19:00:31 #startmeeting Weekly Poppy Meeting 19:00:32 Meeting started Thu Dec 4 19:00:31 2014 UTC and is due to finish in 60 minutes. The chair is amitgandhinz. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:36 The meeting name has been set to 'weekly_poppy_meeting' 19:00:41 #topic RollCall 19:00:47 who do we have today 19:00:50 o/ 19:01:38 o/ 19:01:59 ..hello poopers... 19:02:03 uuoops 19:02:06 poppers... 19:02:11 hello poppers :) 19:02:12 lol 19:02:20 sorry ... 19:02:34 o/ 19:02:34 lol 19:02:37 o/ ? 19:02:38 miqui__: is now officially my favorite teammate 19:02:43 hahha 19:02:45 LOL 19:02:46 LOL 19:03:02 sorry ...too much caffeine... 19:03:07 o/ 19:03:15 haha ok 19:03:20 lets get the show on the road 19:03:29 #link https://wiki.openstack.org/wiki/Meetings/Poppy 19:03:46 #topic Review Last Week 19:03:49 #link http://eavesdrop.openstack.org/meetings/poppy_weekly_meeting/2014/poppy_weekly_meeting.2014-11-20-19.00.html 19:03:59 only one action item 19:04:06 malini to update guidelines regarding running api tests 19:04:11 malini is out this week 19:04:33 i know the readme files have been updated for running api tests 19:04:42 and some work has been going on around using mimic 19:04:55 there is also a patch out there by cpowell on running mimic docker containers 19:05:33 for those who dont know, mimic lets us intercept requests to third parties so that we can run api tests locally without slowing them down hitting external services 19:05:41 such as fastly, akamai, dns, etc 19:06:10 so it mocks/fakes the call to 3rd party CDNs? 19:06:16 ya 19:06:21 cool..neat.. 19:06:22 thahnks. 19:06:33 Hi everyone 19:06:42 sorry I am late 19:06:46 np 19:07:14 #topic Update on Kilo-1 19:07:19 #link https://launchpad.net/poppy/+milestone/kilo-1 19:07:37 ok so kilo-1 cycle officially ended yesterday 19:07:47 so i will be cutting a release today for it 19:08:09 theres a few outstanding items that need their final +1's so if i can merge them after this meeting that would be cool 19:08:16 so...lets go through the list 19:08:27 tonytan4ever: Create caching rules for whole site or specific content 19:08:34 specifically akamai caching rules 19:08:40 (fastly was already merged) 19:08:43 Yeah 19:08:57 https://review.openstack.org/#/c/138449/ 19:09:00 I finished up Akamai and now just need another +1 19:09:11 please review this everyone! 19:09:21 I have reviewd it 19:09:25 I think ready to be merged 19:09:37 cool. 19:09:45 approved! 19:10:05 obulpathi: Rackspace Cloud DNS Driver 19:10:11 Its done 19:10:15 It need sa +1 19:10:20 I just +1ed on this one. 19:10:21 tonytan4ever: just +1'd it 19:10:22 needs a +1 19:10:22 haha 19:10:28 cool 19:10:33 it is ready to ber merged 19:10:36 approved! 19:10:58 thanks! 19:11:02 the following two bp )DNS driver and Defautl DNS driver are related 19:11:16 yep 19:11:25 they are part of the same pr 19:11:28 malini: Specify provider url in the conf file 19:11:36 i know she fixed it for the fastly provider 19:11:42 i think the other ones remain still 19:11:47 i will move this bp to kilo-2 19:12:26 okay 19:12:38 i also moved other bp that i knew werent done yet to kio-2 19:12:57 lets go through the bugs, and then we can take a look at whats planned for k-2 19:13:08 #topic bugs 19:13:36 https://bugs.launchpad.net/bugs/1392578 19:13:37 Launchpad bug 1392578 in poppy "Next link in GET /services is incorrect" [High,Triaged] 19:14:03 ycombinator_: can you take a look at this again? 19:14:25 yes, as soon as I get fastly credentials and get rid of that error 19:14:28 i think this bug is resolved by the enforcement of project_id 19:14:43 ok, i will assign this to you 19:14:46 thanks 19:15:19 done, also moved it to k2 19:15:29 https://bugs.launchpad.net/poppy/+bug/1392568 19:15:31 Launchpad bug 1392568 in poppy "Cannot GET service with name containing dot" [Medium,Triaged] 19:15:50 this also seemed related to the missing projectid issue 19:16:15 mind if i reassign to you ycombinator_ to confirm its fixed? 19:16:22 no problem 19:17:02 https://bugs.launchpad.net/poppy/+bug/1393558 19:17:03 Launchpad bug 1393558 in poppy "api-tests fail when flavor doesn't exist" [Medium,In progress] 19:18:09 i think this was merged already 19:18:12 looking for patch... 19:19:07 yes it was fixed as part of this: https://review.openstack.org/#/c/136505/ 19:20:02 ok so that brings the K1 cycle to an end 19:20:34 everyone deserves a round of applause as that also completes the core functionality we were targeting 19:20:42 good job everyone =) 19:20:58 :) 19:21:00 \o/ 19:21:01 i will cut the release soon 19:21:03 * edward-fastly golf clap 19:21:21 good job folks... 19:21:39 ok, now moving attention to Kilo-2 cycle... 19:21:41 #link https://launchpad.net/poppy/+milestone/kilo-2 19:22:01 the focus on this milestone is to solidify a lot of what was done in K-1 19:22:20 so you will see many bugs that need squishing have been bought over to here 19:22:34 also there is some tech debt that we need to take care of 19:23:14 amitgandhinz: what is the tech debt [recisely? 19:23:26 eg the message queue driver 19:23:47 Oh, we gotta pay that one. 19:23:49 basically the async workers we have now are ok, but it would scale better using a queue (rabbit, zaqar) etc 19:24:04 we also need to refactor the spotlight stuff 19:24:07 you mean learn about the specific backend? 19:24:10 get to py34 19:24:16 like rabbit...zero, zaqer et.c.. 19:24:35 That's the 'backend' for queue driver 19:24:54 ...right.... 19:24:57 so rather than spawn threads to call out to the cdn providers and dns, we post a message on to a queue, and then have scalable workers that perform the task (eg call out to the cdn provider) 19:25:13 celery? 19:25:21 celery could also be an option 19:25:23 celery is a good candidate I think. 19:25:24 celery + rabbit? 19:25:34 but celery is shy'ed upon 19:25:38 boo 19:25:40 hmm.... 19:25:45 i know the barbican team (i think) had incubation comments around it 19:26:23 with celery we can improve the throughput of poppy by a factor atleast 10X 19:26:27 #link http://lists.openstack.org/pipermail/openstack-tc/2013-December/000432.html 19:26:50 its not anything specific with celery, but the processor backend for celery supports gevent 19:27:03 but we should support it as a potential driver 19:27:11 which means we can have potentially hundreds of requests simultaneously going on 19:27:27 And I think Zaqar is not as suitable as celery in this instance. 19:27:31 with process based queue mechanism, scalability will still be an issue 19:27:35 We can talk about it later. 19:27:39 i agree 19:27:53 when we get to that blueprint lets have a design disussion on it 19:27:58 +1 19:28:07 +1 19:28:14 sure. 19:28:31 #action amitgandhinz to set up a design discussion on the Message Queue Driver 19:29:08 i will also do some work to pull some more bp's into the k2 cycle 19:29:18 fyi, k-2 goes until Jan 22 19:29:26 i use draw.io for diagrams.... 19:29:29 so not that far out with the holidays 19:29:49 oh its really short 19:30:13 yeh the cycles are really only 1.5 months 19:30:17 oh ok 19:30:21 we just werent following them until now =P 19:30:24 I though they are 2 months 19:30:29 me too.. 19:30:49 no, they allow time at the end between the end of k3 and the release for bug fiing 19:30:51 fixing 19:31:00 basically a feature freeze 19:31:17 openstack cycles are 6 months total 19:31:24 oh ok, got it 19:31:39 So work harder or less bps ? 19:31:45 in k-2 ? 19:31:58 #link https://wiki.openstack.org/wiki/Kilo_Release_Schedule 19:32:27 in k2 we should focus on bugs, testing, etc and things that improve the work done in k1 19:32:40 basically lets make what we have done so far nice and solid 19:32:50 a hip sprint... 19:32:50 Understood. 19:33:11 ok 19:33:31 i will do more grooming of the k2 plan in the coming days 19:34:04 ok moving on 19:34:08 #topic new items 19:34:15 nothing was set on the agenda here 19:34:30 kind of alreayd covered the k1/k2 discussion i was going to have =p 19:34:36 #topic open discussion 19:34:44 anyone have anything else they want to discuss? 19:34:54 i do have an item, related to 'home doc' bp.. thanks amitgandhinz for the access 19:34:56 i can now edit... 19:35:14 ..now learning pecan routing so i can figure out the routes to document 19:35:45 it would help me a lot if you guys can email me the routes you have in place so that i can put them in apiary.io 19:36:17 the apiary docs should already list the endpoints that exist 19:36:18 at the end of the day all of you can review the home doc and see if anything is missing or not... 19:36:45 right amitgandhinz... but am not sure yet...am going through the code 19:36:47 you can also see the routes we have in pecan under the /poppy/transport/pecan/v1/... folder 19:36:55 yup.... 19:37:26 this might help: https://github.com/stackforge/poppy/blob/master/poppy/transport/pecan/driver.py#L55 19:37:41 ..jsut not used to pecan object routing.... in some other frameworks can you easily see the routes... 19:37:46 ...but thats my problem.... 19:38:07 as long as you cover the top level endpoints like /services then the user can follow the endpoints to get to the rest 19:38:18 hmm k... 19:38:20 you dont need to list each service the user has in the home doc 19:39:11 amitgandhinz: so basically once the user can see the json payload of the services... then he/.she can see the rest... 19:39:17 yes 19:39:29 a good example is the zaqar home doc: https://wiki.openstack.org/wiki/Zaqar/specs/api/v1 19:39:31 k.. sounds good.. thanks for the heads up... 19:40:01 basically you output the link to /services, /health, /ping, /flavors 19:40:03 k. thanks.. 19:40:13 and the hints (GET, POST, DELETE) etc that are available on each endpoint 19:40:14 yeah seen similar doc pattern.. thanks 19:40:31 k, am set for now.. thanks amitgandhinz 19:40:33 once the user follows /services, they will be exposed the links to each one of their services 19:40:34 cool 19:40:50 ok, any other questions? 19:41:03 Not really questions, https://review.openstack.org/#/c/138449/ 19:41:09 needs another round of +1 19:41:18 because of merge conflict. 19:41:23 ok 19:41:29 will do it 19:42:05 thanks guys. 19:42:27 ok, anything else? 19:42:42 not fromme... 19:43:20 only other comment is lets make sure we are reviewing code on a regular basis 19:43:32 everyone reviews 19:44:34 ok, if there is nothing else.... 19:44:39 thanks everyone! 19:44:55 cya... 19:45:01 see you guys later in poppy channel. 19:45:08 later 19:45:14 #endmeeting