17:02:24 #startmeeting service_chaining 17:02:25 Meeting started Thu Dec 15 17:02:24 2016 UTC and is due to finish in 60 minutes. The chair is LouisF. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:02:26 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:02:28 The meeting name has been set to 'service_chaining' 17:02:35 hi all 17:02:39 hi 17:02:44 hi 17:02:52 Hi LouisF 17:04:26 agenda for today https://wiki.openstack.org/wiki/Meetings/ServiceFunctionChainingMeeting#Agenda_for_the_Networking-SFC_Meeting_.2812.2F15.2F2017.29 17:04:54 #topic big stadium work 17:05:20 api reference 17:05:40 pcarver: thanks for the patch update 17:05:41 I just split out the API definition from the API reference 17:06:11 I think the API reference is just about done, but the API definition is turning out to be more complicated than I expected 17:06:50 One question on the API reference though. Is tenant_id a valid field in a request? 17:07:06 I was thinking about whether an admin can make a request on behalf of a tenant 17:07:48 pcarver: that could be a possible scenario 17:08:03 If a tenant makes a request they shouldn't have to supply their tenant ID in the request body, but I'm not sure about the admin scenario. 17:09:01 hi 17:09:07 pcarver: what are the issues with the api definition? 17:09:10 mohankumar: hi 17:09:44 I took the tenant_id field out of the request sections of the api-ref, but I've been having issues with getting SFC to work in devstack lately so I haven't tested it yet. 17:10:22 pcarver: ok thanks 17:10:32 The issue with the API definition is that it uses custom validators and one of the validators uses an SFC specific exception so I had to copy those into neutron-lib as well. However, there are objections to that. 17:11:01 So I need to do some more work with the SFC exceptions and validators before I can make the API definition acceptable. 17:11:36 I can't just import them from where they currently are because that would make neutron-lib depend on networking-sfc 17:11:54 which I suspect won't be accepted by neutron-lib folks 17:11:55 which is not desirable. We want networking-sfc to depend on neutron-lib, but not the other way around. 17:12:14 pcarver: yes 17:12:45 So I'll get the validators and exceptions straightened out. I just didn't want it in the same commit. 17:13:03 The api-ref commit was originally just supposed to be documentation 17:13:12 pcarver: right 17:13:34 I added in the API definition because I thought it was just a simple matter of some copy and paste into neutron-lib 17:13:56 but since it's a bit more than simply copy/paste I'd like to decouple it from the documentation commit 17:14:48 pcarver: ok 17:16:17 mohankumar: what is status of cli port? 17:16:32 https://review.openstack.org/#/c/409759/ 17:18:26 please review this patch 17:19:18 mohankumar: there are few comment on it 17:19:42 LouisF: yes 17:19:51 Working on it 17:20:21 mohankumar: thanks 17:20:54 moving on 17:21:05 #topic current work items 17:22:28 fix flow deletion unit tests https://review.openstack.org/#/c/411194/ 17:22:39 bcafarel: thanks for your work 17:23:17 enable pep8 test everywhere https://review.openstack.org/#/c/411387/ 17:23:29 bcafarel: thanks also 17:23:35 please review these 17:23:50 LouisF: np, just giving you some more review work :) 17:25:13 move db migration tests to functional tests https://review.openstack.org/#/c/410818/ 17:26:28 use project_id instead of tenant_id in DB models/OVS driver https://review.openstack.org/#/c/409162/ 17:26:36 please review these 17:27:52 Use ovs devstack plugin functions from neutron https://review.openstack.org/#/c/409800/ 17:29:07 Replaces uuid.uuid4 with uuidutils.generate_uuid() https://review.openstack.org/#/c/396129/ 17:29:58 Validate all flow-classifier parameters while creating port-chain https://review.openstack.org/#/c/409771/ 17:30:08 priya: thanks for your work 17:30:35 will update the unit tests 17:30:38 priya: can you address the comments 17:30:42 priya: thanks 17:31:48 Revert "Using --strict and add priority for del_flows" https://review.openstack.org/#/c/406938/ 17:32:19 bcafarel: is this still needed? 17:33:01 should it be "use project_id instead of tenant_id" of "use project_id in addition to tenant_id"? 17:33:02 LouisF: should not with https://review.openstack.org/#/c/411194/ 17:33:10 s/of/or/ 17:33:28 i.e., should tenant_id be removed at this time? 17:33:53 bcafarel: ok thanks 17:34:08 I recall Igor mentioning problems with flow deletions, but I am not sure if it was related to this change (anyway this can be done in a later review) 17:34:27 bcafarel: ok 17:35:04 pcarver: it is mostly internal use, so the real field is project_id 17:35:19 bcafarel: right 17:35:28 bcafarel: I was asking mainly in reference to the api-ref I'm working on. 17:35:34 Should tenant_id be deleted? 17:35:41 Or should both be shown? 17:36:10 hmm it is going away and there is some magic in the CLI to accept both 17:36:37 I am not sure what should be shown in doc (like listing tenant_id but as deprecated) 17:37:31 can investigate further 17:38:25 symmetric port chain for ovs driver/agent patch https://review.openstack.org/#/c/410482/ 17:38:31 please review 17:38:57 #topic ocata work 17:39:01 pcarver: we probably can do something like in neutron http://developer.openstack.org/api-ref/networking/v2/#tenant-and-project-attributes-in-requests 17:39:40 bcafarel: that's what I was thinking 17:39:56 in the latest patch set of the api-ref I added project_id but did not delete tenant_id 17:40:07 pcarver: looks like a good approach 17:40:13 They would always be expected to be identical, but the API would return both 17:43:30 sfc graph work is under way https://review.openstack.org/#/c/388802/ 17:44:03 symmetric port chain for ovs in review 17:44:53 todo ovn driver work 17:45:26 little progress - swamped with another project 17:45:37 anyone interested in this contact myself or John 17:45:42 I am trying to carve out some time but hard 17:45:51 doonhammer: ok 17:45:57 happy to help someone else to come up to speed 17:46:02 and participate 17:46:40 i have another meeting and will end meeting 17:46:55 anything else for next week? 17:47:28 LouisF: yes, stable newton branch 17:48:00 there is a bug open from people who tried it and got the ovs build failure, plus pending patches blocked as the gates fail for stable branch 17:48:30 please review https://review.openstack.org/#/c/406152/ to unblock these 17:48:37 bcafarel: ok thanks 17:48:54 it's mostly a backport of tests fixes, plus configuration updates to pick the correct branch 17:49:58 bcafarel: will do 17:50:12 LouisF: thanks! 17:50:55 thanks all 17:51:04 bye 17:51:09 #endmeeting