15:00:32 <kgriffs> #startmeeting marconi 15:00:33 <openstack> Meeting started Tue Mar 4 15:00:32 2014 UTC and is due to finish in 60 minutes. The chair is kgriffs. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:37 <openstack> The meeting name has been set to 'marconi' 15:00:42 <flaper87> 'SUP PEOPLEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE????????????????? 15:00:48 <kgriffs> roll call 15:00:50 <kgriffs> o/ 15:00:57 <malini> \o/ 15:01:06 <flaper87> o/ 15:01:09 <alcabrera> hey all! 15:01:12 <balajiiyer> o/ 15:01:15 <saikrishna_> hi all! 15:01:32 <flwang> o/ 15:01:40 <flaper87> wow, so many folks 15:01:49 <alcabrera> we grow every few weeks. :) 15:01:50 <flaper87> It brings tears to my eyes 15:01:56 * alcabrera hands out the poptarts 15:02:10 <malini> alcabrera: the dentist would love tht 15:02:16 <alcabrera> lol 15:02:19 <alcabrera> oh yes 15:02:21 <alcabrera> [$] 15:02:31 <kgriffs> #topic actions from last time 15:02:46 <alcabrera> #link http://eavesdrop.openstack.org/meetings/marconi/2014/marconi.2014-02-18-15.04.html 15:02:48 <kgriffs> just one was noted 15:02:49 <kgriffs> balajiiyer to follow up with Megan and get us a case study for the summit talk 15:02:52 * flaper87 remembers last time actions once in a week 15:03:00 <alcabrera> not much action last time - how unusual! 15:03:04 <flaper87> guess what, it always happens on Tuesdays at 15:00 UTC 15:03:09 <balajiiyer> kgriffs: can we carry forward it to this week? 15:03:10 <kgriffs> heh 15:03:17 <kgriffs> #action balajiiyer to follow up with Megan and get us a case study for the summit talk 15:03:32 <kgriffs> #topic i-3 15:03:51 <alcabrera> Oh yeah - i3 is close 15:03:53 <alcabrera> very close 15:03:58 <flaper87> like hours close 15:03:58 <malini> it is here! 15:04:23 <kgriffs> flaper87: last time they cut it the following morning. Will they cut during the TC meeting this time do you think? 15:04:28 <kgriffs> sorry, not TC meeting 15:04:30 <kgriffs> release meeting 15:04:43 <kgriffs> I know last time was a special case because of zuul 15:04:45 <flaper87> kgriffs: no, it usually closes at some late time in the west coast 15:04:54 <flaper87> so the cut will happen in the morning 15:05:02 <kgriffs> oh good! 15:05:04 <flaper87> but we should block patches today 15:05:15 <flaper87> those that are not worth shipping with i-3 15:05:21 <kgriffs> ok 15:06:08 <kgriffs> I am going to blitz v1.1 over the next few hours. If I can't get all the patches in by the next 3 hours, let's say v1.1 is going to be an experimental feature 15:06:19 <kgriffs> people will have to go get HEAD to play with it at the summit 15:06:31 <flaper87> sure thing 15:06:52 <kgriffs> ok. sorry, I'm an eternal optimist. :p 15:06:58 <kgriffs> other patches? 15:07:07 <kgriffs> #link https://launchpad.net/marconi/+milestone/icehouse-3 15:07:33 <flwang> kgriffs: /health ? 15:07:54 <flaper87> flwang: I think we should hold it until after i-3 15:08:11 <flaper87> and I'm not sure whether it makes sense to let it land after the FF 15:08:13 <flwang> flaper87: do that mean Juno ? 15:08:14 <flaper87> what do you think? 15:08:24 <kgriffs> hmm 15:08:49 <flwang> flaper87: if you guys believe its overall shape is ok, how about lets try to land it before 6 Mar 15:08:57 <alcabrera> have we solidified the health/ping semantics? 15:09:04 <flwang> i-3 but after FF 15:09:06 <alcabrera> if we have, then I'm favorable towards letting it land soon 15:09:14 <alcabrera> and making small fixes to it as needed 15:09:18 <flwang> and yes, it depends on the overall v1.1 status 15:09:23 <kgriffs> are we "allowed" to land features after i-3 but still for icehouse? 15:09:23 <flaper87> ok, lets work on that 15:09:39 <flaper87> kgriffs: you mean after FF 15:09:47 <kgriffs> flaper87: that's what I mean 15:09:50 <kgriffs> FF for i-3 15:09:50 <flaper87> kgriffs: yest but only if a Feature Freeze Exception is granted 15:09:58 <flwang> 4 Mar is FF, 6 Mar is i-3 15:10:09 <kgriffs> ok 15:10:10 <kgriffs> hmmm 15:10:11 <flwang> https://wiki.openstack.org/wiki/Icehouse_Release_Schedule 15:10:12 <alcabrera> I gotta run guys. I think I'll be back in time for the last 10 minutes of this meeting! 15:10:22 <flaper87> flwang: lets work on getting it ready and then ask a Feature Freeze if necessary 15:10:30 <alcabrera> my basic thoughts: api v1.1 and tempest need help 15:10:30 <flaper87> alcabrera: kk, ttyl 15:10:32 <alcabrera> o/ 15:10:49 <flwang> flaper87: ok 15:10:51 <kgriffs> flaper87: yep, I was just under the impression that FF for the last milestone was also FF for the icehouse release, which seems to be correct 15:10:58 <malini> we also need somebody to start working on sqlalchemy tests 15:11:14 <flaper87> kgriffs: yep 15:11:20 <flaper87> malini: mmh, what tests? 15:11:31 <flaper87> there are tests for sqlalchemy already 15:11:35 <malini> the gate jobs against diff backends 15:11:37 <flwang> flaper87: +1 15:11:50 <malini> flaper87: we have only unit tests, rt? 15:12:00 <flaper87> malini: oh ok, yeah, I'll help you creating some gates for that 15:12:00 <flwang> malini: you mean functional test? 15:12:04 <flaper87> malini: nope, functional too 15:12:05 <malini> we need the functional tests running against diff backends 15:12:31 <malini> flaper87: tht is cool..I didnt know tht !! 15:12:38 <flaper87> malini: I'll help you getting the sqlalchemy gate setup 15:12:39 <flwang> kgriffs: seems we're still missing the func test against v1.1 15:12:40 <malini> I'll chk it out 15:12:46 <flaper87> malini: I replaced all the sqlite tests with sqlalchemy 15:12:47 <flaper87> :) 15:12:51 <kgriffs> flwang: yes we are 15:12:53 <flwang> flaper87: malini: cound me 15:13:22 <flwang> kgriffs: do you think it's a blocker for landing the v1.1? 15:13:26 <malini> I was planning to start working on fun tests for 1.1, if somebody can tackle the gate part..I have a good idea of what/how to do tht 15:13:30 <flwang> kgriffs: if so, seems we can't make it 15:14:16 <flwang> malini: count me, after I complete the /health 15:14:32 <malini> flwang: awesome! thabks! 15:14:35 <kgriffs> flwang: well, our only possible solution would be to add functional tests to the mainline and if we find any bugs, we would need to backport those to the icehouse RC 15:14:35 <malini> thanks! * 15:15:22 <flwang> kgriffs: make sense for me, thanks for the clarification 15:15:34 <kgriffs> ok, so I would like to propose a 2-hour super hero sprint 15:15:39 <kgriffs> lots of code, lots or reviewing 15:15:54 <flaper87> kgriffs: sounds good 15:16:00 <kgriffs> then we meet back in #openstack-marconi and decide what our final bp 15:16:05 <flwang> kgriffs: +1 15:16:05 <kgriffs> set of bp's will be 15:16:30 <flaper87> kgriffs: we could probably make today's meeting shorter and use that time for coding too 15:16:34 <kgriffs> ok 15:16:39 <kgriffs> let's speed this along then 15:16:47 * kgriffs seconds flaper87's motion 15:16:53 <kgriffs> #topic tempest 15:16:55 <flwang> flaper87: competing with time? 15:16:59 <kgriffs> malini: status? 15:17:15 <malini> I have a patch outstanding to address the devstack issue 15:17:22 <flaper87> flwang: I'm always competing with time :P 15:17:32 <kgriffs> malini: have you confirmed it will work? 15:17:34 <malini> https://review.openstack.org/77832 15:17:36 <flaper87> malini: mind giving a summary of the latest findings 15:17:41 <flaper87> ? 15:17:57 <malini> kgriffs: it does work on a brand new devstack install when I tried..so hope it'll work 15:18:09 <flaper87> malini: cool 15:18:13 <malini> I pinged a few core reviewers..but havent heard back yet 15:18:25 <malini> flaper87: Its basically a workaround till we figure the stdout issue 15:18:42 <flaper87> looks like a fix to me 15:18:44 <flaper87> :P 15:18:59 <kgriffs> flaper87: the gist is that something still requires stdout but we can't figure out what 15:19:05 <malini> flaper87: hmm…I would like to know why ..but can live with this for now 15:19:11 <kgriffs> we tried monkey-patching sys.stdout and nobody called write on it 15:19:23 <flaper87> mmhh, that's interesting 15:19:34 <flaper87> what's even more interesting is that all projects are using the same logging module 15:19:45 <flaper87> so, I can't think OTOH what's wrong with marconi 15:19:46 <kgriffs> so, we got logging to stop using stdout 15:19:53 <malini> flaper87, flwang: if you are buddies with any devstack core reviewers, try bribing them to look at the patch ;) 15:19:53 <kgriffs> so I don't think it's logging any more 15:19:55 <kgriffs> something else 15:20:12 <malini> flaper87: I think there is something different 15:20:28 <malini> only ceilometer & us specify the log dir in devstack scripts 15:20:30 <kgriffs> anyway, if we can get this to land and revisit for Juno, that would be WUNDERBAR 15:20:50 <flaper87> malini: mmhh 15:20:52 <flaper87> ok 15:21:09 <flaper87> malini: and now you know what project I copied our devstack code fromn 15:21:11 <flaper87> :D 15:21:16 <malini> :D 15:21:17 <kgriffs> devstack does crazy stuff with FD redirects 15:21:17 <flaper87> from* 15:21:17 <flwang> malini: may i know the patch link? I will see what I can do 15:21:29 <flaper87> ok, moving forward 15:21:31 <kgriffs> yep 15:21:32 <malini> flwang: https://review.openstack.org/#/c/77832 15:21:40 <flwang> malini: thanks 15:21:49 <malini> flwang: thank You!! 15:21:57 <kgriffs> kk 15:22:01 <kgriffs> let's git-r-done 15:22:15 <kgriffs> #topic documentation 15:22:18 <kgriffs> balajiiyer: ^^^ 15:22:33 <balajiiyer> kgriffs: Latest patch set 20 needs some review 15:22:53 <flaper87> balajiiyer: gtk 15:22:58 <malini> balajiiyer: can you post the review link plz? 15:23:06 <balajiiyer> https://review.openstack.org/#/c/72001/ 15:23:16 <balajiiyer> There was a trademark issue on 19, that was fixed. 15:23:20 <flaper87> #link https://review.openstack.org/#/c/72001/ 15:23:43 <balajiiyer> We also have operations doc now 15:24:17 <kgriffs> balajiiyer: is that part of the same patch? 15:24:18 <balajiiyer> Oz will be submitting a patch today 15:24:21 <kgriffs> oic 15:24:22 * flaper87 wants an operation gummy bears 15:24:27 <kgriffs> ok, let's get these merged today 15:24:28 <cpallares> lol 15:24:42 <kgriffs> if needed, please file bugs for minor nits and we will follow up 15:24:52 <kgriffs> let's make sure the major stuff is OK 15:25:01 <flwang> kgriffs: +1 15:25:12 <balajiiyer> kgriffs: ok 15:25:21 <kgriffs> #topic sqlalchemy 15:25:32 <flaper87> IT'S DONEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE 15:25:35 <kgriffs> w00t 15:25:40 <flwang> haha 15:25:41 <kgriffs> this one should be done right? https://blueprints.launchpad.net/marconi/+spec/sql-sharding-controllers 15:25:48 <balajiiyer> flaper87: awesome 15:25:56 <flaper87> kgriffs: yeah, alej completed the sharding part 15:26:05 <kgriffs> ETOO_MANY_EES 15:26:11 <flaper87> we need to test it carefully with other backends now 15:26:14 <kgriffs> k, I will update that bp 15:26:18 <flaper87> kgriffs: you mean, EEEEEEEEEEEEEEEEEEETOMANYEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE 15:26:31 <kgriffs> heh 15:26:33 <flwang> kgriffs: I would suggest to make sqlalchemy as the default backend to be more friendly for the new comer 15:26:43 <flwang> kgriffs: flaper87: make sense? 15:26:48 <flaper87> so, I'll help amlini with getting the gate done for that 15:26:53 <flaper87> flwang: it is the default backend 15:27:04 <flwang> flaper87: now? 15:27:07 <flaper87> yes 15:27:08 <flwang> flaper87: coooooooooooool 15:27:15 <flaper87> haha 15:27:27 <kgriffs> can I mark as implemented the SQL Storage Driver? 15:27:29 <flaper87> it's funny how I swapped it and people still thinks they're using sqlite 15:27:30 <kgriffs> I assume so... 15:27:33 <flaper87> muahahahah muahahha muahahaha 15:27:35 <flaper87> :D 15:27:35 <flwang> flaper87: I'm always using mongodb, so i'm not aware of that 15:27:44 <flaper87> well, they're using sqlite but through sqla 15:27:49 <alcabrera> lol 15:27:55 * alcabrera returns sooner than expected 15:27:56 <flaper87> kgriffs: yeah, mark it as done 15:28:00 <flaper87> alcabrera: coool, wb 15:28:04 <alcabrera> kgriffs: +1 for sql marked as done 15:28:13 <flwang> alcabrera: $$$$$$ 15:28:20 * ametts is going to try it with Microsoft Access 15:28:24 <alcabrera> at this point, it's just bug fixes, since it's functionally feature complete 15:28:29 <flaper87> alcabrera: hahahahahahahaha 15:28:32 <flaper87> hahaha 15:28:33 <kgriffs> ametts: that's crazy talk! 15:28:35 <flaper87> ametts: ^ 15:28:49 <alcabrera> ametts: try out the experimental Microsoft Excel backend. :P 15:28:55 <kgriffs> #topic pecan 15:29:13 <balajiiyer> kgriffs: working on perf benchmarks 15:29:26 <balajiiyer> I have numbers with tsung, wrking with ab now 15:29:31 <kgriffs> This can land after the FF, but I want to make sure we are on track to deliver a good, objective report 15:30:04 <balajiiyer> kgriffs: yeah, have to do some due diligence before I can publish numbers, will have it by end of this week 15:30:30 <kgriffs> balajiiyer: rock on 15:30:37 <alcabrera> cool 15:30:46 <alcabrera> #note balajiiyer making good progress on evaluating pecan 15:31:10 <alcabrera> any other thoughts on this topic? 15:31:23 <kgriffs> The more things we can quantify, the better 15:31:55 <balajiiyer> Im using mongodb as the storage driver, and possibly look at in-memory storage drivers 15:31:59 <balajiiyer> any other ideas what else I can do? 15:32:32 <flaper87> balajiiyer: sqlalchemy + mysql 15:32:39 <alcabrera> flaper87: +1 15:32:50 <alcabrera> so, another thought came to mind 15:32:58 <balajiiyer> Im treating this as 'Marconi in Pecan', vs 'Marconi in Falcon', so Im swapping out transport drivers, keeping everything else a constant. 15:33:24 <balajiiyer> flaper87: ok 15:33:29 <alcabrera> There's a lot of talk of performance, and I understand that's a point of interest. It's one of the easier things to measure objectively. 15:33:50 <alcabrera> I'm very interested in the subjective side of the equation, too 15:34:01 <alcabrera> how easy is it to extend marconi using one vs. the other? 15:34:07 <alcabrera> Things of that nature. 15:34:13 <flaper87> balajiiyer: also, lets make sure we come up with some ideas as to how both pecan and falcon can be improved 15:34:18 <alcabrera> I understand that's hard to be objective about, though. :/ 15:34:45 <flaper87> the whole point is not just to find out which is faster, easier etc but to come up with an idea of how we can contribute back to both communities regardless of our final choice 15:34:54 <alcabrera> good words, flaper87 15:34:57 <balajiiyer> alcabrera: yeah, I have a set of evaluation criteria, that is one of those. 15:35:17 <flwang> flaper87: +1 15:35:17 <alcabrera> people will continue to use pecan and falcon for their own projects. Let's try to make both more awesome 15:35:36 <kgriffs> good thoughts 15:35:58 * flaper87 for president 15:36:10 <alcabrera> #note pecan+falcon: how can we make both better? Share feedback with community 15:36:15 * alcabrera votes flaper87 - every time 15:36:49 <balajiiyer> *takes notes* 15:37:10 <flaper87> ok, open discussion and then code ? 15:37:25 <alcabrera> api v1.1 discussion is next, I believe 15:37:45 <flaper87> yeah, we were thinking to make the meeting shorter this week and do a full code immersion 15:37:50 <alcabrera> ah 15:37:52 <alcabrera> okay 15:37:59 <alcabrera> sounds good to me 15:38:01 <kgriffs> #note Goals for Pecan evaluation include: identify areas of improvement for both frameworks where we can contribute, provide guidance on when to use each framework, deliver metrics, and provide well-thought-out arguments for subjective metrics 15:38:12 <flaper87> not sure what kgriffs thoughts are about moving the v1.1 discussion to the enxt meeting 15:38:36 <kgriffs> oh, I'm pretty much ignoring today's agenda. :p 15:38:40 <flaper87> LOL 15:38:44 <kgriffs> crunch time 15:38:45 <flaper87> hahaha 15:38:48 <flaper87> ROFL 15:39:07 <kgriffs> ok, let me just emphasize this point and then let's get to work 15:39:38 <kgriffs> Imagine if we could rewind 18 months or so 15:39:46 <kgriffs> and our team were the ones floating the idea of Pecan 15:40:30 <kgriffs> now, what kind of a report would we create and circulate to other OS teams? 15:40:53 <kgriffs> maybe that isn't quite the right thought experiment, but you get the idea 15:41:28 <kgriffs> actually 15:41:33 <alcabrera> kind of fuzzy 15:41:35 <kgriffs> let's adjust this 15:41:54 <kgriffs> so, here is how I would personally go about introducing pecan 15:42:17 <kgriffs> first, I would find a different project and ask them to basically do what we are doing 15:42:33 <kgriffs> I would want an "outside" opinion 15:42:58 <kgriffs> and I would consider carefully a set of criteria 15:43:19 <kgriffs> then I would take the results to the community and see if I could get another project to give Pecan a try 15:43:43 <kgriffs> preferably a core (AKA "kernel") project: network, storage, compute 15:44:15 <alcabrera> what that tells me, is that this is only the beginning of the evaluation question 15:44:44 <alcabrera> what would API v3 in Nova look like with Falcon is the first thing that comes to mind 15:44:50 <alcabrera> things of that nature 15:45:01 <alcabrera> I'm favorable towards an ongoing experiment 15:45:22 <alcabrera> tool diversity is a powerful thing 15:45:27 <alcabrera> that's it from me 15:46:05 <flaper87> sounds good to me too 15:46:22 <kgriffs> I tried to argue last year for a broader evaluation period on the ML. Not that I think Falcon is inherently "better" but that this is how you create standards in a community; you do it organically and with evaluations, gathering the data over time and working towards a rough consensus. 15:46:27 <kgriffs> But, I was shouted down 15:46:51 <alcabrera> :( 15:46:54 <kgriffs> in any case, I am trying to keep our evaluation as objective and serious as possible 15:47:11 <balajiiyer> *hopes to deliver* 15:47:12 <kgriffs> that is why I explicitly wanted balajiiyer to lead that effort. 15:48:08 <kgriffs> he is brand new to openstack, and started with little to no experience in either framework 15:48:40 <kgriffs> and I have been telling everyone every chance I get to quantify things, be open-minded, and let's do this the right way 15:49:17 <kgriffs> Ideally, we would take our findings to the community and see if we can get another project to do the same excercise 15:50:08 <kgriffs> Could everyone come back and say Falcon totally sucks and nobody wants to use it? 15:50:19 <kgriffs> sure. 15:51:13 <kgriffs> but I suspect the answer is going to be more gray than that and we will find that arguing over which is better: forks or knives 15:51:33 <kgriffs> ...is just silly and a huge waste of time 15:52:04 <kgriffs> whenever someone asks me what framework they should use for their new project, I say: try both and decide which one works best for you, your team, and your project 15:52:16 <alcabrera> +1 15:52:18 <kgriffs> I honestly do 15:52:31 <kgriffs> All I am asking is for promoters of other frameworks to do the same 15:52:41 <kgriffs> this is how you build a community 15:52:55 <flwang> kgriffs: +1 15:52:55 <kgriffs> and yes, I am well aware this is all going on the record. 15:53:07 <kgriffs> someone needed to recognize the elephant in the room 15:53:07 * alcabrera cheers 15:53:40 <alcabrera> #note kgriffs advocates for open-minded evaluation of tools and technologies 15:54:10 <kgriffs> ok, sorry, this ended up taking the full hour after all. 15:54:18 <kgriffs> #topic open discussion 15:55:05 <flaper87> #link https://plus.google.com/events/ckjhm8rggnqvrnspftna845kubk 15:55:23 <alcabrera> oh yeah 15:55:25 <flaper87> Marconi hangout 15:55:26 <alcabrera> March 27 15:55:39 <flwang> flaper87: I will watch it 15:55:50 <alcabrera> #note Marconi G+/Hangout happening March 27 15:56:03 <malini> what is tht for? 15:56:21 <flwang> general or any best practice 15:56:54 <kgriffs> ok, so everyone have a good idea of what they will focus on the next 2 hours 15:57:19 <flwang> I have watched all the Marconi I can get from youtube, including kgriffs', flaper87 and alcabrera's, haha 15:57:27 <kgriffs> oh oh. :) 15:57:39 <kgriffs> #endmeeting