08:30:03 <anac1> #startmeeting Yardstick work meeting 08:30:03 <collabot> Meeting started Thu Oct 29 08:30:03 2015 UTC. The chair is anac1. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:30:03 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic. 08:30:03 <collabot> The meeting name has been set to 'yardstick_work_meeting' 08:30:10 <anac1> #info Ana Cunha 08:30:52 <QiLiang1> #info QiLiang 08:31:00 <kubi1> #info gao liang 08:31:01 <YiminWang> #info Yimin Wang 08:31:24 <anac1> #topic Lmbench bug fix 08:31:57 <anac1> what is the status on this ? 08:32:00 <jingwen> #info Jingwen Hou 08:32:38 <QiLiang1> fix in kristian 's patch 08:33:06 <patrick1> #info patrick 08:33:08 <rexlee1> #info Rex 08:33:26 <krihun> I added it to my patch when I rebased it so it could pass all unit tests 08:33:29 <anac1> ok, so nothing to discuss, the patch https://gerrit.opnfv.org/gerrit/#/c/2623/ is fixing the issue? 08:33:45 <krihun> yes 08:33:56 <anac1> #info patch https://gerrit.opnfv.org/gerrit/#/c/2623/ fix the problems 08:34:00 <anac1> thanks 08:34:13 <anac1> #topic test result API 08:35:07 <jose_lausuch> #info Jose Lausuch 08:35:12 <jose_lausuch> I have comments for this section :) 08:35:24 <QiLiang1> yardstick2dashboard is merged in releng 08:35:24 <anac1> ok, go ahead 08:35:38 <jose_lausuch> for all the test projects reporting results to the DB 08:35:54 <anac1> #info yardstick2dashboard is merged in releng 08:35:55 <QiLiang1> jiust add ping for poc 08:36:03 <jose_lausuch> we have to agree on a common way to have an option to push the results or not 08:36:14 <jose_lausuch> we talked about this last week in functest 08:36:31 <jose_lausuch> by default we shouldnt push anything 08:36:35 <jose_lausuch> and if we want, we add a flag to our scripts 08:36:42 <anac1> why we shouldn't push anything? 08:36:47 <jose_lausuch> something like --push_results=true or something like that 08:36:55 <jose_lausuch> because we dont want everyone that tries opnfv to push it 08:37:08 <jose_lausuch> there will be credentials in the DB soon 08:37:18 <jose_lausuch> and if someone doesnt have access, that push function will fail 08:38:11 <anac1> jose_lausuch: if we are using a result api, the default should NOT be DONT PUSH, doesn't make sense to me... 08:38:28 <jose_lausuch> imagine, I am individual and Im testing opnfv on my private lab 08:38:33 <jose_lausuch> and then I run the tests 08:38:39 <jose_lausuch> why should I have to push? 08:38:49 <jose_lausuch> I wont have credentials, it will fail anyway 08:38:52 <jose_lausuch> what do you think? 08:39:08 <QiLiang1> ci triggerd test can push result ? 08:39:09 <jose_lausuch> (this is not 100% decided, so we can share different views) 08:39:33 <jnon> yes a flag to push or not is ok 08:39:33 <anac1> i think CI triggered should push results 08:39:34 <jose_lausuch> now it does 08:39:57 <jose_lausuch> CI connected labs is ok, they will have credentials to the target DB 08:40:11 <jose_lausuch> but Im concerned about everyone else testing opnfv 08:40:21 <jose_lausuch> this could cause a lot of pushes from different sources 08:40:22 <QiLiang1> so private test cannot use the api ? 08:40:26 <jnon> we are currently not continously pushing in ci job but im working on that 08:40:53 <jose_lausuch> we should all have credentials in the target db, thats for sure, all the opnfv community labs for sure 08:41:15 <anac1> yes, agree to the need to have credentials 08:41:32 <jose_lausuch> but in the end if I triy it on my private server, and another one does the same, and so 100 people, we can have 10000 different sources in the DB 08:41:42 <jose_lausuch> to avoid it, we will use credentials 08:41:51 <jose_lausuch> so what do you thing about the flag ? 08:41:51 <jnon> yes we need credentials tpo avoid that 08:41:54 <jose_lausuch> and by default dont push? 08:42:20 <jnon> yes 08:42:51 <anac1> so when is the credentials implemented? 08:43:01 <jose_lausuch> not sure when 08:43:04 <jose_lausuch> but we will know :) 08:43:28 <anac1> when? 08:43:41 <jose_lausuch> I dont have the details now 08:43:57 <anac1> jnon, qiliang: do we need a local dB? 08:44:15 <jnon> not for this , we can dumpt to file 08:44:35 <anac1> ok 08:44:41 <QiLiang1> agree with jnon 08:44:46 <anac1> ok 08:44:46 <jose_lausuch> that was all what I had to say, please add a flag to push results to DB as an option 08:44:47 <jose_lausuch> :) 08:45:09 <QiLiang1> ok, thanks 08:45:15 <anac1> jose_lausuch: how is the documentation for the api? 08:45:33 <jose_lausuch> that is what morgan has been working on 08:45:40 <jose_lausuch> what do you mean by "how" ? :) 08:45:53 <anac1> i mean, wiki, gerrit ? 08:46:08 <anac1> :) 08:46:15 <jose_lausuch> https://wiki.opnfv.org/collection_of_test_results 08:46:30 <jose_lausuch> guy rodrigue is working on that 08:46:37 <anac1> ok, will it always be in the wiki? 08:46:44 <jose_lausuch> for now... I think so 08:46:55 <anac1> i need a link to the documentation, that's why i ask 08:46:55 <jose_lausuch> not sure if morgan has plans to put it somewhere else, I will ask 08:46:59 <jose_lausuch> ok 08:47:04 <anac1> thans ! 08:47:10 <anac1> thanks! 08:47:36 <anac1> so, jnon, qiliang: we agree to put a flag, default dont push? 08:48:01 <QiLiang1> agree 08:48:07 <jnon> yes that is a flag to the ci scripts 08:48:38 <QiLiang1> ci will push right ? 08:48:45 <jose_lausuch> yes 08:48:45 <anac1> #agreed to add a flag to result api, default DONT push 08:48:46 <jnon> i already added a flag to set db server address to the scripts (not pushed yet) 08:49:13 <jose_lausuch> thanks :) 08:49:14 <jnon> if not set it dumps to file 08:50:09 <anac1> jose_lausuch: do you know if the installer type is a variable from all installers? 08:50:34 <patrick1> actually we use the yarsick.conf to config to sent to DB or just dump to file. so pay attention to this config should not to conflict with the flag value. 08:50:38 <jose_lausuch> thats a tricky question 08:50:45 <jose_lausuch> even worse is the user and password of the installer 08:51:03 <jose_lausuch> we know the defaults, but you can install it without defaults (different usr/pass) 08:51:07 <jose_lausuch> for the installe type 08:51:12 <jose_lausuch> I give the parameters in the jenkins job 08:51:21 <jose_lausuch> installer_type and installer_ip 08:51:25 <jnon> patrik1: yes the flag updates yardstick.conf in docker image 08:52:01 <QiLiang1> jose_lausuch i have check the code today :) 08:52:20 <patrick1> jnon: that's the point, tks. 08:52:41 <jose_lausuch> are you guys using fetch_os_credssh? 08:52:53 <jnon> yes 08:53:08 <jose_lausuch> so we have a common problem 08:53:27 <jose_lausuch> someone contacted us because he was not using root/r00tme in fuel, he changed the defaults 08:53:31 <jose_lausuch> and then that script failed 08:53:38 <jose_lausuch> so we will figure out how to do that 08:53:41 <jose_lausuch> not clear yet 08:53:58 <jnon> ok 08:54:07 <jose_lausuch> the only thing I can imagine now is to provide also together with installer_type and ip, installer_user and installer_password 08:54:18 <jnon> pod name is needed also 08:54:26 <jose_lausuch> but thats not very safe ... 08:54:33 <jose_lausuch> providing passwords like this :) 08:54:42 <jose_lausuch> ok jnon 08:54:49 <anac1> ok, moving on 08:54:50 <jose_lausuch> you mean for fetch_os_creds? 08:55:00 <anac1> #topic HA test cases 08:55:17 <jnon> no for other things 08:55:26 <anac1> patrick1: are you ok with the 5 test cases in the etherpad ? 08:55:51 <anac1> #link https://etherpad.opnfv.org/p/yardstick_ha 08:55:58 <patrick1> not yet . I gave the first, and try to finish all of them this weekend. 08:56:18 <patrick1> but I gave the five test case name and scope. 08:56:28 <anac1> ok, but that is the list to develop for r2? 08:56:35 <jnon> we need pod name to choose test suite, and in info dumped to db 08:56:50 <patrick1> yes. 08:56:59 <jose_lausuch> ok 08:57:21 <anac1> #info the list in etherpad is the R2 target 08:57:22 <patrick1> I try to finish all of them in R2, 08:57:44 <anac1> the first is almost done, where will you test? 08:57:57 <patrick1> but if some are not finished for some reason, I will delay it to C 08:58:01 <anac1> ok 08:58:19 <patrick1> Now I test it on huawei lab. 08:58:30 <anac1> and then on china mobile's lab? 08:58:54 <anac1> for ha project, i mean 08:59:04 <patrick1> I have talked to Fuqiao, I hope so 08:59:12 <anac1> ok, good 08:59:26 <patrick1> but I think maybe we can share a demo to HA team first 08:59:36 <patrick1> and talk about some details 08:59:54 <anac1> yes, when the code is ready 09:00:14 <patrick1> when the test case stable , we can deploy and run on China Mobiles lab 09:00:40 <anac1> sounds like a plan 09:00:56 <patrick1> In OPNFV Summit , HA has a session 09:01:08 <anac1> you mean breakout? 09:01:33 <patrick1> I will share the five test cases on it. And discuss with other guys from HA team. 09:02:12 <anac1> ok 09:02:25 <anac1> that's all for today, thanks everyone 09:02:28 <anac1> #endmeeting