17:02:12 #startmeeting service_chaining 17:02:12 Meeting started Thu Oct 6 17:02:12 2016 UTC and is due to finish in 60 minutes. The chair is LouisF. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:02:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:02:16 The meeting name has been set to 'service_chaining' 17:02:21 hi all 17:02:23 hi 17:02:29 hi 17:02:32 hi 17:03:16 here is agenda for today https://wiki.openstack.org/wiki/Meetings/ServiceFunctionChainingMeeting#Agenda_for_the_Networking-SFC_Meeting_.2810.2F6.2F2016.29 17:03:27 hello 17:03:48 anything to add? 17:04:26 hi 17:04:39 hi all 17:05:11 #topic progress status 17:05:16 greetings 17:06:44 Replace OvsSfcAgent by L2 Agent Extension https://review.openstack.org/#/c/351789/ 17:07:16 there are some +1 reviews now, nothing noted in the code itself after last fix (thanks igordcard), 17:07:36 what is progress on testing this? 17:08:00 LouisF: I got the flows I was expecting, with the last patchset 17:08:09 i know that some testing has been done 17:08:42 I'll fix the functional test which in fact does nothing at the moment, and I want to add some unit tests on the extension itself (most are on the ovs driver), but nothing blocking 17:09:22 i would like to see some more testing done before final approval 17:10:18 i will get more testing done 17:10:50 next item - Move to using OpenStack Client (OSC) https://review.openstack.org/#/c/344920/ 17:11:13 mohankumar: what is status of testing this? 17:11:47 LouisF, Got distracted with SFC-heat support and other work items . this patch needd some more testing , will open to review soon with UT 17:12:20 mohankumar: ok thanks for your efforts on this 17:13:09 i can also start testing on osc for sfc 17:13:23 LouisF , okay 17:13:39 mohankumar: thnaks 17:13:53 next item - Symmetric chain API https://review.openstack.org/#/c/308274/ 17:14:42 pavel had started this work - I will be updating with a new patch 17:15:25 i will also post a new patch on the changes for the ovs driver/agent 17:15:51 LouisF , great ! 17:16:12 next - Networking-sfc / OVN driver spec https://review.openstack.org/#/c/333172/ 17:16:16 LouisF: is that based on the new agent? 17:16:37 igordcard: it will be once that is merged 17:17:26 all - please review this 17:17:53 doonhammer: there is a presentation at the summit on this - right? 17:17:54 LouisF: I think one of next steps is to get the draft OVN spec done and to the OVN/OVS team 17:18:11 LouisF: If we get it put together :-) 17:18:30 doonhammer: yes 17:18:36 LouisF: Also I think we might be able to have a f2f in Barcelona 17:19:03 doonhammer: yes that will be very useful 17:19:21 with the OVS/OVN team so iif we have some details to discuss by then it would be good 17:20:14 doonhammer: we can go over the proposed changes to ovn-nbctl and the SB logical flows with them 17:20:33 LouisF: yup that would be ideal 17:21:44 * sridhar_ram has couple of items to discuss on tacker+sfc, if time permits 17:21:57 sridhar_ram: ok 17:22:28 LouisF: thanks, can i go now ? 17:22:42 sridhar_ram: go ahead 17:23:08 LouisF: 1st.. it will be good if tacker & sfc team can meet for a F2F in Barcelona.. 17:23:24 sridhar_ram: definitely 17:24:05 LouisF: i'm not travelling .. but sripriya is the acting PTL.. we have identified a design summit slot for this 17:24:30 sridhar_ram: let me know the details 17:24:59 LouisF: here is the agenda .. https://etherpad.openstack.org/p/tacker-ocata-summit 17:25:32 LouisF: one possible slot is Friday AM (already penciled in).. but I'll ask sripriya to coordinate with you and the sfc team 17:25:39 sripriya: ^^^ 17:26:01 sridhar_ram: catching up 17:26:07 LouisF: next, the issue related to network_src_port_id restriction in the classifier.. 17:26:11 sridhar_ram: i will contact her 17:26:23 LouisF: http://lists.openstack.org/pipermail/openstack-dev/2016-October/105153.html 17:26:43 LouisF: I've described the downside of this here ^^^ 17:26:57 LouisF: anyway we can address and remove this restriction ? 17:28:50 sridhar_ram: there are several issues - 1. if the logical source port is not specified the ovs classifier rules must be installed on all compute nodes 17:30:43 there is also an issue with matches on traffic entering via a router 17:31:23 and this causes a chain path loop :( 17:31:31 LouisF: I see 17:32:03 LouisF: I'd like to request for some ideas.. perhaps out of the box to atleast reduce this restriction.. 17:32:12 LouisF: can this classifier be applied only when a vif matching a specific "source critieria" .. perhaps we can use port tagging ? 17:32:40 sridhar_ram: we are open to explore ideas on this 17:32:45 LouisF: btw, traffic entering via a router is a huge use-case that we need to address 17:32:59 sridhar_ram: agree 17:33:08 LouisF: okay.. didn't mean to hold your meeting, just want to bring this up.. 17:33:46 LouisF: perhaps this can be discussed further in the F2F summit meeting 17:33:55 sridhar_ram: noted - we are actively looking at options as we are aware of the issue 17:34:06 LouisF: thanks 17:34:16 yes we should definitely discuss 17:34:48 sridhar_ram: any other issues from your side? 17:34:58 LouisF: no, that is it for now :) 17:35:53 sridhar_ram: ok 17:36:29 sridhar_ram: btw anil did a demo at the odl summit - right? 17:36:52 LouisF: yes, he did .. neutron-sfc + ODL-sfc .. i heard it went really well 17:37:08 sridhar_ram: is there video on that? 17:37:26 LouisF: for our FFG talk we are preparing a demo of Tacker --> n-sfc--> ODL-SFC ..!! 17:37:52 LouisF: building on top of Anil's 17:38:26 sridhar_ram: great - we should discuss the talk later 17:38:35 LouisF: sounds good 17:40:50 igordcard: i missed the nsh work https://review.openstack.org/#/c/373465/ 17:41:05 what is the status of that? 17:41:08 LouisF: working right now 17:41:11 it's almost done 17:41:26 it's given me more trouble than I antecipated, but it's almost done now 17:41:44 igordcard: great - looking foward to the update 17:41:51 I'm fixing the classification flows when there no sfc-proxy is needed 17:41:59 igordcard: ok 17:42:00 s/there// 17:42:27 I plan to start asap the sfc graphs work, which will directly build on top of this one 17:42:34 igordcard: will you be in barcelona? 17:42:59 I will rebase the patches to bernard's ovs agent later, likely after the summit 17:43:04 yes 17:43:26 igordcard: we should have a design session 17:43:42 LouisF: yes! 17:44:11 we can discuss future work 17:45:51 I plan on doing some work on the common classifier for the next cycle as well 17:46:09 igordcard: ok 17:46:20 when is networking-sfc expecting to shift to a neutron classifier? when there's one 17:46:28 igordcard: yes 17:47:18 igordcard: regaing this bug: https://bugs.launchpad.net/networking-sfc/+bug/1630657 17:47:19 Launchpad bug 1630657 in networking-sfc "project_id not being leveraged by the OVS driver" [Undecided,New] 17:47:40 can you attach steps to reproduce 17:48:26 LouisF: yes, but it's how cathy says (just read) 17:48:43 igordcard: likewise https://bugs.launchpad.net/networking-sfc/+bug/1630503 17:48:44 Launchpad bug 1630503 in networking-sfc "DB transaction issues since switching to Newton" [Undecided,New] 17:49:18 igordcard: ok 17:49:55 okay 17:50:29 ok this is all i have 17:50:42 anything else to discuss? 17:51:34 ok thanks all 17:51:43 'til next week 17:51:55 bye all 17:52:03 bye 17:52:04 bye 17:52:34 #endmeeting