00:33:17 #startmeeting Yardstick Work Meeting 00:33:17 Meeting started Tue Mar 14 00:33:17 2017 UTC. The chair is kubi001. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:33:17 Useful Commands: #action #agreed #help #info #idea #link #topic. 00:33:17 The meeting name has been set to 'yardstick_work_meeting' 00:33:23 #topic roll call 00:33:35 #info Jack 00:33:57 https://global.gotomeeting.com/join/156176189 00:34:34 #info Ross 00:34:49 #info Jing 00:34:50 let's wait for one minute 00:35:23 Trevor said he couldn't make it 00:36:06 rbbratta: got it, thanks 00:36:33 JackChan: JingLu would you like to join the GTM 00:36:39 https://global.gotomeeting.com/join/156176189 00:37:06 otherwise, we can talk with IRC 00:37:27 kubi001: yes, I would join right now 00:38:01 #topic action item follows up 00:38:08 http://ircbot.wl.linuxfoundation.org/meetings/opnfv-yardstick/2017/opnfv-yardstick.2017-03-07-00.32.html 00:38:24 #info AP1: rex would move the documentation before branch stable 00:38:49 #info AP1: Done https://gerrit.opnfv.org/gerrit/#/c/30013/ 00:39:39 #topic stable branch 00:40:01 #info we have created the stable branch at last Friday by Aric 00:40:38 #info any changes which you want to release with Danube would need be cherry pick to the stable branch. 00:40:42 https://gerrit.opnfv.org/gerrit/#/c/30139/ 00:40:53 This is a bugfix for tc072 00:42:35 #info Danube 1.0 is Mar. 27 th 00:42:56 #info https://wiki.opnfv.org/display/SWREL/Danube?preview=/6827418/8689182/OPNFV%20Release%20%2522Danube%2522%20r4.pdf 00:43:35 #info MS8 is Formal test execution complete 00:43:53 6 days for MS8 00:44:51 #info MS9 is JIRA issues assigned to release closed or deferred 00:45:37 https://jira.opnfv.org/projects/YARDSTICK/versions/10632 00:46:17 #info MS10 Documentation completed 00:47:20 #topic CI status 00:47:52 https://gerrit.opnfv.org/gerrit/#/c/30331/ 00:52:11 the Summit CFP closes on March 27th and we hope that several folks from your group (and you of course) decide to submit a session(s) in the following tracks: 00:52:12 • NFV Applications and Orchestration 00:52:12 • Testing, Infrastructure, and DevOps 00:52:12 • NFV Strategy and End User Stories 00:52:12 • NFV Platform Requirements 00:52:12 • Community and Upstream 00:52:12 • Futures and Research 00:55:06 SR-IOV standalone context https://gerrit.opnfv.org/gerrit/#/c/29555/ 00:58:21 I guess another meeting has been started 00:58:38 Deepak gave a walk-through of vnf_generic.py recently, he might be able to present it 00:58:52 rbbratta: got it. thanks 00:58:58 The main thing with NSB is all the yaml files for topology and VNF descriptors. 00:59:15 There are several yaml files that all refer to each other. 01:00:05 I'll check with Deepak 01:00:28 rbbratta: a quick question, do we have a available open source VNF for NSB now 01:00:54 I think vPE is there somewhere, or the samplevnf project will have them. 01:01:49 where is the vPE come from? 01:02:02 vPE_vnf, 01:02:53 nsb_setup.sh: dpdk-16.07/examples/ip_pipeline/ 01:03:39 https://gerrit.opnfv.org/gerrit/gitweb?p=yardstick.git;a=blob;f=nsb_setup.sh;hb=HEAD 01:04:25 https://gerrit.opnfv.org/gerrit/gitweb?p=yardstick.git;a=blob;f=nsb_setup.sh;hb=HEAD#l171 01:04:30 so that means vPE has been included in dpdk-16.07 package 01:04:38 yes 01:04:54 rbbratta: got it, thanks 01:04:55 so it is really basic 01:05:11 rbbratta: i see 01:05:23 you send it commands 01:05:47 maybe the SampelVNF would be a good source for reference VNF 01:06:20 so this is the real config https://gerrit.opnfv.org/gerrit/gitweb?p=yardstick.git;a=blob;f=samples/vnf_samples/nsut/vpe/vpe_config/vpe_config;hb=HEAD 01:06:37 it creates 9 pipelines to simualte vPE 01:07:01 this is all in vpe_vnf.py 01:07:26 yes, they plane to opensource all our DPDK sample VNFs in the samplevnf project 01:07:39 rbbratta: sounds good 01:08:33 so you have a topology yaml, a VNF descriptor yaml, and a traffic profile yaml, etc. 01:09:17 rbbratta: Is it possible to run the NSB with CI in E-release, 01:09:31 https://etherpad.opnfv.org/p/samplevnf 01:09:55 It depends on the context. NSB with Heat doing East/West with no external traffic generator, probably easy. 01:10:31 NSB doing on SR-IOV or baremetal requires re-imaging the machine, and maybe physical wiring topology 01:11:10 rbbratta: specific pod would be need if NSB depends on physical wiring topology 01:11:28 yes 01:12:01 but maybe if we have the dynamic POD stuff 01:12:13 NSB with Heat looks like the sample testcase in https://jira.opnfv.org/browse/YARDSTICK-579 01:13:17 There are several pieces there, one we need to provide MAC address, PCI BUS ID and kernel driver to vnf_generic so we can DPDK bind/unbind the interfaces 01:13:40 We also want to create flavors using Heat context 01:14:07 And then we need better code to create samplevnf images for glance. 01:16:12 rbbratta I see. I am also working on dpdk testcase recently. 01:16:28 rbbratta: sounds good, I'm glad to see that the enhancement of context. 01:16:56 JingLu: dpdk testcase? 01:19:03 rbbratta I was trying to do some dpdk performance test in my local, so I learn a little about DPDK. 01:20:51 rbbratta so in the NSB only trex is used as the TG? 01:21:25 We have some Ixia code somewhere 01:22:07 rbbratta so both hardware and software TGs? 01:22:10 pktgen as well 01:22:39 yes, HW and software. Ideally we integrate with VSPERF and use moongen, etc. 01:23:01 There is also PROX which is new DPDK traffic generator app. 01:23:29 https://01.org/intel-data-plane-performance-demonstrators/prox-overview 01:24:06 rbbratta I see. Thanks. 01:24:21 there are too many traffic generators, too hard to compare and figure out which give correct results for RFC2544, etc. 01:24:48 Of course I'm not the expert, have to ask Trevor. 01:25:37 OK, thanks a lot 01:25:49 I guess we need to end today's meeting 01:26:03 #endmeeting