13:00:54 <fuqiao> #startmeeting High Availability project meeting
13:00:54 <collabot`> Meeting started Wed Mar 14 13:00:54 2018 UTC.  The chair is fuqiao. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:54 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
13:00:54 <collabot`> The meeting name has been set to 'high_availability_project_meeting'
13:01:03 <fuqiao> #chair fuqiao
13:01:03 <collabot`> Current chairs: fuqiao
13:01:23 <fuqiao> please info yourself for the meeting
13:01:38 <Kanglin> #info Kanglin
13:01:38 <mardim> #info Dimitrios Markou
13:02:48 <georgk> #info Georg Kunz
13:02:50 <fuqiao> #topic HA testcase update
13:03:17 <fuqiao> #info have upload the two test cases for VM HA and hypervisor HA yesterday
13:04:44 <georgk> fuqiao: I looked at them and I am not sure if additional implementation is missing
13:04:59 <fuqiao> #link https://gerrit.opnfv.org/gerrit/#/c/53641/
13:05:27 <fuqiao> #link https://gerrit.opnfv.org/gerrit/#/c/53645/
13:05:29 <Kanglin> I have updated test cases on other openstack components a few days ago.
13:05:52 <Kanglin> https://gerrit.opnfv.org/gerrit/#/c/53147/
13:05:59 <Kanglin> https://gerrit.opnfv.org/gerrit/#/c/53159/
13:06:07 <Kanglin> https://gerrit.opnfv.org/gerrit/#/c/53229/
13:06:09 <fuqiao> #link https://gerrit.opnfv.org/gerrit/#/c/53147/
13:06:13 <Kanglin> https://gerrit.opnfv.org/gerrit/#/c/53243/
13:06:39 <fuqiao> Kanglin: would you mind help add"#link" before the link so that the meetbot can catch that info automatically?
13:07:02 <Kanglin> fuqiao: got it
13:07:03 <fuqiao> Georg: what do you mean by additional implementation?
13:07:16 <Kanglin> #link https://gerrit.opnfv.org/gerrit/#/c/53147/
13:07:25 <Kanglin> #link https://gerrit.opnfv.org/gerrit/#/c/53159/
13:07:34 <Kanglin> #link https://gerrit.opnfv.org/gerrit/#/c/53229/
13:07:43 <Kanglin> #link https://gerrit.opnfv.org/gerrit/#/c/53243/
13:08:01 <fuqiao> Kanglin: Thank you
13:08:09 <georgk> I added a few comments to the test case implementation
13:08:17 <georgk> #link https://gerrit.opnfv.org/gerrit/#/c/53641/1/tests/opnfv/test_cases/opnfv_yardstick_tc086.yaml
13:08:46 <georgk> just curious: I was wondering where  the monitor scripts are
13:08:59 <mardim> fuqiao, Are you sure that the node-panic script
13:09:04 <mardim> exists for example
13:09:06 <mardim> ?
13:09:25 <mardim> I mean is it actually a script there to execute the attack ?
13:09:45 <georgk> mardim: can I add you as reviewer? You have some experience now :-)
13:09:58 <fuqiao> I will drag Wangmeng here for the question
13:09:59 <mardim> I think I am already a reviewer
13:10:06 <mardim> just didn't review it so far
13:10:10 <mardim> I was a bit busy
13:10:17 <georgk> not for all of them :-)
13:10:17 <mardim> I will review it after the meeting
13:10:20 <georgk> no problem
13:10:23 <fuqiao> as far as I know, that script is something we are using when we are testing internally for vendor products
13:10:27 <mardim> Ah yes please add me
13:10:45 <georgk> mardim: thanks
13:11:03 <wangmeng> #info Wangmeng
13:11:09 <mardim> fuqiao, Yes but if that script doesn't exist in Yardstick repo
13:11:18 <mardim> the testcases will fail in CI
13:11:25 <mardim> OPNFV CI*
13:11:40 <mardim> georgk, no pro!
13:11:55 <fuqiao> Wangmeng: could you please track on the quesitons Georg propose?
13:12:06 <fuqiao> it is about the script we use the generate the panic
13:12:39 <fuqiao> I think we have already tested the case in our internal lab environment, right?
13:13:18 <wangmeng> yes, I have tested the case in our lab environment.
13:13:42 <fuqiao> it seems that we will have some problem when such script is not included in the yardstick repo
13:14:10 <mardim> fuqiao, exactly
13:14:26 <mardim> I had the same problem witht the ODL noha testcase
13:14:36 <mardim> the script were missing
13:14:41 <wangmeng> Yes, I think we have to upload all the script not only the yaml files.
13:14:46 <mardim> and I created patches to include them in Yardstick repo
13:15:06 <fuqiao> OK. Sorry, my fault... I only upload the yaml....
13:15:10 <mardim> like this : https://gerrit.opnfv.org/gerrit/#/c/53027/
13:15:23 <fuqiao> I will upload the script tomorrow morning
13:15:36 <mardim> check all the scripts that you are using
13:15:44 <fuqiao> OK. Got that
13:15:49 <mardim> I think you are missing some monitors scripts too
13:15:54 <mardim> :)
13:16:17 <mardim> #info ODL noHA testcase status
13:16:30 <mardim> #link https://gerrit.opnfv.org/gerrit/#/c/52159/
13:16:35 <wangmeng> OK, we will upload all the scripts.
13:16:54 <mardim> #info Yardstick guys send mail and they said that it looks ok
13:17:08 <mardim> #info They are waiting for KAnglin to review it
13:17:21 <mardim> Kanglin, *
13:17:22 <Kanglin> fuqiao: can you also update the jira ticket status in yardstick of these test cases?
13:17:33 <fuqiao> OK. sure
13:17:34 <Kanglin> mardim: ok, I'll review them
13:17:54 <fuqiao> #action fuqiao will update the jira status of these cases
13:17:56 <mardim> Kanglin, thanks
13:18:23 <fuqiao> Mardim, how about the ODL HA case?
13:18:48 <mardim> we are gonna have some problems implementing the ODL HA testcase
13:19:05 <mardim> we need to find installer which supoprts ODL clustring
13:19:18 <mardim> and has a pod.yaml in Yardstick too
13:19:41 <mardim> It will take time to implement it because we need to do an investigation first
13:19:56 <mardim> we are not gonna catch the MS6 deadline for that testcase
13:20:29 <Kanglin> so there is no deployment scenarios for ODL ha in existing installers?
13:20:43 <fuqiao> I actually is a bit confuse for this. why do we need an installer support this for our test case. what if we just upload the test case, and if any installer do not support ODL HA, it just fail the test?
13:20:43 <mardim> Kanglin, In fuel no
13:20:52 <mardim> Kanglin, I am not sure for compass
13:20:54 <mardim> need to check
13:21:14 <fuqiao> Yes, this is the same problem for VM HA and Hypervisor HA cases, no scenario
13:21:31 <fuqiao> does that mean we should not upload the test case at this time as well?
13:21:40 <mardim> fuqiao, Beacuse we will merge a testcase which doesn't run
13:21:47 <Kanglin> mardim: I'm afraid compass also no
13:22:10 <mardim> fuqiao, I do not know I though that we should test it first in our development envirronment
13:22:15 <mardim> georgk, ^
13:22:20 <mardim> what do you think >
13:22:22 <mardim> ?
13:22:32 <georgk> yeah, good question...
13:23:07 <fuqiao> OK got that. the VM HA and hypervisor HA test cases are tested in our lab. however it is really the case that we have no scenario for these two case
13:23:18 <mardim> The question here is does Yardstick release Frase with failing testcases in CI ?
13:23:18 <georgk> uploading the code doesn´t hurt for the time being
13:23:26 <mardim> Fraser*
13:23:31 <georgk> but if they will get merged by Yardstick is a big question
13:23:48 <georgk> well, the test cases just cannot be enabled
13:23:54 <georgk> that is still configurable
13:23:56 <Kanglin> mardim: such cases won't be added to CI
13:24:04 <mardim> georgk, Yes but what's the point to upload a patch for review which doesn't work ?
13:24:29 <mardim> georgk, ok I think I can make a yaml
13:24:29 <georgk> mardim: well, to make some progress, I would say
13:24:33 <fuqiao> OK. got that. so our two cases also should wait until Fraser release, when we have a scenario
13:24:34 <mardim> but this will not be tested
13:24:46 <georgk> we can push for the inclusion of Masakari in the platform because we have test cases for it
13:25:07 <mardim> georgk, I do not knkow what Masakari is
13:25:08 <georgk> not in Fraser at least
13:25:38 <georgk> mardim: just an example. Masakari is an OpenStack project for migrating VMs on compute failures
13:25:48 <mardim> georgk, oh ok
13:25:50 <mardim> thanks
13:26:03 <georgk> this is not related directly to the ODL clustering test case
13:26:33 <mardim> I can create a patch for the ODL HA testcase
13:26:41 <mardim> but they will not merge it propably
13:26:47 <mardim> are you ok with that guys ?
13:26:48 <fuqiao> OK. Then I guess the conclusion is we'd better not have the two test cases related with VM migration merged so far, right?
13:27:09 <georgk> it would just be a basis for now
13:27:18 <fuqiao> Mardim: I am OK, we have the same problem
13:27:29 <mardim> ok great!
13:28:01 <fuqiao> Got that. this is important info for me. I will check with Greg see if we can promote the progress of masakari
13:28:41 <fuqiao> #info the VM HA case, hypervisor HA case and ODL HA case should not be merged for now, since no scenario can be tested
13:30:06 <georgk> mardim: did you check Apex already for support for ODL clustering? sorry if I forgot...
13:30:20 <mardim> georgk, Apex and XCI support clustring
13:30:29 <georgk> mardim: but we would have to anyway check their future plans for Gambia
13:30:30 <mardim> but they do not have a pod.yaml in Yardstick
13:30:43 <georgk> right...
13:30:59 <mardim> georgk, I can create a pod.yaml for Apex I think or XCI
13:31:08 <mardim> but I need to investigate soem stuff
13:31:19 <mardim> I think I will go with XCI
13:31:22 <mardim> first
13:31:28 <georgk> ok, cool. let me know if you need support
13:31:30 <mardim> because my server has ubuntu and not Centos
13:31:39 <mardim> georgk, ok thanks!
13:31:56 <mardim> but XCi doesn;t have any adpater for Yardstick in releng
13:32:03 <mardim> and that will be an additional problem
13:32:36 <mardim> we have some pros and cons to take into account here
13:32:43 <georgk> I do have a CentOS system in the lab.... but I wanted to use that for an ONS demo
13:33:21 <mardim> Maybe I can ask from Dianfeng to install Centos in my server
13:33:30 <mardim> that is possible I think
13:33:45 <mardim> But I will do this after the ODL noHA testcase get merged
13:34:01 <mardim> because curremtly I need my environment
13:34:07 <mardim> currently*
13:34:32 <georgk> fine, thanks. let me know, I could also borrow you a lab as a service machine I have
13:34:42 <georgk> or you order one for yourself
13:34:45 <georgk> :-)
13:34:51 <georgk> so, there are options
13:34:56 <mardim> georgk, ok we can discuss it offline thanks!
13:35:03 <georgk> sure
13:35:33 <fuqiao> Great. Any other topics?
13:35:49 <mardim> georgk, We can discuss some things about testcases in ONS
13:35:52 <mardim> I will be there
13:35:55 <fuqiao> Seems that we can at least make sure some cases catch the MS6
13:38:24 <georgk> my IRC client crashed...
13:38:57 <fuqiao> I will be at ONS as well. We can continue the discussion for HA test cases there
13:39:04 <mardim> good
13:39:07 <georgk> oh, cool
13:39:19 <georgk> great, yes, absolutely
13:39:54 <fuqiao> OK. If no other topic, I will close the meeting.
13:39:58 <fuqiao> Have a good day!
13:40:32 <fuqiao> #endmeeting