08:29:53 #startmeeting Yardstick work meeting 08:29:53 Meeting started Mon Aug 8 08:29:53 2016 UTC. The chair is kubi. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:29:53 Useful Commands: #action #agreed #help #info #idea #link #topic. 08:29:53 The meeting name has been set to 'yardstick_work_meeting' 08:30:15 #topic call role 08:30:21 #info Mingjiang 08:30:28 #info Jing 08:30:50 #info Kanglin 08:32:09 #topic CI and release 08:33:04 all, please help to keep the Yardstick Job success. 08:33:21 https://build.opnfv.org/ci/view/yardstick/ 08:33:26 #link https://build.opnfv.org/ci/view/yardstick/ 08:33:48 #info zhihui 08:34:30 #info LiHuan 08:34:31 Mingjiang: could you help to monitor the compass job 08:34:46 Jing: fuel Job? 08:35:07 #link https://build.opnfv.org/ci/view/yardstick/ 08:35:09 sure 08:35:46 kubi: yes 08:35:48 zhihui: could you help to keep the yardstick job which is triggered at Zte-pod 08:35:55 I wil keep an eye on fuel job 08:36:15 zhihui: I mean monitor the status of Yardstick job 08:36:39 Mingjiang: thanks 08:36:41 ok, I will do it 08:36:43 Jing: thanks 08:37:09 #info Mingjiang for compass, Jing for Fuel , zhihui for zte-pod 08:37:19 zhihui: thanks 08:37:28 you are welcome 08:37:37 #topic DPDK testing 08:38:45 I met 08:39:01 zhihui: I merged the patch about building image 08:39:12 I met some problem. one is vm booted on sam blade. 08:40:22 zhihui: anti-affinity? 08:40:52 Another is sla of latency. The latency is not acceptable. 08:41:01 zhihui: i tried ping in my environrment, it is ok. two VMs are on differnet hosts 08:41:37 Yes, I do write anti-affinity. but sometime they are still installed on same blade. 08:42:11 how many VMs in total do you installed? just 2? 08:42:22 About latency's sla, please see https://gerrit.opnfv.org/gerrit/#/c/17947/ to details. 08:42:30 just 2 08:43:54 zhihui: you could miss the SLA at first. and to find the reason why the latency is not acceptable 08:44:25 ok I will think about it. Can anyone help? 08:44:53 Thank you, Jing. 08:45:34 SLA is not the release criterion. 08:45:40 we just configure hugepage on compute node, there are not NUMA, cpu ping, so the performace could not satify. 08:46:38 zhihui: Ok 08:46:58 JiangYunhong from kvm project is also interested in this test case. 08:47:56 zhihui: OK. I see 08:48:01 So I think this test case just prove vm can install DPDK but the performace is not well without NUMA, CPU PING, etc. 08:48:18 zhihui: any suggestion from Yunhong? 08:49:49 I plan to ask for his help via email. 08:50:23 zhihui: one risk is the deadline is Aug 15th. 08:50:32 Yes, I see. 08:50:56 there are still some work to do and some problem to solve. 08:50:59 Only Bugfix will be allowed after Aug 15th 08:51:20 can I put it to release c 2.0? 08:51:48 zhihui: that's a idea. It depends on your progress. 08:52:06 I will do my best. 08:52:13 zhihui: thanks 08:52:30 zhihui: I know DPDK testing is a very technical test case. 08:52:49 #topic others 08:53:26 I saw there was a comment in the channel 08:53:27 pava2: we at armband woudl also want to add our pod (arm-pod1) to the CI and include yardstick results can someone give us guidance in what we have to do ? 08:53:48 pava2 want to add pod to CI and include yardstick result 08:54:06 anyone could help him? 08:55:49 OK, I will find a volunteer later 08:55:54 i think i can do it 08:56:05 Mingjiang: that' great , thanks 08:56:26 anything else? 08:57:07 about ha-yardstick meeting, still got no update? 08:57:56 Mingjiang: yes, I didn't get update from HA team, I will confirm with HA team after meeting 08:58:03 Mingjiang: thanks for reminder 08:58:43 thanks all for today 08:58:46 #endmeeting