17:00:33 #startmeeting service_chaining 17:00:34 Meeting started Thu Oct 1 17:00:33 2015 UTC and is due to finish in 60 minutes. The chair is LouisF. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:38 The meeting name has been set to 'service_chaining' 17:00:57 o/ 17:01:13 hi all 17:01:13 o/ 17:01:40 hello 17:01:56 hi 17:02:06 vikram: hi 17:02:18 LouisF: hello ;) 17:02:26 hi, long time no see :) 17:02:41 amotoki: welcome to the party ;) 17:02:53 hi 17:03:02 lets discuss issues relating to the patches 17:03:41 vikram: do you have questions about the patches? 17:04:02 LouisF: Yes I do! 17:04:09 vikram: go ahead 17:04:36 LouisF: amotoki has raised a concern on having migration,db and extension changes together 17:04:49 LouisF: I believe he has a valid point 17:05:19 what is the opinion on this? 17:05:37 we split them up so the review process would be manageable 17:06:15 LouisF: generally speaking, db models and db migration scripts need to be reviewed at the same time. 17:06:56 LouisF: and they need to be consistent. 17:07:07 This is my poiint. 17:07:18 amotoki: +1 17:07:38 we can combine commits 2/6, 3/6 4/6 if all agree 17:07:58 LouisF: Actually, we can have separate patches .. 17:08:12 LouisF: 2/6: Just address db-migration framework 17:08:44 I am fine to combine them. Another option is that 2/6 introduces db-migraetion framework only. 17:08:48 LouisF: 3/6: Flowclassifier db + data model + extension 17:09:01 LouisF: 4/6: for port chain 17:09:21 vikram: i would prefer to have them combined 17:09:26 amotoki: If we have only one then it will become very dense 17:09:33 vikram: I thought amotoki mentioned we want to have DB model and migration in the same patch? 17:09:56 s3wong: yes. right. 17:09:59 s3wong: I mean to have separate patches for flowclassifer and port chain 17:10:18 vikram: prefer to have fc and pc in same patch 17:10:22 addressing their API and DB model togethet 17:10:27 vikram: so no need to have a DB migration framework patch, right? 17:10:49 LouisF: I have no issues, but the patch will become dense again 17:10:58 The original patch was just too huge, but in breaking them up we need to have them in usable chunks 17:11:31 I think it makes sense to group DB model and migration as the first, because that needs to be merged before the other parts can do anything 17:11:47 pcarver_: agree we can have a separate patch for services, drivers 17:12:01 pcarver_, Louis: I don't have issues 17:12:03 pcarver_: +1 17:12:06 but we don't want to have a situation where we have to merge everything simultaneously, otherwise it's no different than one big patch with multiple Gerrit numbers 17:12:10 pcarver_, Louis: We only got to review ;) 17:12:17 In usual convention in neutron, we introduce extensions and db models in a same patch and then implements drivers or other stuffs. 17:12:20 pcarver_: +1 17:12:28 +1 17:12:35 pcarver_: in that case, we should set up the patchset dependencies accordingly 17:12:36 amotoki: that we are anyways doing.. 17:12:50 yeah 17:12:51 amotoki: We have separate patch for services and driver 17:13:10 Only concern is 2/6, 3/6 and 4/6 17:13:22 if we all agree we can combine these 17:13:25 vikram: can you create a patch with api, db and migration and I will create one with services, driver.. 17:13:33 s3wong: I agree. Ideally we would have a linear series of dependent patches where we can focus on reviewing and merging each in sequence 17:14:03 LouisF: does API have to be combined with DB? 17:14:06 LouisF: Ok, I can merge those 17:14:20 LouisF: One more question 17:14:23 pcarver_: and if there exists a patch set that doesn't depend on anything, or if multiple patch sets depend on another patch set, we should publish them accordingly 17:14:37 pcarver_: that what amotoki suggested 17:14:49 that way, we can review them in proper sequence 17:14:56 LouisF: I think we shouldn't merge ovs db changes with that.. You have to do it with the driver patch 17:15:11 LouisF: My question (and I don't have a preconceived answer in mind) is does DB depend on API? 17:15:14 vi ok 17:15:21 I know the API won't work if the DB isn't there 17:15:26 pcarver_: LouisF: extensions can be added separately, but in most cases it has no meaning without DB. 17:15:50 But I think it would be possible to merge the DB + migration changes without needing the API 17:16:08 amotoki: i think they should be in the same patch 17:16:31 we can test a patch with DB and extensions :-) 17:16:34 LouisF: Patch size will come close to 6K 17:16:55 LouisF: I believe Kyle will not have any concern 17:17:19 mestery: ping 17:17:25 I think the size itself is not a problem as long as it is not complicated. 17:17:49 vikram: he did have a comment on the size of our first patch set 17:17:53 amotoki: No issues then .. If we all agree then I can merge 17:18:01 +1 17:18:17 +1 17:18:20 +1 17:18:25 +1 17:18:43 #agree vikram to combine api, db and migration 17:18:44 okay then merge 2/6, 3/6, 4/6 17:19:05 LouisF: You got to take care of OVS data model 17:19:19 vikram: ok will do 17:19:23 LouisF: ok 17:19:47 another issue is the py34 failure 17:20:07 LouisF: Which patch 17:20:26 3/6 17:20:52 LouisF: My mistake.. Will fix it up 17:21:00 ok 17:21:10 LouisF: You won't see any failures in the merged patch ;) 17:21:20 vi great 17:21:36 Another issue I want to bring to notice 17:21:51 vikram: go ahead 17:22:10 Let's review "https://review.openstack.org/#/c/229841"... this address amotoki concerns on parameter issues 17:22:20 We got to agree on this quick 17:22:28 *quickly 17:23:33 it comes from the discussion in the CLI patch https://review.openstack.org/#/c/210008/ 17:23:33 the parameter naming? 17:23:43 yes 17:23:46 yes 17:24:26 specifally? 17:24:34 vikram: are you going to address amotoki's nit comments on the next patchset for https://review.openstack.org/#/c/229841? 17:24:42 we accept both name and ID for neutron port. Then source-port and source-port-id cannot be distinguished. 17:24:43 specifically? 17:25:09 s3wong: yes, i will .. if it gets merged 17:25:31 s3wong: vikram: my comment on 229841 is just a nit. we can merge it as-is. 17:25:48 amotoki: ok 17:26:00 amotoki: ok 17:26:07 vikram, amotoki: OK 17:26:08 I will take care of these in the next patch-set.. 17:26:22 both server + cli patches 17:27:00 sorry for coming in late, but I believe it is worth addressed before the initial merge. 17:27:02 Request cores to kindly review and approve the changes ;) 17:27:51 vikram: reviewed and voted 17:28:00 s3wong: thanks 17:28:48 Some polishing I have done while reviewing 17:28:49 https://review.openstack.org/#/c/229765/ 17:29:08 https://review.openstack.org/#/c/229777/ 17:29:20 Request team to kindly provide their opinion 17:29:33 LouisF: I am done ;) 17:29:49 vikram: ok will review 17:29:59 LouisF: Thanks 17:30:40 vikram: we can add content to those docs? 17:30:56 LouisF: I didn't get? 17:31:50 ignore my q 17:31:57 LouisF: ok 17:32:28 any other comments on these patches? 17:33:07 #topic cli commands 17:33:26 is mohan on? 17:33:41 LouisF: Mohan has verified the cli's.. all are working fine 17:33:56 i see he updated the CLI patch 17:34:01 mohan not able to make it today 17:34:33 LouisF: We need to push another patch-set addressing latest API changes and then we are done ;) 17:34:34 vikram: good to hear, we are also starting to test with those cli commands 17:34:53 LouisF: I will post the patch tomorrow 17:34:59 vikram: great 17:35:17 #topic horizon 17:35:49 vikram: do you know when mohan can post the latest horizon code? 17:36:27 LouisF: Horizon needs more testing.. 17:36:44 LouisF: He will be able to post by next week 17:36:49 meeting 17:36:56 vikram: thanks 17:37:04 I can review it from the perspective of horizon team next week. 17:37:15 amotoki: thanks 17:37:17 amotoki: thanks 17:37:36 #topic testing 17:38:06 i think the unit tests are in good shape 17:38:18 LouisF: how about the agent code 17:38:26 LouisF: by when we can post them 17:38:34 vikram: post today 17:38:45 LouisF: Great! 17:39:12 LouisF: Can you please summarize the patch sequence.. 17:39:20 LouisF: Just to clarify 17:39:40 LouisF: Sorry to interrupt :( 17:40:50 vikram: requirements; combined db, migration and api; services and driver manager; drivers 17:41:17 LouisF: Perfect.. I am on the same page.. thanks 17:41:42 vikram: great 17:41:55 back to testing 17:42:08 we need to add api, full stack .. 17:43:11 any volunteers to start to look at that? 17:44:02 api test i can take care for db patch 17:44:22 db patch i can take care E2E.. 17:44:40 vikram: ok thanks 17:45:04 LouisF: I feel API test we only got to write for db patch.. 17:45:23 as it will have new API's 17:45:34 vikram: yes, i would agree 17:45:38 Please correct me if I am wrong 17:46:30 ok, i will have a look at full stack 17:47:09 any other questions, concerns? 17:47:35 vikram: what do you mean by "API test by db patch"? It is not a unit test. 17:48:01 amotoki: I mean tempest tests 17:48:08 vikram: got it. thanks 17:49:13 vikram: when will you post the combined patch? 17:49:47 LouisF: Will post by tomorrow EOD my time 17:49:56 LouisF: Without API tests ;) 17:50:08 vikram: np, thanks 17:50:26 ok thanks all 17:50:30 LouisF: Will post both CLI and server changes.. so that we are not blocked 17:50:40 vikram: thanks 17:51:09 bye all 17:51:47 #endmeeting