17:03:45 <LouisF> #startmeeting service_chaining
17:03:46 <openstack> Meeting started Thu Jan 26 17:03:45 2017 UTC and is due to finish in 60 minutes.  The chair is LouisF. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:03:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:03:50 <openstack> The meeting name has been set to 'service_chaining'
17:04:04 <LouisF> hello all
17:04:25 <bcafarel> hi
17:05:06 <LouisF> agenda for today https://wiki.openstack.org/wiki/Meetings/ServiceFunctionChainingMeeting#Agenda_for_the_Networking-SFC_Meeting_.281.2F26.2F2017.29
17:06:05 <LouisF> #topic API reference
17:06:38 <LouisF> mohankumar_: hi
17:06:48 <mohankumar_> Hi LouisF
17:07:10 <LouisF> pcarver: how is progress on the api reference?
17:07:27 <pcarver> I got most of it done but was running into HTTP 500 server error when trying to create port chain. I'm not sure if that's still a problem.
17:07:59 <pcarver> I had to switch over to working on the BGPVPN API ref and def, and am now switching back.
17:08:25 <LouisF> pcarver: ok thanks
17:08:29 <pcarver> When I switched back I was having issues with Devstack coming up without any flavors available so that side tracked me for a bit.
17:09:01 <pcarver> I'm almost at the point again where I can try to create port chains
17:09:23 <bcafarel> pcarver: if you still get 500 errors on port chains creation, it may still be https://bugs.launchpad.net/networking-sfc/+bug/1630503
17:09:23 <openstack> Launchpad bug 1630503 in networking-sfc "DB transaction issues since switching to Newton" [Undecided,In progress] - Assigned to Bernard Cafarelli (bcafarel)
17:09:26 <LouisF> pcarver: great
17:09:42 <pcarver> I'm going to spend about 90 minutes on it immediately after this meeting, so if I run into problems I can share that in the networking-sfc IRC channel
17:10:40 <pcarver> If I can't get through it in 90 minutes it's going to get bumped out to tomorrow due to meetings the rest of the day
17:10:49 <LouisF> ok thanks
17:11:25 <LouisF> #topic OSC CLI work
17:11:38 <LouisF> mohankumar_: how is progress on the cli?
17:12:17 <mohankumar_> LouisF , i got som suggestion from reviewers , working on those
17:12:35 <mohankumar_> LouisF , testing chaneges . maybe tmrw will add patchset
17:12:55 <LouisF> mohankumar_: when will you be able to post?
17:13:02 <LouisF> mohankumar_: ok thanks
17:13:14 <mohankumar_> LouisF  :)
17:13:39 <LouisF> #topic tempest testing issues
17:14:12 <LouisF> we have had tempest failures
17:14:29 <LouisF> bcafarel: you have resolved those?
17:15:06 <bcafarel> LouisF: looking good now, there was also a functional test to fix
17:15:18 <bcafarel> but all rechecks went fine so we are good at the moment
17:15:32 <bcafarel> (for the voting tests)
17:15:41 <LouisF> what exactly was the fix?
17:16:48 <bcafarel> there are some details at https://bugs.launchpad.net/networking-sfc/+bug/1655618
17:16:48 <openstack> Launchpad bug 1655618 in networking-sfc "Tempest runs: intermittent single test failure" [Undecided,Fix released] - Assigned to Bernard Cafarelli (bcafarel)
17:17:03 <bcafarel> basically our tempest tests can not run in parallel, there is some configuration for that before
17:17:36 <bcafarel> but in recent changes, API and scenario test sets were running in parallel (probably because of bigger test VMs)
17:18:36 <LouisF> so the  fix is to prevent the tests from running in  parallel?
17:19:21 <bcafarel> short term yes, I think that if we give more specific flow classifier parameters per test, they should not conflict anymore
17:19:42 <bcafarel> but that is a longer fix, I preferred to clear the reviews queue first
17:19:48 <LouisF> that is done in post_test_hook.sh?
17:20:10 <bcafarel> exactly: https://git.openstack.org/cgit/openstack/networking-sfc/commit/?id=d31008bc0c8a829995a310b64519d125f6c055c1
17:21:00 <LouisF> so the tests are invoked one after the other?
17:21:26 <bcafarel> at the moment yes
17:22:03 <LouisF> bcafarel: great thanks for your help
17:22:32 <bcafarel> LouisF: np, I will probably create a bug for the longer term fix so it does not get forgotten
17:22:41 <bcafarel> but in the meantime we can merge reviews :)
17:23:05 <LouisF> bcafarel: yes please do that
17:23:21 <LouisF> there is still a non-voting multnode test that is failing
17:23:32 <LouisF> i will open a bug for that
17:24:10 <LouisF> #topic ocata work
17:25:58 <LouisF> i want to discuss Support MPLS correlation without SFC Proxy https://review.openstack.org/#/c/420339/
17:27:13 <LouisF> igor, bernard can you give a summary?
17:27:36 <igordcard> LouisF: basically it exposes the mpls labels to the vm
17:28:06 <igordcard> the nsh patch included that change, but then I splitted it in 2
17:28:20 <LouisF> so this is a step towards nsh-aware vms?
17:28:47 <igordcard> this is because sfc graphs requires the encapsulation to be preserved.... but we can't merge nsh support yet
17:29:00 <igordcard> yes
17:29:55 <LouisF> looks like correlation is mapped to sfc_encap?
17:31:11 <igordcard> LouisF: I use sfc_encap in some situations, esoecially to differentiate between pp corr and pc corr; but I can change that
17:31:38 <LouisF> can you elaborate?
17:32:21 <LouisF> is there a need to use a different name: sfc_encap vs. correlation?
17:33:35 <LouisF> is there a difference in intended behavior?
17:33:54 <igordcard> no need, I will rename to pc_corr and pp_corr, but simply correlation will become ambiguous
17:34:37 <igordcard> yes, if there is no pp correlation,sfc-proxy flows will be installed, otherwise the encapsulation is exposed to the vms
17:35:17 <LouisF> so pc_corr is the desired encap for the whole chain, and pp_corr is the encap capability of a SF?
17:36:10 <igordcard> yes
17:37:06 <LouisF> good - can you add some description on that in the code and docs
17:37:35 <igordcard> LouisF: I believe I have something but I will improve,sure
17:37:44 <LouisF> igordcard: thanks
17:37:56 <igordcard> LouisF: when is the code freeze for ocata  sfc?
17:38:26 <LouisF> igordcard: not sure
17:40:24 <LouisF> looks like jan 23-27 https://releases.openstack.org/ocata/schedule.html
17:41:30 <bcafarel> yes for major projects it is this week
17:41:47 <bcafarel> (feature freeze that is)
17:42:54 <LouisF> i am not sure on code freeze vs features
17:43:47 <igordcard> perhaps a better question is when would be a good time to release ocata networking-sfc
17:43:50 <bcafarel> mostly it is only smaller patches accepted from next week on to the release
17:44:43 <bcafarel> but yes networking-sfc is marked as independant for releases so it can be different
17:45:08 <LouisF> igordcard: agree - although the release should closely aligned with ocata
17:45:21 <bcafarel> though it will be nice to have a tested ocata release not too far away from the "official" releases
17:45:33 <LouisF> bcafarel: yes
17:45:54 <LouisF> within 3-4 weeks?
17:47:38 <LouisF> i think that major items for this are: sfc reference, osc cli, sfc graph, symmetric chains
17:48:08 <LouisF> doonhammer: hi
17:48:54 <igordcard> LouisF: alright
17:49:03 <LouisF> the first two are really important are they are part of the ocata assessment
17:49:07 <doonhammer> LouisF: Hi
17:49:50 <LouisF> hopefully we can  get them completed in the next week or so
17:50:44 <igordcard> LouisF: graphs are still pointing at ocata though?
17:50:45 <LouisF> other items can be done in pike release
17:50:57 <LouisF> igordcard: ideally yes
17:51:59 <LouisF> we have  done a lot to be compliant with ocata
17:52:47 <LouisF> there has been less focus on new features and more on being good citizens
17:54:13 <LouisF> for pike the project will have a base of ocata compliance
17:54:40 <LouisF> apis, cli, docs, tempest, grafana, etc.
17:55:11 <LouisF> #topic other items
17:55:17 <igordcard> I'm a bit concerned on whether it will be too complicated to change the api in neutron-lib, after ocata
17:55:25 <LouisF> anything else?
17:55:25 <igordcard> to add the graph resource
17:56:34 <LouisF> igordcard: when can you get the patch complete?
17:57:04 <LouisF> i see we have a few weeks...
17:57:20 <igordcard> next week I'll have limited availability, but I just need to finish the driver and add docs
17:57:29 <igordcard> so maybe 1.5 weeks
17:57:41 <LouisF> igordcard: ok
17:58:11 <LouisF> i would prefer to have that work in ocata
17:58:58 <LouisF> ok thanks all
17:59:10 <LouisF> bye
17:59:18 <bcafarel> bye!
17:59:18 <pcarver> bye
17:59:27 <igordcard> bye
17:59:30 <LouisF> #endmeeting