17:14:21 <LouisF> #startmeeting service_chaining
17:14:22 <openstack> Meeting started Thu Jan 12 17:14:21 2017 UTC and is due to finish in 60 minutes.  The chair is LouisF. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:14:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:14:26 <openstack> The meeting name has been set to 'service_chaining'
17:14:56 <LouisF> hi all apologies for late start - my train was late
17:14:59 <doonhammer> Hi all - happy new year
17:15:10 <bcafarel> hi LouisF, no problem
17:15:15 <LouisF> doonhammer: happy new year
17:15:20 <pcarver> hello
17:16:23 <LouisF> #topic agenda
17:16:58 <LouisF> last weeks we discussed the API ref, and the OSC client work
17:18:39 <LouisF> pcarver: is there progress on the api ref work?
17:19:31 <pcarver> LouisF: Sort of, indirectly. I've been working on both bgpvpn and sfc API ref. I was having devstack issues which prevented me from doing verification of all examples.
17:19:57 <pcarver> I've got it all sorted out on the bgpvpn side and running stack.sh right now on an sfc devstack.
17:20:09 <LouisF> pcarver: thanks
17:20:36 <pcarver> hopefully everything is all fixed up. Then I'm just going to run curl commands to validate the accuracy of all the examples and I should be done
17:21:09 <LouisF> pcarver: great
17:21:58 <LouisF> mohankumar: great timing - how is sfc for osc work?
17:22:13 <mohankumar> Hi louis , its going good
17:22:53 <mohankumar> Got some comments , will address tmrw
17:23:03 <LouisF> https://review.openstack.org/#/c/409759/  has some comments
17:23:17 <mohankumar> LouisF: yes
17:23:41 <mohankumar> Will address shortly
17:23:51 <LouisF> mohankumar: when will you post a patch to address comments?
17:23:55 <LouisF> mohankumar: ok thanks
17:25:10 <mohankumar> LouisF:  I wsbooked with othrt
17:25:24 <LouisF> mohankumar: ok understand
17:25:45 <LouisF> #topic latest patches
17:26:30 <LouisF> look like latest patches are encountering a tempest failure
17:26:57 <LouisF> for example http://logs.openstack.org/00/409800/2/gate/gate-tempest-dsvm-networking-sfc-ubuntu-xenial/e7994eb/
17:27:58 <bcafarel> indeed, I created https://bugs.launchpad.net/networking-sfc/+bug/1655618 for it
17:27:58 <openstack> Launchpad bug 1655618 in networking-sfc "Tempest runs: intermittent single test failure" [Undecided,In progress] - Assigned to Bernard Cafarelli (bcafarel)
17:28:14 <LouisF> http://logs.openstack.org/62/409162/1/check/gate-tempest-dsvm-networking-sfc-multinode-ubuntu-xenial-nv/41b4466/
17:28:37 <LouisF> bcafarel: thoughts on what the problem is?
17:29:29 <bcafarel> Looks like a race condition, where a flow classifier from a previous test still is in database, when we want to create a new one
17:29:41 <bcafarel> tempest tests use a new port each time, but the parameters are the same
17:30:19 <bcafarel> I followed up on igordcard email, as this is a problem he got recently:
17:30:38 <bcafarel> flow classifiers with same parameters, but different logical ports
17:30:57 <bcafarel> you can create them, but they conflict when adding it to the port chain
17:31:12 <LouisF> bcafarel: that is valid functionality to block that case
17:31:27 <bcafarel> http://lists.openstack.org/pipermail/openstack-dev/2017-January/110025.html
17:32:02 <LouisF> the reason is that when re-classification is done when a packet is received from the egress port of a  SF
17:32:41 <LouisF> the N-tuple is the only way to distinguish the chain the packet is a part of
17:33:11 <LouisF> so ... the classifier N-tuples must be different
17:33:57 <bcafarel> ah ok :/
17:34:03 <LouisF> and hence the case Igor refers to where 2 classifers with same N-tuples but different logical source ports is blocked
17:34:16 <igordcard> LouisF: so effectively it's 1 classification per port chain
17:34:40 <igordcard> LouisF: no, 1 port chain per classification
17:34:51 <LouisF> igordcard: yes
17:35:11 <igordcard> tests on lines 602 and 1079 fail if the full conflict method is called: https://review.openstack.org/#/c/419212/1/networking_sfc/tests/unit/db/test_sfc_db.py
17:35:55 <igordcard> so this is another limitation of using the sfc-proxy in networking-sfc
17:36:15 <igordcard> which means that for the sfc graphs patch it's OK since the graphs require full encapsulation
17:36:31 <LouisF> igordcard: agree - using nsh-aware SF will remove this problem
17:36:58 <LouisF> then there is no need for proxy devices
17:37:03 <igordcard> LouisF: thanks for clarifying why the constraint exists
17:37:12 <LouisF> igordcard: np
17:37:26 <igordcard> great
17:37:27 <bcafarel> thanks, I will look for another solution for the tempest failure then
17:38:00 <LouisF> bcafarel: is there a way to enure the db is flushed after a test?
17:38:06 <LouisF> ensure
17:38:54 <bcafarel> I do not know yet, but I will look into it, as this would fix the problem indeed
17:39:22 <LouisF> bcafarel: ok thanks
17:40:12 <LouisF> bcafarel: maybe post a question to openstack-infra for help
17:41:42 <LouisF> bcafarel: thanks
17:41:47 <bcafarel> LouisF: np, I will also check with tempest folks
17:42:16 <LouisF> #topic ocata work
17:42:56 <LouisF> symmetric chain patch https://review.openstack.org/#/c/410482/
17:43:06 <LouisF> all  - please review
17:43:42 <LouisF> sfc graph patch https://review.openstack.org/#/c/388802/
17:43:49 <LouisF> all - please review
17:44:44 <LouisF> doonhammer: on ovn integration - i see you posted a patch to ovs
17:44:47 <igordcard> LouisF: yep that is not expected to have any further changes except its dependency
17:45:06 <LouisF> igordcard: thanks
17:45:16 <igordcard> LouisF: sfc graph api patch will not depend on the nsh patch anymore, but rather on a different patch I will submit, which simply enables MPLS encap without proxying
17:45:23 <doonhammer> LouisF: Yes trying it in
17:45:29 <doonhammer> or at least reviewed
17:46:20 <LouisF> doonhammer: for the sfc to ovn driver - have you started work on that?
17:47:07 <doonhammer> LouisF: I have a version but it is pretty old
17:47:43 <LouisF> doonhammer: that is on github?
17:47:56 <doonhammer> LouisF: yes
17:48:28 <LouisF> do you need help on that?
17:48:46 <doonhammer> LouisF: Would appreciate it
17:49:24 <LouisF> all - volunteers on sfc/ovn driver welcome
17:49:43 <igordcard> just 1 thing, I've also submitted a graph patch including db and plugin logic/code, I'm not expecting major changes to it so I'd say it's ready for an initial review: https://review.openstack.org/#/c/419212/
17:50:28 <LouisF> igordcard: thanks - all please review
17:51:14 <LouisF> #topic other items
17:51:31 <LouisF> any other items to discuss?
17:51:48 <igordcard> LouisF: is the team going to the PTG?
17:52:50 <LouisF> igordcard: i don't think i will be there - not sure about cathy
17:53:28 <LouisF> this is in atlanta in feb https://www.openstack.org/ptg/
17:54:33 <LouisF> ok anything else?
17:55:20 <LouisF> ok thanks all
17:55:22 <LouisF> bye
17:55:25 <igordcard> I will be there, if anyone from the team can come, great
17:55:28 <igordcard> bye
17:55:45 <LouisF> #endmeeting