Monday, 2013-10-21

*** nosnos has joined #openstack-marconi00:56
*** nosnos has quit IRC01:10
*** nosnos has joined #openstack-marconi01:11
*** whenry has quit IRC03:13
*** nosnos has quit IRC06:00
*** nosnos_ has joined #openstack-marconi06:00
*** nosnos has joined #openstack-marconi06:47
*** nosnos_ has quit IRC06:48
*** flaper87|afk is now known as flaper8707:13
*** yassine has joined #openstack-marconi08:20
*** nosnos has quit IRC09:01
*** nosnos has joined #openstack-marconi09:01
*** flaper87 is now known as flaper87|afk10:33
*** fifieldt has quit IRC10:49
*** flaper87|afk is now known as flaper8711:24
*** tedross has joined #openstack-marconi11:51
flaper87kk, tests seem to be working again12:23
flaper87some of these tests were wierdly succeeding12:23
*** jraim has quit IRC12:47
*** malini_afk is now known as malini12:53
*** malini is now known as malini_afk13:04
*** malini_afk is now known as malini13:05
*** nosnos has quit IRC13:09
*** nosnos has joined #openstack-marconi13:10
*** nosnos has quit IRC13:14
*** mpanetta has joined #openstack-marconi13:14
*** malini is now known as malini_afk13:16
*** jraim has joined #openstack-marconi13:19
*** jraim_ has joined #openstack-marconi13:19
*** jraim_ has left #openstack-marconi13:20
openstackgerritFlavio Percoco proposed a change to openstack/marconi: Migrate from nosetest to testr  https://review.openstack.org/5244313:31
openstackgerritFlavio Percoco proposed a change to openstack/marconi: Move misplaced test under the right package  https://review.openstack.org/5292513:34
*** amitgandhi has joined #openstack-marconi13:38
*** alcabrera has joined #openstack-marconi13:40
alcabreraMorning! :D13:40
*** oz_akan_ has joined #openstack-marconi13:41
flaper87alcabrera: GOOOD MORNING13:42
alcabreraflaper87: w00t. How are you? :)13:42
flaper87alcabrera: I'm good, you?13:45
* flaper87 is still fighting with testr13:45
alcabreraDoes it break some tests?13:45
alcabreraahh, I see Jenkins is not enthused about the change. :P13:46
flaper87alcabrera: yeah, plus, testr is very cryptic13:47
flaper87no idea what's failing13:47
flaper87:P13:47
alcabreraick. :/13:48
alcabreraparallel, but at the cost of comprehensibility?13:48
flaper87alcabrera: yeah, I'm asking my self the same thing13:58
flaper87what keeps me going is that nose did hide some tests and that was not cool13:58
*** malini_afk is now known as malini14:02
alcabreraflaper87: we'll see after using it for some while.14:03
alcabreraIf nothing else, if there's a problem, there's probably something we can contribute to make it better.14:03
zyuani don't know why this fails either https://review.openstack.org/#/c/52692/14:04
zyuanit works on my machine and py27 box, just not py26 one ran by jenkins14:05
alcabrerazyuan: it is pretty weird. Nothing stood out to me that would make it explode.14:07
*** jcru has joined #openstack-marconi14:07
zyuananyway.. please review. that on is done actually14:10
alcabrerazyuan: will do shortly. :)14:12
flaper87alcabrera: indeed, for now, I'll keep that review with -214:15
flaper87to make sure no one tries to review it14:15
openstackgerritFlavio Percoco proposed a change to openstack/marconi: Move misplaced test under the right package  https://review.openstack.org/5292514:17
flaper87alcabrera: ^ quick one14:17
alcabreraflaper87: very quick +2!14:18
flaper87alcabrera: thank you!14:25
alcabreranp. :)14:25
flaper87btw guys, I'll be out for the next 2 days, same reason why I haven't been around that much!14:26
flaper87kgriffs_afk: make sure you read this :)14:26
flaper87^14:26
flaper87damn, stupid irc14:26
alcabreramore conferences, flaper87? :)14:28
flaper87alcabrera: hehe, nope, family in town!14:29
flaper87There's MongoTorino this w/e, though!14:29
* flaper87 will be there14:29
alcabreraooooh, right. Family!14:30
alcabreraHow goes the visit?14:30
*** jraim has quit IRC14:31
*** jraim_ has joined #openstack-marconi14:33
flaper87alcabrera: amazing, having a lot of fun after 2 years w/o seeing each other! :D14:35
alcabrerawonderful, I'm glad to hear it. Enjoy the good times! :)14:36
fvolleroHi fellas :)14:38
alcabrerafvollero: Hey hey! :)14:39
fvolleroi was surfing on the internet yesterday and i saw this https://github.com/Comcast/cmb from here I got the idea, why not also supporting cassandra as they're doing? :)14:40
alcabreraThat'd be pretty awesome. Honestly, I'm all for supporting more storage backends!14:41
alcabreraI read this (http://andrewwdeane.blogspot.com/2013/05/the-reliability-of-go.html) yesterday and made me curious about LevelDB.14:42
alcabreraThere's so many flavors nowadays.14:42
flaper87fvollero: you do like procrastination, don't you?14:50
flaper87I made this yday: https://github.com/FlaPer87/cookiecutter-marconi-transport14:51
flaper87not sure if it works, though :D14:51
flaper87I haven't tested it14:51
flaper87I'll create one for storage later14:51
fvolleroflaper87: no :(14:51
fvolleroalcabrera: :)14:52
fvolleroalcabrera: how's thing going ?14:53
flaper87fvollero: no what ? I don't trust you! >.>14:53
flaper87:D14:53
fvolleroflaper87: If i have to change everytime tasks...14:53
fvolleroflaper87: you already been in MXP office?14:54
fvolleroflaper87: <314:54
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(shard): queue listing  https://review.openstack.org/5269214:54
flaper87fvollero: nope, but I alredy made sure my package got here safe! :D14:54
fvolleroflaper87: ahahhaahaha14:55
alcabrerafvollero: things have been going well since this weekend. There's a lot of things I'm working to figure out lately beyond the tech space. :)14:55
*** vkmc has joined #openstack-marconi14:55
*** vkmc has quit IRC14:55
*** vkmc has joined #openstack-marconi14:55
fvolleroalcabrera: same here :) i needed to refresh my mind14:55
flaper87duuuuuude, you know that stuff I didn't test? it works14:57
flaper87o.014:57
flaper87I'll have to add more base code for tests and it's done14:58
fvollerommmm14:58
flaper87next step: cp THAT_STUFF STORAGE_STUFF && git push14:58
flaper87:D14:58
fvollerohahahaha14:58
alcabrerawhich stuff that you didn't test? o.o14:58
flaper87alcabrera: https://github.com/FlaPer87/cookiecutter-marconi-transport14:59
flaper87:)14:59
flaper87basically, newcomers (or old ones) that want to create a transport for Marconi, won't have to struggle with the code structure15:00
flaper87they can just do: cookiecutter https://github.com/FlaPer87/cookiecutter-marconi-transport.git15:00
flaper87answer 4 questions and BOOM15:00
alcabreraahhh15:00
alcabreraI see15:00
alcabreraOkay, that thing that would use cookiecutter. :)15:00
alcabreraNice!15:01
flaper87yup :)15:01
flaper87alcabrera: what? did you expect it to have any kind of documentation?15:01
flaper87:D15:01
alcabrera;)15:01
openstackgerritZhihao Yuan proposed a change to openstack/marconi: fix: one less unnecessary modification  https://review.openstack.org/5069215:01
alcabreraI'm going to review zyuan's patches, then proceed to re-re-re-rebase the admin API feature line. :P15:02
*** JRow has joined #openstack-marconi15:05
*** JRow has left #openstack-marconi15:06
*** kgriffs_afk is now known as kgriffs15:11
* flaper87 does that as well15:14
alcabreraflaper87: jenkins blew up on this one (https://review.openstack.org/#/c/52925/) because of py26 random errors. I'm running a recheck and reapproving if we can get past the bug.15:15
flaper87alcabrera: just set the reverify trigger15:16
flaper87alcabrera: btw, when the patch was approved, you need to use reverify instead of recheck15:16
alcabreraahh15:17
alcabreragtk15:17
openstackgerritA change was merged to openstack/marconi: Move misplaced test under the right package  https://review.openstack.org/5292515:19
flaper87alcabrera: quick one: https://review.openstack.org/#/c/52692/15:20
alcabreraflaper87: already reviewed that one. :P15:21
flaper87alcabrera: mmh, don't see your comment / vote there.. stupid gerrit15:22
flaper87alcabrera: I meant this: https://review.openstack.org/#/c/50692/15:22
flaper87:D15:22
flaper87sorry15:22
* flaper87 slaps himself15:22
alcabreralol15:24
alcabrerazyuan, flaper87: +2 - approved!15:24
openstackgerritA change was merged to openstack/marconi: fix: one less unnecessary modification  https://review.openstack.org/5069215:28
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: split queues API into public/admin  https://review.openstack.org/5070115:30
alcabrerazyuan, flaper87: Can I get a review on ^^? I'd love to get that one in soon. :D15:30
flaper87alcabrera: yup, right away15:31
kgriffslooking at it now15:31
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: add shard management resource  https://review.openstack.org/5070215:31
alcabreraflaper87, kgriffs: Thanks!15:31
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: shards storage controller interface  https://review.openstack.org/5072115:32
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: shards mongodb driver + tests  https://review.openstack.org/5081515:33
flaper87alcabrera: why do we need a 'public' package at this level? https://review.openstack.org/#/c/50701/11/marconi/queues/public/bootstrap.py15:36
flaper87I understand the need under transport/wsgi/ but not sure about that one15:36
*** berryciderspider has joined #openstack-marconi15:38
alcabreraflaper87: hmm... the rationale is a little blurry to me now, but I did this to make it easy to have a public wsgi app and an admin wsgi app.15:38
berryciderspiderHow does falcon fair against node?15:38
flaper87alcabrera: mmh, TBH, I don't like having a 'public' package at that level. Let me review the patch a bit further and see if I can come up with a better solution15:39
alcabreraThe correct storage drivers are chosen, for example (though that's not implemented yet - separate storage drivers for the admin/public interfaces, since they target very different responsibilities).15:39
alcabreraI'm open to suggestions, though!15:39
alcabreraberryciderspider: kgriffs - thoughts on node vs. falcon?15:39
berryciderspiderthanks15:40
flaper87alcabrera: mmh, AFAIU, it's not actually being used now, you added both packages but they're not used - besides from public.15:41
flaper87alcabrera: My thinking is that the bootstrap process should be agnostic of what API version / type it's loading15:42
flaper87it should just know what API to load15:42
flaper87I mean, it should know where the transport driver is, load it and run it!15:43
alcabrerahmmm, I wonder how I could make that happen...15:43
flaper87what about a admin flag ?15:43
alcabreraI see your point, though.15:43
flaper87in the config15:43
alcabreraahh, something like "admin_mode"?15:43
flaper87that can be used throughout Marconi15:43
flaper87yup15:43
flaper87That can be used for the proxy, notification and queues15:43
alcabrerathat would do it15:44
alcabreraoh, hey... hmm...15:44
alcabreraI just realized something.15:44
alcabreraMeteing in 15 minutes?15:44
alcabrera*Meeting15:44
flaper873 birds with 1 bazuca15:44
flaper87:P15:44
flaper87yup, I'm already in m-alt15:44
flaper87kgriffs: you around for the meeting?15:45
alcabreraI'll be popping in there now, because I'm liable to forget again in 5 minutes. :P15:45
kgriffsyes, I'll be there15:47
flaper87kgriffs: cool beans!15:47
* flaper87 checks the action bullets and starts targetting them15:48
flaper87alcabrera: I added a comment based on our recent discussion to that patch15:51
kgriffsflaper87, alcabrera: just thinking about the bootstrap thing15:52
kgriffsIt is helpful to consider what the operator's experience would be15:52
kgriffsIf I go to deploy Marconi, I need to run it twice15:53
kgriffsif I am deploying with, say, gunicorn15:53
kgriffswhat would be the ideal way I would want that to work?15:53
flaper87kgriffs: I was thinking about admin nodes also exposing the public API15:53
flaper87but they would be used internally15:54
flaper87mmh, ok, I understand what you mean15:54
kgriffsso, today there is that WSGI app module15:54
* alcabrera listens15:54
flaper87you don't actually need to 'deploy' it twice, you just need to run it twice15:55
flaper87I'd say we should also have and admin_app15:55
flaper87that can be run under gunicorn15:55
flaper87meeting time15:55
flaper87oh no, 5 mins left15:55
flaper87well, the other meeting ended already15:56
kgriffsyep, mtg soon15:56
kgriffslooking at alcabrera's patch, there is15:56
kgriffswsgi.admin.app15:56
kgriffsand15:56
kgriffswsgi.public.app15:56
flaper87kgriffs: yup15:57
kgriffsthat seems to make sense. The driver is in there two, where the routes are set up, which are obviously unique to each api15:57
kgriffss/two/too15:57
flaper87yup15:57
* kgriffs is constantly amazed at the kinds of typos his brain comes up with15:57
kgriffslast word15:58
kgriffsI think that we are OK having each of those apps just pass in access mode to bootstrap15:58
kgriffsI guess we would also need two commands15:58
alcabreramarconi-server, marconi-server-admin15:58
kgriffsor one command with an option15:58
kgriffsmarconi-server --admin15:59
alcabreraahh, that would work, too15:59
flaper87kgriffs: yeah, that's what we were thinking as well15:59
flaper87an admin parameter15:59
alcabreraoption 2 probably leads to less duplication of code (options)15:59
flaper87that will mark a node as an admin node15:59
flaper87it can also be used in other modules like proxy and notification15:59
kgriffssee you guys over in #openstack-meeting-alt16:00
*** berryciderspider has quit IRC16:22
*** yassine has quit IRC16:38
flaper87awesome meeting, yet again17:12
flaper87ttyl guys!17:12
flaper87take care!17:12
kgriffsttfn17:12
*** flaper87 is now known as flaper87|afk17:13
alcabreraflaper87|afk: take care! :)17:14
*** kgriffs is now known as kgriffs_afk17:25
alcabrerazyuan: Check this out: http://paste.openstack.org/show/48888/17:46
*** kgriffs_afk is now known as kgriffs17:46
alcabrerathe object_size approach is pretty good - 14us in Python 3.3.2 for a fairly large document.17:46
alcabrera**pretty good perf-wise17:47
*** jdaggett has joined #openstack-marconi17:54
*** vkmc has quit IRC17:55
*** jdaggett has quit IRC17:58
*** jdaggett has joined #openstack-marconi17:58
*** kgriffs is now known as kgriffs_afk17:59
zyuanalcabrera: that's large but not complex18:13
zyuanhmm, caching18:14
alcabrerayup, it's not a comprehensive benchmark.18:15
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: add catalogue storage driver for queues  https://review.openstack.org/5216018:16
zyuanus = ? ┬Ás?18:16
alcabrerayup18:25
alcabrerazyuan: ^18:25
zyuan...18:25
*** kgriffs_afk is now known as kgriffs18:28
*** ametts has joined #openstack-marconi18:32
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: split queues API into public/admin  https://review.openstack.org/5070118:34
alcabreraSee you guys later. :)18:35
*** alcabrera has quit IRC18:35
zyuankgriffs: ping18:45
*** jdaggett has quit IRC19:06
*** reed has joined #openstack-marconi19:21
*** malini is now known as malini_afk19:38
*** jdaggett has joined #openstack-marconi19:41
*** whenry has joined #openstack-marconi19:42
*** jdaggett1 has joined #openstack-marconi19:42
*** jdaggett1 has quit IRC19:46
*** jdaggett has quit IRC19:46
*** vkmc has joined #openstack-marconi20:05
*** whenry has quit IRC20:23
zyuanhack simplejson for document size checking; throw JSONDecodeError when oversized during parsing (python impl only):20:32
zyuanhttp://paste.openstack.org/show/48899/20:32
*** jdaggett has joined #openstack-marconi20:45
zyuan... i noticed that the biggest blocker is openstack requirements... i don't think they will allow a hacked version...20:58
*** jcru has quit IRC21:16
*** malini_afk is now known as malini21:17
*** malini is now known as malini_afk21:26
*** oz_akan_ has quit IRC21:37
zyuani also noticed that the use of parse_int hook may decrease the performance of JSON parsing.  simplejson has a fast path to parse with a C function, PyInt_FromString21:37
*** tedross has quit IRC21:55
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(shard): queue listing  https://review.openstack.org/5269222:11
*** mpanetta has quit IRC22:12
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(shard): queue listing  https://review.openstack.org/5269222:17
zyuani surrender; that sucks22:24
*** amitgandhi has quit IRC22:33
*** kgriffs is now known as kgriffs_afk22:35
*** jdaggett has quit IRC22:39
*** vkmc has quit IRC22:42
*** jdaggett has joined #openstack-marconi23:10
*** jdaggett has quit IRC23:19
*** jdaggett has joined #openstack-marconi23:45
*** jdaggett has quit IRC23:50

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!