17:40 the first is about the use of oslo libraries 19:17:53 the second is about releases and versioning 19:18:37 I've talked with various people about all this, including emaisl with swift-core. 19:18:50 importantly, I don't want to characterize this as a "fight" 19:19:37 but it is important to realize that this is unresolved differences that have existed for the past 4+ years. so there is, unfortunately, baggage on each side of the debate 19:19:55 What do the *'s mean in the etherpad next to a few of the oslo libraries? 19:20:02 mattoliverau: I'll get there :-) 19:20:08 sorry :) 19:20:32 basically, the situation is that we should either adopt more of the oslo libraries and the common release cycles or convice the TC that it's ok that Swift not do this 19:21:04 so, as mattoliverau referred to, there is an entherpad where we can collect our thoughts and use as a common scratchpad 19:21:09 #link https://etherpad.openstack.org/p/swift_gap_scratchpad 19:21:28 I've started it with the 2 issues bolded and then some subpoints under each 19:21:45 today in this meeting I want to focus a little on the oslo usage 19:22:14 I've listed what I think are the 3 big questions that need to be answered for each oslo library 19:22:28 (I found the list of oslo libraries by searching pypi for "oslo" 19:22:59 so for each, I'd like to answer what the library is doing, what are we doing (ie what would it replace in swift), and what's the impact of using it 19:23:17 some of them (oslo.vmware being the simplest) are just silly to consider for swift 19:23:29 that's why I put a strikethrough on oslo.vmware 19:23:43 the ones I put a * on are the Big Ones (tm). 19:23:52 ie those are the ones that people have specifically called out 19:24:06 config, messaging, and logging 19:24:34 logging is sorta special because it's not a separate library at this time, but I still want to see an analysis of it 19:24:46 here's what I would like to see 19:25:07 I'd like us, as a group, to each take one or two of these oslo libraries and answer the questions for them 19:25:48 eg with oslo config, figure out what it would look like to use it, what the impact to swift deployers is, migration paths, gaps in the oslo library, etc 19:26:19 the ones with a * are bigger, and therefore I think that if one person takes it, that's the only one they should take 19:26:24 make sense? 19:26:31 (I'm about to ask for volunteers) 19:26:39 any questions so far? 19:27:06 makes sense to me. should we add ourselves to the etherpad to the libs we are going to investigating? 19:27:19 s/investigating/investigate 19:27:43 cschwede_: yes, feel free to add yourself to the etherpad. also, say somethere in here now when you do :-) 19:28:16 that's my call for volunteers. who wants to look at what oslo libraries 19:28:44 re: logging, it seems a bit dirty to me as its still incubated but is suppose to be treated as released. But incubated means no pip.. so does that mean a copy and paste job? 19:29:11 notmyname: k. i’ll have a look at oslo.serialization first 19:29:19 i'll look at oslo.config 19:29:38 made a start already on that 19:29:49 mattoliverau: we're not going to use copy/paste code that is externally managed in swift. so that's a non-starter. but assuming that it will be released, the common perspective will be that we should use it until we convince the TC it's ok that we don't 19:29:54 acoles: thansk 19:29:58 cschwede_: thanks 19:30:17 let's put our names by our comments; I also have stuff to say about oslo.config, and I don't want to put words in acoles's mouth 19:30:31 yes, good point 19:30:48 it's hard to track who said what on an etherpad. 19:31:11 torgomatic: what about simply writing your name at the top so there is a color match to your name? 19:31:34 isn't there an oslo.nametracker? 19:31:38 lol 19:31:41 heh 19:32:12 yes, please put your name next to your comments in the etherpad 19:32:43 cschwede_: can you take another one too? I think i18n is small. also rootwrap 19:32:58 notmyname: yes, i was thinking about i18n 19:33:15 mattoliverau: hurricanerix: gvernik: cutforth: any you can look in to? 19:33:30 and kota_ :-) 19:33:53 oh, now maybe we can volunteer zaitcev for one :-) 19:34:08 oslo.db 19:34:15 i will look over it also 19:34:18 portante_: is that for you? 19:34:19 I'm happy to look at any, I guess I'll start with logging as it's highlighted then maybe using oslo.db and see how it could connect to sqllite etc 19:34:30 for zaitcev ... :) 19:34:36 happy to volunteer others ... 19:34:36 heh! 19:34:39 hurricanerix: thanks for taking logging. I think there's a lot to consider there. let me know if you have any questions 19:34:45 portante: heh 19:34:48 That "db" thing sounds like a database. 19:34:59 Are we importing a database from Oslo now? 19:35:00 mattoliverau: hurricanerix got logging. can you take messaging? 19:35:12 sure 19:35:27 zaitcev: how else will we store all our container listings in Mongo? /s 19:35:29 mattoliverau: thanks. also, I think that one is kinda big, so please talk to me at any time about it 19:35:57 mattoliverau: specifically, the point made by the TC is that we aren't using oslo.messaging and are instead using statsd 19:36:27 I'll take oslo.rootwrap 19:36:28 torgomatic: can you take 3 small ones? db, rootwrap, and oslotest 19:36:43 kota_: ok. can you also look at middleware? 19:36:44 notmyname: k, may need to talk to you about it then.. when I'm more awake :P 19:36:50 mattoliverau: no worries :-) 19:36:58 notmyname: k 19:37:06 notmyname: sure, I can look at those... I think portante is taking db though 19:37:08 mattoliverau: or maybe you can ask torgomatic nicely and he can do messaging and you can take a few smaller one s: -) 19:37:17 unless I misunderstood what he said 19:37:27 (which is entirely possible) 19:37:30 I think portante has only volunteered others 19:37:45 nobody told me he got promoted to management! congratulations! ;) 19:37:49 lol 19:37:58 lol 19:38:39 torgomatic: db and oslotest are all that's left 19:38:48 torgomatic: if you want messaging it's yours, but happy to give it a whirl. If you do take it I'll take the smaller ones that are left :) 19:38:49 notmyname: I can take a look at those two, sure 19:38:55 torgomatic: yes, thanks, it is great feeling ... 19:39:05 torgomatic: thanks 19:39:11 cool, done then 19:39:21 thanks, everyone 19:39:25 * portante would like to take a minute at the end of the meeting to esplain ... 19:39:53 I'm going to the openstack ops meetup in san antonio in a few weeks. I'm specifically going to bring up some of these things to get feedback there 19:40:18 dhellmann: you're editing stuff now? 19:40:33 dhellmann: we already use pbr, that's why it's not on the list (like oslosphinx) 19:40:56 we aren't going to use stevedore and taskflow, are we 19:41:23 notmyname: ah, sorry, I was filling out your list when I saw you said you'd only searched for "oslo" on pypi 19:42:00 I'll look at stevedore and taskflow 19:42:01 zaitcev: stevedore would be used if you're loading driver code, but I don't know if swift does that 19:42:02 zaitcev: I have no diea 19:42:41 I've talked to clayg briefly about stevedore. I'll ping him again about it 19:42:59 dhellmann: what is taskflow? 19:43:20 it manages multi-step processes in a way that lets you pause, restart, and roll them back 19:43:38 josh harlow can fill in more details, he manages that 19:43:50 dhellmann: ok, thanks 19:43:54 np 19:44:03 ok, let's move on 19:44:28 thanks everyone for taking these. please try to have some information for next week's meeting (some, not all) 19:44:45 peluse: around? 19:44:51 I don't think he's here today 19:45:01 #topic reviews 19:45:15 #link https://wiki.openstack.org/wiki/Swift/PriorityReviews 19:45:40 thanks for working on the migration middleware and the keystone reviews. I've seen a bit of back and forth on those this week 19:46:02 mattoliverau: you assigned the affinity+DELETEs bug to yourself on LP. are you working on that? 19:46:06 mattoliverau: if so, what's the status? 19:46:15 #link https://bugs.launchpad.net/swift/+bug/1318375 19:46:19 Launchpad bug 1318375 in swift "DELETE operation not write affinity aware" [High,New] 19:47:45 maybe he stepped off to get a sandwich 19:47:52 or went back to sleep :-) 19:47:59 lol i'm here 19:48:08 heh, ok 19:48:16 lol 19:48:26 mattoliverau: are you working on that bug? 19:48:33 ie a patch 19:48:53 yup, I've just thrown together a swift multi region SAIO for testing it. Will get a version 0.1 up today for a flaming 19:49:02 mattoliverau: awesome. thanks 19:49:17 sorry, irc decided to hang 19:49:25 I understand. been there :-) 19:50:23 the other bug that was reraised this week is addressed by https://review.openstack.org/#/c/113180/. several people have looked at it 19:50:46 I think the only outstanding question there is to prove one way or another if a circular logging reference is properly GC'd or not 19:50:54 which seems like a big deal :-) 19:51:14 so if you like such things, I'd appreciate your comments on the patch :-) 19:51:28 not for a logger whose lifetime virtually matches the process's lifetime 19:51:30 IMO 19:52:33 as to other patches, the migration middleware has had some back and forth from acoles and torgomatic with gvernik 19:52:44 is there anything on that to bring up in the meeting? 19:53:08 can somebody dump the gerrit link here for convenience? 19:53:13 yes, sorry 19:53:24 thanks 19:53:24 https://review.openstack.org/#/c/64430/ 19:53:49 Hi, I would like to get core review about EfficientReplication 19:53:54 https://review.openstack.org/#/c/99824/ 19:54:27 kota_: yes, me too :-) 19:54:27 It's writtend down in PriorityReviews on wiki 19:54:33 (Hi, pot. I'm kettle) 19:55:50 mattoliverau: thanks for a review! 19:56:22 kota_: I think that one is pretty important. I like the ideas, and I'm sorry I haven't had a chance to look at it personally yet 19:56:44 notmyname: ok, 19:56:47 kota_: of course, we've had some back and forths on it, but looks good to me now :) 19:56:51 to all, please keep looking at the priority reviews page, and work your way down it 19:56:54 notmyname: no worries 19:57:04 just a few minutes left... 19:57:10 #topic open discussion 19:57:19 anything else to bring up in the meeting this week? 19:57:35 the SwiftService patch for python-swiftclient could still use another set of eyes 19:58:00 torgomatic: can you add that to the priority reviews page, please? 19:58:00 notmyname: working on keystone V3 review for python swift client 19:58:06 portante: thanks 19:58:06 notmyname: will do 19:58:19 just fyi, i will be away 8/15 - 9/1 19:58:29 acoles: ah, ok. thanks for the head's up 19:58:40 acoles: vacation? 19:58:50 it will be one of my last reviews for a while, as my role going forward on the performance team is changing to more consulting and less development work on swift 19:58:59 notmyname: yep, italy 19:59:03 acoles: enjoy! 19:59:06 acoles: enjoy your vacation! 19:59:17 portante: we'll try to keep finding ways to pull you back! 19:59:20 thanks 19:59:21 portante: :( that's a shame 19:59:24 notmyname: thanks 19:59:37 portante: appreciate your efforts on that review 19:59:44 acoles: welcome 20:00:01 last 30 seconds for comments in the meeting.... 20:00:07 and for others, if I have dropped the ball on anything, please let me know so that I can wind that down properly ... 20:00:36 time's up 20:00:45 thanks, everyone, for coming today 20:00:45 #endmeeting