| vkmc | kgriffs|afk, tjanczuk_ hey! :) let's catch up with AMQP tomorrow | 00:04 |
|---|---|---|
| vkmc | peoplemerge, o/ | 00:10 |
| *** nosnos has joined #openstack-marconi | 01:09 | |
| *** reed has quit IRC | 01:50 | |
| *** malini_afk is now known as malini | 02:12 | |
| *** rossk has quit IRC | 02:19 | |
| *** Joe_knock has quit IRC | 02:23 | |
| *** shakamunyi has quit IRC | 02:31 | |
| *** flwang has quit IRC | 03:13 | |
| *** nosnos has quit IRC | 03:20 | |
| *** flwang has joined #openstack-marconi | 03:28 | |
| *** balajiiyer has joined #openstack-marconi | 03:30 | |
| *** balajiiyer has quit IRC | 03:31 | |
| *** whenry has quit IRC | 03:47 | |
| *** oz_akan_ has quit IRC | 03:49 | |
| *** fifieldt is now known as fifieldt-afk | 03:53 | |
| *** whenry has joined #openstack-marconi | 03:59 | |
| *** nosnos has joined #openstack-marconi | 04:00 | |
| *** oz_akan_ has joined #openstack-marconi | 04:21 | |
| *** oz_akan__ has joined #openstack-marconi | 04:23 | |
| *** oz_akan_ has quit IRC | 04:25 | |
| *** oz_akan__ has quit IRC | 04:27 | |
| *** flwang has quit IRC | 04:32 | |
| *** flwang has joined #openstack-marconi | 04:56 | |
| *** vkmc has quit IRC | 05:05 | |
| *** malini is now known as malini_afk | 05:07 | |
| *** oz_akan_ has joined #openstack-marconi | 05:08 | |
| *** haomaiwang has joined #openstack-marconi | 05:17 | |
| *** oz_akan_ has quit IRC | 05:24 | |
| *** haomaiwang has quit IRC | 05:31 | |
| *** haomaiwang has joined #openstack-marconi | 05:32 | |
| *** ykaplan has joined #openstack-marconi | 06:56 | |
| *** renlt has joined #openstack-marconi | 07:05 | |
| *** haomaiw__ has joined #openstack-marconi | 07:15 | |
| *** haomaiwang has quit IRC | 07:18 | |
| *** oz_akan_ has joined #openstack-marconi | 07:23 | |
| *** oz_akan_ has quit IRC | 07:27 | |
| *** oz_akan_ has joined #openstack-marconi | 08:23 | |
| *** oz_akan_ has quit IRC | 08:27 | |
| *** flaper87|afk is now known as flaper87 | 08:41 | |
| *** AAzza_afk is now known as AAzza | 08:43 | |
| flaper87 | kgriffs|afk: mmhh, FWIW, last time vkmc and I talked, we discussed the AMQP 1.0 implementation and I think we'll go down that road. | 08:49 |
| flaper87 | kgriffs|afk: what vkmc is doing now, AFAIK, is helping us with figuring out what endpoints are actually supported by the AMQP driver | 08:50 |
| flaper87 | since we won't be able to fully-support Marconi's API | 08:50 |
| *** oz_akan_ has joined #openstack-marconi | 09:24 | |
| *** oz_akan_ has quit IRC | 09:28 | |
| *** flaper87 is now known as flaper87|afk | 09:50 | |
| openstackgerrit | A change was merged to openstack/marconi: V1 Tests JsonSchema https://review.openstack.org/94212 | 09:56 |
| openstackgerrit | A change was merged to openstack/marconi: Fix sqlalchemy storage driver for Python 3 https://review.openstack.org/95504 | 09:56 |
| *** oz_akan_ has joined #openstack-marconi | 10:25 | |
| *** oz_akan_ has quit IRC | 10:29 | |
| *** oz_akan_ has joined #openstack-marconi | 10:38 | |
| *** oz_akan_ has quit IRC | 10:42 | |
| *** flaper87|afk is now known as flaper87 | 11:04 | |
| *** ykaplan has quit IRC | 11:17 | |
| *** renlt has quit IRC | 11:23 | |
| *** oz_akan_ has joined #openstack-marconi | 11:39 | |
| *** oz_akan_ has quit IRC | 11:43 | |
| *** flaper87 is now known as flaper87|afk | 11:46 | |
| *** vkmc has joined #openstack-marconi | 11:48 | |
| *** vkmc has quit IRC | 11:48 | |
| *** vkmc has joined #openstack-marconi | 11:48 | |
| *** mwagner_lap has quit IRC | 11:53 | |
| *** jamie_h has joined #openstack-marconi | 12:04 | |
| *** abettadapur has joined #openstack-marconi | 12:20 | |
| *** sriram has joined #openstack-marconi | 12:42 | |
| *** sriram has joined #openstack-marconi | 12:42 | |
| *** jchai has joined #openstack-marconi | 12:58 | |
| *** openstackgerrit has quit IRC | 13:06 | |
| *** openstackgerrit has joined #openstack-marconi | 13:07 | |
| *** prashanthr_ has joined #openstack-marconi | 13:09 | |
| *** oz_akan_ has joined #openstack-marconi | 13:12 | |
| *** nosnos has quit IRC | 13:12 | |
| *** jmckind has joined #openstack-marconi | 13:13 | |
| *** jchai is now known as jchai_afk | 13:14 | |
| *** jchai_afk is now known as jchai | 13:16 | |
| sriram | kgriffs|afk: ping | 13:19 |
| abettadapur | sriram: so what kinds of things are we putting in these base classes | 13:22 |
| abettadapur | im thinking enforcing headers there is not good, because then we can't test the absence of headers | 13:22 |
| sriram | alright, lets look at whats currently there in our current base class. | 13:25 |
| abettadapur | all the request methods | 13:25 |
| abettadapur | mock response, app, configs | 13:25 |
| *** malini_afk is now known as malini | 13:26 | |
| sriram | abettadapur: are we talking about this? https://github.com/openstack/marconi/blob/master/marconi/tests/base.py | 13:27 |
| sriram | https://github.com/openstack/marconi/blob/master/marconi/tests/queues/transport/wsgi/base.py | 13:28 |
| abettadapur | yes | 13:29 |
| sriram | https://github.com/openstack/marconi/blob/master/marconi/tests/queues/storage/base.py | 13:29 |
| abettadapur | marconi/tests/base | 13:29 |
| *** malini is now known as malini_afk | 13:29 | |
| sriram | is this what we are gonna split in two? and make them inherit from a single base class? | 13:30 |
| abettadapur | ok, first let me see if my refactoring is ok | 13:31 |
| *** malini_afk is now known as malini | 13:31 | |
| abettadapur | wait sorry, | 13:32 |
| *** Obulpathi has joined #openstack-marconi | 13:32 | |
| *** Obulpathi has quit IRC | 13:32 | |
| abettadapur | the base file /we/ are modifying is /marconi/teests/queues/transport/wsgi/base | 13:32 |
| abettadapur | right? | 13:32 |
| *** Obulpathi has joined #openstack-marconi | 13:32 | |
| sriram | yeah, I think so. since its the transport that has v1 and v1.1 | 13:34 |
| abettadapur | ok | 13:34 |
| abettadapur | so i have split up marconi/tests/queues/transport/wsgi in to v1 and v1.1 | 13:34 |
| abettadapur | i have made four new base classes, v1base v1faultydriverbase v1_1base v1_1faultydriverbase | 13:35 |
| sriram | awesome | 13:35 |
| abettadapur | and put them in marconi/tests/queues/transport/wsgi/base.py | 13:35 |
| abettadapur | and made everything inherit accordingly | 13:35 |
| abettadapur | i think that is what we want, yes? | 13:35 |
| *** oz_akan_ has quit IRC | 13:36 | |
| sriram | yes that is correct, and one way to check is to make sure that it works is running test_v1_0.py and test_v1_1.py | 13:37 |
| *** oz_akan_ has joined #openstack-marconi | 13:37 | |
| sriram | after making changes in each. | 13:37 |
| *** alcabrera|afk is now known as alcabrera | 13:37 | |
| sriram | they do not affect each other. | 13:37 |
| abettadapur | yeah, good idea | 13:38 |
| *** mpanetta has joined #openstack-marconi | 13:43 | |
| *** shakamunyi has joined #openstack-marconi | 13:51 | |
| *** mpanetta has quit IRC | 13:52 | |
| *** mpanetta has joined #openstack-marconi | 13:52 | |
| abettadapur | sriram: they are well separated | 13:54 |
| sriram | woot! | 13:54 |
| abettadapur | what was the problem you were having with tests? | 13:55 |
| sriram | I couldnt make changes to the tests without tests for v1 breaking. | 13:55 |
| abettadapur | oh ok | 13:55 |
| abettadapur | nothing specific then? | 13:56 |
| sriram | if you see both test_v1_1 and test_v1_0 both inherit from the same base class. so essentially there is just one set of tests. | 13:56 |
| abettadapur | right | 13:56 |
| abettadapur | ^(also no longer true) | 13:57 |
| abettadapur | ok, so i'm just going to put the headers in for every request in the v1_1 tests | 13:57 |
| abettadapur | and we can figure out what to put in the base classes | 13:57 |
| sriram | allright. | 13:57 |
| sriram | I would say go ahead and submit a patch, we can work our way through based on the comments. | 13:58 |
| sriram | so do we have 2 sets of tests now? | 13:58 |
| abettadapur | yeah, they are identical | 13:58 |
| sriram | cool. | 13:58 |
| sriram | thats how the functional tests are as well, correct? | 13:58 |
| abettadapur | functional tests inherit from the same base class | 13:59 |
| abettadapur | but its much less of an issue | 13:59 |
| abettadapur | since its just api calls | 13:59 |
| abettadapur | (but there are two sets, yes) | 13:59 |
| *** cpallares has joined #openstack-marconi | 13:59 | |
| sriram | and lets get you assigned to the bp | 14:00 |
| sriram | https://blueprints.launchpad.net/marconi/+spec/decoupling-unit-tests | 14:00 |
| sriram | abettadapur: whats your launchpad id? | 14:00 |
| sriram | malini: ping | 14:00 |
| abettadapur | alexbettadapur | 14:01 |
| sriram | alrighty then, reassigned the bp abettadapur :) | 14:01 |
| abettadapur | awesome | 14:02 |
| abettadapur | ill get some patches up | 14:02 |
| sriram | woohoo :D | 14:02 |
| malini | sriram: pong | 14:03 |
| sriram | I was just wondering if one needed to be a core to reassign bp, I was able to do it. thanks malini! | 14:03 |
| sriram | also were you following the conversation we now have 4 baseclasses. | 14:04 |
| malini | I was not.. | 14:05 |
| * malini scrolls up | 14:05 | |
| malini | sounds good | 14:07 |
| *** balajiiyer has joined #openstack-marconi | 14:10 | |
| *** kgriffs|afk is now known as kgriffs | 14:11 | |
| kgriffs | malini: will you be able to make the mtg today? | 14:14 |
| malini | kgriffs: I will be around for the first 15 minutes | 14:15 |
| kgriffs | ok | 14:15 |
| malini | Then need to leave for the airport | 14:15 |
| openstackgerrit | Sriram Madapusi Vasudevan proposed a change to openstack/python-marconiclient: Enable deleting of message with claim id https://review.openstack.org/97246 | 14:16 |
| malini | kgriffs: & FYI..I am not so keen on the Specs idea. But just wanted us to give it a thought, since most of Openstack seems to be moving in tht direction. | 14:17 |
| malini | So I am perfectly fine if we don't use it :) | 14:17 |
| sriram | hey kgriffs,malini: could I get a review on https://review.openstack.org/#/c/97246/ | 14:17 |
| *** flaper87|afk is now known as flaper87 | 14:17 | |
| sriram | flaper87: hey | 14:18 |
| sriram | flaper87: I have addressed your comments on the patch for the client . Thanks for reviewing it! :) | 14:18 |
| flaper87 | sriram: yo yo yo | 14:20 |
| kgriffs | malini: we could try it with one bp and see how it goes | 14:20 |
| flaper87 | :) | 14:20 |
| * flaper87 opens the request | 14:20 | |
| kgriffs | flaper87: when you have a sec, I wanted to ask about your comment on my oslo cache patch (https://review.openstack.org/#/c/95303/) | 14:21 |
| *** mwagner_lap has joined #openstack-marconi | 14:21 | |
| kgriffs | I didn't realize there was somewhere else to submit patches. Where is that? | 14:21 |
| flaper87 | mmhh, no wait | 14:22 |
| flaper87 | your patch is actually correct. I somehow thought you proposed it on marconi | 14:22 |
| flaper87 | damn, I feel stupid | 14:22 |
| kgriffs | no worries... it's not like you are involved in, like 10 different OS projects or anything. ;) | 14:23 |
| *** haomaiw__ has quit IRC | 14:23 | |
| flaper87 | :P | 14:23 |
| flaper87 | kgriffs: +2 | 14:23 |
| *** haomaiwang has joined #openstack-marconi | 14:23 | |
| kgriffs | cool | 14:23 |
| kgriffs | flaper87: btw, what's the status on the memcached patch? | 14:25 |
| flaper87 | kgriffs: bryansd was working on it. I know he had an almost working/updated version back in the summit | 14:26 |
| flaper87 | but I don't recall seeing it on the oslo reviews | 14:27 |
| malini | kgriffs: Specs will have a lot of overhead..All specs go to a different repo & will need +2s before it is merged to the specs repo | 14:27 |
| kgriffs | yeah, I didn't see it either, so was wondering where it went. | 14:27 |
| flaper87 | malini: re specs, I like the idea of we adopting the process but I've this feeling we're not there yet | 14:28 |
| flaper87 | if you guys think we should, I'm happy with it | 14:29 |
| kgriffs | I think it becomes more useful the larger your team. With a smaller team it is easier to keep everyone on the same page just through the mailing list and IRC | 14:29 |
| kgriffs | and meetings | 14:29 |
| flaper87 | kgriffs: agreed | 14:29 |
| *** oz_akan__ has joined #openstack-marconi | 14:31 | |
| malini | kgriffs, flaper87: +1..I don't think we meed to start using specs yet | 14:32 |
| malini | flaper87: Can you respond to the ML with your thoughts -so the wider community knows why we are not using it ? | 14:33 |
| flaper87 | malini: yup, I'm on it ;) | 14:33 |
| malini | great..thanks :) | 14:33 |
| *** oz_akan_ has quit IRC | 14:34 | |
| *** haomai___ has joined #openstack-marconi | 14:39 | |
| *** haomaiwang has quit IRC | 14:41 | |
| malini | abettadapur: Congratulations on your first merge :) | 14:46 |
| sriram | woot :D | 14:48 |
| sriram | kgriffs: ping | 14:54 |
| sriram | shall I submit a patch for marconi-bench with the producer and consumer? or should we wait until client patch makes it through? | 14:56 |
| sriram | kgriffs: ^ | 14:56 |
| *** balajiiyer has quit IRC | 14:58 | |
| kgriffs | sriram: hmm. did you modify the producer and consumer to check for errors? | 14:58 |
| sriram | not yet, I'll get to that now. | 14:59 |
| vkmc | hiiii | 15:00 |
| *** reed has joined #openstack-marconi | 15:00 | |
| alcabrera | o/ | 15:00 |
| alcabrera | are we having our meeting now: kgriffs, malini, flaper87, flwang? | 15:01 |
| Obulpathi | o/ | 15:01 |
| kgriffs | yeah one sec | 15:01 |
| alcabrera | kk | 15:02 |
| abettadapur | malini: Thanks! | 15:08 |
| openstackgerrit | Alex Bettadapur proposed a change to openstack/marconi: Decoupled Unit Tests https://review.openstack.org/97534 | 15:11 |
| *** jchai is now known as jchai_afk | 15:13 | |
| *** tonytan4ever has joined #openstack-marconi | 15:16 | |
| *** jchai_afk is now known as jchai | 15:21 | |
| *** oz_akan__ has quit IRC | 15:30 | |
| *** oz_akan_ has joined #openstack-marconi | 15:30 | |
| *** balajiiyer has joined #openstack-marconi | 15:30 | |
| *** malini is now known as malini_afk | 15:30 | |
| *** balajiiyer has quit IRC | 15:55 | |
| *** balajiiyer has joined #openstack-marconi | 15:55 | |
| *** jmckind has quit IRC | 15:59 | |
| *** haomai___ has quit IRC | 16:00 | |
| *** abettadapur is now known as abettadapur_lunc | 16:02 | |
| *** flaper87 is now known as flaper87|afk | 16:03 | |
| kgriffs | vkmc: can you summarize some notes around "why 1.0" and some of your initial design thoughts on the wiki page link that I added here? https://blueprints.launchpad.net/marconi/+spec/storage-amqp-1.0 | 16:04 |
| vkmc | kgriffs, sure, I'll update it soon | 16:05 |
| vkmc | I'm currently mapping AMQP 1.0 core functions with Marconi API and already finding some design issues we will have to solve | 16:05 |
| vkmc | https://gist.github.com/vkmc/1816061a3e601a93a42b | 16:06 |
| vkmc | I'll add it to the whiteboard | 16:07 |
| *** flaper87|phone has joined #openstack-marconi | 16:07 | |
| *** flaper87|phone has quit IRC | 16:07 | |
| prashanthr_ | kgriffs: I wanted to ask you if there are any plans of adding Redis as a cache backend to marconi along with using it as a storage backend ? | 16:08 |
| kgriffs | prashanthr_: yes, although the plan is to do that via oslo cache. I've been meaning to write a redis driver for oslo cache for months, but haven't had time to do it yet. If someone <wink, wink> would like to take that on, I think a lot of people in the community would thank them. | 16:10 |
| *** haomaiwa_ has joined #openstack-marconi | 16:10 | |
| prashanthr_ | kgriffs: I would love to :) | 16:11 |
| kgriffs | prashanthr_: btw, once oslo cache has a redis driver, we should try to get the keystone middleware to migrate to oslo so it is no longer tightly coupled with memcached. That would mean splitting the middleware to it's own package, separate from the keystone client, since the client can't afford to take on the oslo chain of dependencies. FYI... | 16:11 |
| kgriffs | prashanthr_: the oslo cache lib lives in the openstack/oslo-incubator repo | 16:12 |
| kgriffs | https://github.com/openstack/oslo-incubator | 16:12 |
| kgriffs | specifically | 16:12 |
| kgriffs | https://github.com/openstack/oslo-incubator/tree/master/openstack/common/cache | 16:12 |
| kgriffs | that would be awesome if you could help us get this done! | 16:13 |
| kgriffs | FWIW, I think bryansd was working on the memcached driver... | 16:13 |
| kgriffs | (I haven't seen it submitted for review yet, however) | 16:14 |
| prashanthr_ | Sure. I will start reading a bit into this. Ohh awesome. memcached would also be a great candidate for caching i guess. | 16:14 |
| kgriffs | https://review.openstack.org/#/q/status:open+project:openstack/oslo-incubator,n,z | 16:14 |
| kgriffs | prashanthr_: thanks! | 16:15 |
| * kgriffs gives prashanthr_ a box of Pop-Tarts™ | 16:15 | |
| prashanthr_ | Sure :). he he caught it ;) | 16:15 |
| * prashanthr_ starts munching | 16:17 | |
| kgriffs | FWIW everyone: Blueprints should link to a wiki page where notes and TBD's are captured. The page name should follow this schema: | 16:18 |
| kgriffs | https://wiki.openstack.org/wiki/Marconi/specs/{blueprint-name} | 16:18 |
| *** cpallares has quit IRC | 16:20 | |
| *** haomaiwa_ has quit IRC | 16:24 | |
| kgriffs | alcabrera, prashanthr_: as part of that work, we may want to submit a patch to global-requirements for hiredis | 16:31 |
| kgriffs | ("that work" meaning redis driver for oslo cache) | 16:32 |
| prashanthr_ | kgriffs: ha ha got that :) Sure. Is there any specific process to follow to doing that ? | 16:32 |
| kgriffs | I believe it is as simple as submitting a patch (via git review) to https://github.com/openstack/requirements | 16:33 |
| kgriffs | specifically: https://github.com/openstack/requirements/blob/master/global-requirements.txt | 16:33 |
| *** haomaiwang has joined #openstack-marconi | 16:33 | |
| kgriffs | the commit comment will need to justify it... basically would like to use redis for caching, and hiredis makes the redis library faster blah blah | 16:34 |
| kgriffs | we are supposed to have any dependency used in an individual project also listed in global-requirements.txt AFAIK | 16:35 |
| alcabrera | kgriffs: that's my understanding, too | 16:35 |
| *** tonytan4ever has quit IRC | 16:35 | |
| prashanthr_ | alcabrera, kgriffs: Sure. I will do that first thing tomorrow. | 16:36 |
| kgriffs | if a hiredis patch to global-requirements.txt is accepted, help me remember to circle back and add it to marconi's requirements.txt | 16:37 |
| prashanthr_ | kgriffs: Sure. | 16:37 |
| kgriffs | prashanthr_: cool | 16:37 |
| kgriffs | prashanthr_: btw, this needs a rebase - https://review.openstack.org/#/c/96017/ | 16:38 |
| *** haomaiw__ has joined #openstack-marconi | 16:44 | |
| *** haomaiwang has quit IRC | 16:45 | |
| kgriffs | vkmc: we should probably also summarize the email thread wrt Rabbit's 1.0 support in the AMQP 1.0 blueprint wiki page | 16:46 |
| vkmc | kgriffs, makes sense, thanks | 16:47 |
| kgriffs | rock on | 16:47 |
| *** haomaiw__ has quit IRC | 16:58 | |
| *** mpanetta has quit IRC | 16:59 | |
| *** rossk has joined #openstack-marconi | 17:02 | |
| *** tonytan4ever has joined #openstack-marconi | 17:06 | |
| *** tonytan4ever has quit IRC | 17:11 | |
| kgriffs | bbl (lunch) | 17:14 |
| *** kgriffs is now known as kgriffs|afk | 17:15 | |
| *** Obulpathi has quit IRC | 17:15 | |
| *** Obulpathi has joined #openstack-marconi | 17:17 | |
| *** mpanetta has joined #openstack-marconi | 17:17 | |
| *** mpanetta has quit IRC | 17:19 | |
| *** mpanetta has joined #openstack-marconi | 17:20 | |
| *** Obulpathi has quit IRC | 17:20 | |
| *** Obulpathi has joined #openstack-marconi | 17:21 | |
| *** whenry has quit IRC | 17:23 | |
| *** tonytan4ever has joined #openstack-marconi | 17:24 | |
| *** shakamunyi has quit IRC | 17:24 | |
| vkmc | lunch brb | 17:26 |
| *** whenry has joined #openstack-marconi | 17:30 | |
| *** AAzza is now known as AAzza_afk | 17:42 | |
| *** kgriffs|afk is now known as kgriffs | 17:45 | |
| tjanczuk_ | vkmc: are you online? | 17:50 |
| *** balajiiyer has quit IRC | 17:54 | |
| *** kgriffs is now known as kgriffs|afk | 17:55 | |
| vkmc | tjanczuk_, I am! | 17:57 |
| tjanczuk_ | what TZ are you working in? | 17:58 |
| vkmc | UTC-3 | 17:58 |
| vkmc | and you? | 17:58 |
| *** balajiiyer has joined #openstack-marconi | 18:00 | |
| tjanczuk_ | UTC-8 (Seattle) | 18:01 |
| tjanczuk_ | Did you have a chance to look into AMQP 1.0 implementation of a storage driver? Do you have any thoughts/issues/design we can discuss? I suppose vast majority of design aspects are going to be the same between AMQP 1.0 and 0.9. | 18:02 |
| vkmc | well there isn't much done over AMQP 1.0 currently | 18:04 |
| vkmc | so I'm still drafting the storage driver to fit Marconi | 18:05 |
| tjanczuk_ | For example, how do you plan to deal with claims, in particular message deletion? | 18:06 |
| vkmc | the main design aspect we will have to deal with is the fact that queues are not a first class citizen in AMQP | 18:06 |
| *** flaper87|afk is now known as flaper87 | 18:06 | |
| tjanczuk_ | Do you think you can get away without introducing additional metadata storage? | 18:07 |
| vkmc | I'm working on that today... do you have an idea in mind? | 18:08 |
| flaper87 | o/ | 18:09 |
| vkmc | flaper87, o/ | 18:09 |
| tjanczuk_ | I read through Mongo implementation yesterday. I am starting to work on this today, I don't have specific ideas at this point. I will drop in questions/ideas back here as I go. | 18:10 |
| *** abettadapur_lunc is now known as abettadapur | 18:11 | |
| vkmc | awesome! I'll keep working on it so we can share some thoughts later :) | 18:12 |
| flaper87 | the design aspects that worry me about the AMQP driver are the changes required to the API. I'd like us to figure that out before we start submitting patches for the driver. We need to evaluate the impact of the driver and how many things we're willing to give up to support this driver | 18:16 |
| flaper87 | jenkins is soooo slow today | 18:18 |
| *** cpallares has joined #openstack-marconi | 18:18 | |
| vkmc | flaper87, I agree with that... | 18:23 |
| tjanczuk_ | I agree. For me the best way to flesh out the issues and understand options is to prototype that thing. | 18:24 |
| peoplemerge | flaper87: tell me about msgpack, is this ticket for removing it as a dependency? | 18:25 |
| peoplemerge | I see the Jan review that dropped it but I see the blueprint describes adding it back | 18:26 |
| flaper87 | peoplemerge: nope, the idea is to add support for msgpack | 18:27 |
| flaper87 | peoplemerge: so, using msgpack instead of json | 18:28 |
| flaper87 | well, the idea is not to remove json but to support both | 18:28 |
| flaper87 | re AMQP, we know the protocols, based on the protocol capabilities we can start mapping the functionality to Marconi's endpoints | 18:29 |
| peoplemerge | flaper87: got it | 18:30 |
| *** oz_akan_ has quit IRC | 18:30 | |
| *** oz_akan_ has joined #openstack-marconi | 18:31 | |
| *** balajiiyer has quit IRC | 18:32 | |
| *** balajiiyer has joined #openstack-marconi | 18:34 | |
| *** kgriffs|afk is now known as kgriffs | 18:34 | |
| *** oz_akan__ has joined #openstack-marconi | 18:34 | |
| tjanczuk_ | re AMQP I don't know the full intent behind the endpoints well enough to be confident doing this exercise "on paper". I need to spend some time with code before I know enough to have such discussion. | 18:36 |
| *** oz_akan_ has quit IRC | 18:36 | |
| tjanczuk_ | Give me a day or two. | 18:36 |
| *** kgriffs is now known as kgriffs|afk | 18:44 | |
| vkmc | flaper87, if you have a moment later, I would love to hear your comments on this https://gist.github.com/vkmc/1816061a3e601a93a42b | 18:48 |
| * flaper87 clicks | 18:54 | |
| *** kgriffs|afk is now known as kgriffs | 18:54 | |
| vkmc | is a WIP, I'm working on claims now | 18:56 |
| *** AAzza_afk is now known as AAzza | 19:02 | |
| *** Obulpathi has quit IRC | 19:04 | |
| *** Obulpathi has joined #openstack-marconi | 19:05 | |
| *** Obulpathi has quit IRC | 19:14 | |
| *** Obulpathi has joined #openstack-marconi | 19:15 | |
| *** oz_akan__ has quit IRC | 19:25 | |
| *** oz_akan_ has joined #openstack-marconi | 19:26 | |
| *** Obulpathi has quit IRC | 19:30 | |
| *** Obulpathi has joined #openstack-marconi | 19:31 | |
| *** Obulpathi has quit IRC | 19:33 | |
| *** Obulpathi has joined #openstack-marconi | 19:33 | |
| *** mkoderer has quit IRC | 19:42 | |
| *** cpallares has quit IRC | 19:59 | |
| *** flaper87 is now known as flaper87|afk | 20:00 | |
| vkmc | kgriffs, alcabrera do you have a minute to discuss some implementation decisions? | 20:03 |
| alcabrera | vkmc: sure thing | 20:03 |
| vkmc | alcabrera, :) awesome | 20:03 |
| *** sriram has quit IRC | 20:03 | |
| *** tonytan4ever has quit IRC | 20:03 | |
| *** sriram has joined #openstack-marconi | 20:03 | |
| alcabrera | :) | 20:03 |
| alcabrera | what's on your mind, vkmc? | 20:04 |
| vkmc | alcabrera, I started to write some code for the AMQP driver using the proton lib in my deployment... in the future I'll add the corresponding dependency | 20:04 |
| vkmc | my main concern is about the way queues are handled | 20:04 |
| vkmc | in Marconi we manage queues as a separate entity from messages whereas in AMQP queues are just another attribute for messages | 20:05 |
| vkmc | some time ago we were discussing about topics, an idea flaper87|afk suggested, and this is exactly the same | 20:05 |
| alcabrera | I gotcha | 20:06 |
| alcabrera | lazy queues | 20:06 |
| vkmc | so... yeah | 20:06 |
| alcabrera | so... | 20:06 |
| alcabrera | I wonder how we'll make this consistent with marconi 1.1? | 20:06 |
| alcabrera | it sounds like | 20:06 |
| alcabrera | it might not even be possible | 20:06 |
| alcabrera | hmmm | 20:06 |
| alcabrera | unless we enforce strict queues from marconi side | 20:06 |
| alcabrera | since marconi tracks that information | 20:06 |
| vkmc | yeah it will require some thinking | 20:07 |
| vkmc | one possibility is to mimic that behaviour in the controller | 20:07 |
| vkmc | but is not efficient nor elegant | 20:07 |
| vkmc | and quite error prone :o | 20:09 |
| alcabrera | yes... :/ | 20:09 |
| vkmc | for instance | 20:10 |
| alcabrera | ah, yes - marconi wouldn't even be tracking the existence of the queue | 20:10 |
| alcabrera | that's right | 20:10 |
| alcabrera | so hmmm | 20:10 |
| alcabrera | (since amqp is storage) | 20:10 |
| vkmc | one thought that crossed my mind (I know this is ugly, embrace yourselves) | 20:11 |
| * alcabrera prepares | 20:11 | |
| vkmc | if the user creates a queue, we just return a 200 OK... because nothing changed from the Marconi side | 20:11 |
| vkmc | the queue is created only when the user sends a message | 20:11 |
| vkmc | we forget about what to do with the queue creation/deletion... but we are hiding the real behaviour | 20:12 |
| vkmc | so... yeah, we have to find a better approach | 20:13 |
| alcabrera | hmmm | 20:13 |
| vkmc | I also thought about sending a dummy message | 20:13 |
| alcabrera | what if we always returned a 201/204 for create/delete queue, respectively? | 20:13 |
| alcabrera | or a 204 even, for create queue? | 20:14 |
| vkmc | but the problem with it is that we are wasting resources | 20:14 |
| vkmc | hmm... we are not doing anything and the "queue" has no content | 20:14 |
| alcabrera | hmm | 20:16 |
| alcabrera | which resource is being wasted? | 20:16 |
| vkmc | that applies to the second idea.... if we send a dummy message just for queues creation | 20:16 |
| alcabrera | ah | 20:17 |
| alcabrera | in that case | 20:17 |
| alcabrera | there's two ways to look at it, I think | 20:17 |
| alcabrera | in both cases, we let users know: "Hey, AMQP + Marconi 1.1 queue ops are really no-ops. No need to use them." | 20:17 |
| alcabrera | so POV 1) if users keep this in mind, they won't send those messages | 20:18 |
| alcabrera | POV 2) we return some code that indicates that this operation is not implemented, some 4xx level code | 20:18 |
| vkmc | hmm | 20:19 |
| alcabrera | in both cases, we can translate it at the storage layer as a no-op that just bubbles back up to the transport layer as a success | 20:19 |
| vkmc | maybe a forbidden? 504 | 20:19 |
| alcabrera | oh wow, hmmm | 20:20 |
| alcabrera | I'd forgotten that existed | 20:20 |
| vkmc | heh | 20:20 |
| vkmc | :) | 20:20 |
| alcabrera | 500s - The Codes We Do Not Speak Of | 20:20 |
| alcabrera | oh wait | 20:20 |
| alcabrera | 504 Gateway Timeout | 20:20 |
| alcabrera | I think 403 is Forbidden | 20:20 |
| alcabrera | and that refers to refusal based on authentication considerations | 20:21 |
| alcabrera | so that's a no-go for us | 20:21 |
| vkmc | sounds like a better approach... we don't lie of what's going on underlying (200 OK yeah you have a queue!) and we get away from changing the queues handling in our API | 20:21 |
| vkmc | now that I'm thinking... the error code for not allowed operations were 405 | 20:27 |
| vkmc | Any http operation , other than the specified will return a 405 (?) <-- | 20:27 |
| alcabrera | lemme check | 20:27 |
| alcabrera | ah | 20:27 |
| alcabrera | method not allowed | 20:27 |
| alcabrera | 405 | 20:27 |
| alcabrera | yeah | 20:27 |
| vkmc | https://wiki.openstack.org/wiki/Marconi/specs/api/v1/responsecodes | 20:27 |
| vkmc | yes :) | 20:28 |
| alcabrera | a tricky thing here is | 20:28 |
| alcabrera | it sounds like it'd take some transport-level modifications to achieve that | 20:28 |
| alcabrera | which owuldn't be applicable to the other storage drivers | 20:28 |
| alcabrera | *wouldn't | 20:28 |
| vkmc | hmm >:| | 20:29 |
| alcabrera | it's another reason I'm leaning towards the 201/204 approach. what do you think? | 20:29 |
| vkmc | yeah that sounds better | 20:29 |
| vkmc | it's still semantically correct | 20:30 |
| alcabrera | cool | 20:30 |
| vkmc | ok then, I'll tackle messages and queues | 20:31 |
| vkmc | after that we have to deal with claims | 20:31 |
| alcabrera | yup! :) | 20:33 |
| alcabrera | I gotta run, vkmc. | 20:33 |
| alcabrera | catch you tomorrow morning! | 20:34 |
| vkmc | k! | 20:34 |
| vkmc | ttyl alcabrera, thanks for the feedback :) | 20:34 |
| *** sriram has quit IRC | 20:36 | |
| *** abettadapur has quit IRC | 20:37 | |
| *** alcabrera is now known as alcabrera|afk | 20:54 | |
| *** reed has quit IRC | 21:00 | |
| *** kgriffs is now known as kgriffs|afk | 21:00 | |
| *** kgriffs|afk is now known as kgriffs | 21:05 | |
| *** oz_akan_ has quit IRC | 21:09 | |
| *** oz_akan_ has joined #openstack-marconi | 21:09 | |
| *** oz_akan_ has quit IRC | 21:14 | |
| *** balajiiyer has quit IRC | 21:23 | |
| *** jamie_h has quit IRC | 21:25 | |
| *** jchai has quit IRC | 21:26 | |
| *** Obulpathi has quit IRC | 21:30 | |
| *** AAzza is now known as AAzza_afk | 21:36 | |
| flwang | kgriffs: ping | 21:43 |
| kgriffs | o/ | 21:46 |
| flwang | kgriffs: a quick question, as for a typical marconi deployment, seems there are only the web server and the backend DB server, am I right? | 21:46 |
| kgriffs | more or less | 21:47 |
| kgriffs | a minimal deployment would look something like: | 21:47 |
| kgriffs | load balancer ---> 2-3 web servers ---> DB | 21:48 |
| mpanetta | hmm | 21:48 |
| mpanetta | Is that really minimal? heh | 21:48 |
| kgriffs | mpanetta: really small machines. ;) | 21:48 |
| mpanetta | Ahh ;) | 21:49 |
| kgriffs | i guess you'd want a couple LB boxes for HA too | 21:49 |
| kgriffs | anyway, I'll defer to mpanetta. :) | 21:50 |
| kgriffs | I gotta run | 21:50 |
| mpanetta | Have a good night :) | 21:50 |
| kgriffs | ciao! | 21:50 |
| kgriffs | o/ | 21:50 |
| mpanetta | And yeah... haproxy type LB would be good. | 21:50 |
| *** kgriffs is now known as kgriffs|afk | 21:51 | |
| flwang | mpanetta: so haproxy (LB server) -> uwsgi + memcached (web server) + mongodb (DB server), is it? | 21:54 |
| mpanetta | pretty much | 21:55 |
| flwang | mpanetta: cool | 21:55 |
| mpanetta | no requirement to use uwsgi, but it is pretty fast | 21:55 |
| flwang | mpanetta: is the memcached a best practice? | 21:56 |
| flwang | for Marconi? | 21:56 |
| mpanetta | It helps speed up queue lookups in a sharded deployment | 21:57 |
| mpanetta | and it is used to cache auth tokens. | 21:58 |
| flwang | nice | 21:58 |
| mpanetta | So in that respect, I suppose so :) | 21:58 |
| flwang | mpanetta: thanks for info, I'm trying to setup marconi in a pre-production env | 21:58 |
| flwang | so you're tips will be very helpful for me | 21:59 |
| mpanetta | Ah cool! | 21:59 |
| flwang | really appreciate that :) | 21:59 |
| mpanetta | Well I will offer any assistance I can :) | 21:59 |
| flwang | mpanetta: i will definitely bug you in the future about Marconi deployment :D | 22:00 |
| mpanetta | ok :) not a bother though :) | 22:00 |
| *** mwagner_lap has quit IRC | 22:02 | |
| *** mpanetta has quit IRC | 22:11 | |
| *** rossk has quit IRC | 22:19 | |
| tjanczuk_ | anyone there? | 22:21 |
| *** rossk has joined #openstack-marconi | 22:31 | |
| *** flwang has quit IRC | 22:52 | |
| *** rossk has quit IRC | 22:54 | |
| *** rossk has joined #openstack-marconi | 22:55 | |
| *** rossk has quit IRC | 22:59 | |
| *** rossk has joined #openstack-marconi | 23:06 | |
| *** rossk has quit IRC | 23:11 | |
| *** lvh_ has joined #openstack-marconi | 23:14 | |
| *** peoplemerge has quit IRC | 23:20 | |
| *** lvh has quit IRC | 23:20 | |
| *** peoplemerge has joined #openstack-marconi | 23:23 | |
| *** amalagon has quit IRC | 23:37 | |
| *** amalagon has joined #openstack-marconi | 23:38 | |
| openstackgerrit | A change was merged to openstack/python-marconiclient: Initial support for openstack-client https://review.openstack.org/87937 | 23:42 |
| openstackgerrit | A change was merged to openstack/python-marconiclient: Updated from global requirements https://review.openstack.org/96826 | 23:42 |
| openstackgerrit | A change was merged to openstack/marconi: Fix handling of request/response body for Python 3 https://review.openstack.org/97096 | 23:47 |
| *** flwang has joined #openstack-marconi | 23:53 | |
| *** oz_akan_ has joined #openstack-marconi | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!