08:14:52 <joehuang> #startmeeting multisite
08:14:52 <collabot`> Meeting started Thu Nov 10 08:14:52 2016 UTC.  The chair is joehuang. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:14:52 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
08:14:52 <collabot`> The meeting name has been set to 'multisite'
08:15:01 <sorantis> hi
08:15:09 <joehuang> hi,
08:15:14 <joehuang> #topic rollcall
08:15:14 <SAshish> Hi Dimitri
08:15:18 <joehuang> #info joehuang
08:15:20 <Goutham> hi
08:15:38 <Goutham> #info goutham
08:15:47 <sorantis> #info dimitri
08:15:49 <SAshish> #info Ashish
08:16:05 <joehuang> #topic  CI Job unstable
08:16:25 <joehuang> hello, the CI job in intel lab is unstable recently, about 20 days
08:16:36 <SAshish> keystone is not working
08:16:49 <joehuang> #link https://build.opnfv.org/ci/view/multisite/builds
08:17:39 <SAshish> I have checked, there is not much memery left on that device
08:17:52 <SAshish> have checked on jump host
08:18:23 <joehuang> Is there any memory leak?
08:19:23 <sorantis> could be anything
08:19:36 <SAshish> libvirt process takes lot of memory
08:20:03 <joehuang> MOS should be production grade
08:21:11 <sorantis> short term - we can restart and see if the proble is gone
08:21:13 <joehuang> so for the function test, it may be better to use one time environment
08:21:40 <sorantis> I’ve already started working on centralization of Keystone
08:21:48 <sorantis> first patch https://gerrit.opnfv.org/gerrit/#/c/23975/
08:21:53 <joehuang> for short term, restart services works
08:22:05 <sorantis> please spend some time reviewing
08:22:30 <joehuang> sure, will do that
08:22:32 <sorantis> this work will be slow however due to other responsibilities
08:22:41 <joehuang> understand
08:23:06 <SAshish> yeah, will do
08:23:32 <joehuang> I also think we can add tempest check and gate job in OpenStack pipeline
08:23:37 <joehuang> first in single region
08:23:53 <joehuang> so that function can be executed in live environment
08:23:56 <SAshish> so every time we set MR before functest run for Multistie?
08:24:24 <SAshish> assuming there are multicontroller set up working on a POD?
08:24:33 <joehuang> in OpenStack, there is no MR test bed, but multi-node is there.
08:25:04 <joehuang> I am also consulting OpenStack infra how to setup MR devstack based gate/check environment
08:25:46 <joehuang> to Ashish, you mean installer supported MR in OPNFV testbed
08:26:36 <SAshish> right now as there is no MR installer we plan to do it from Dimtri's script right
08:27:09 <sorantis> there will be no mr installer
08:27:20 <joehuang> installer only need to install single region from last meeting slides
08:27:21 <sorantis> at least not in the scope of work I’m doing
08:27:26 <sorantis> yes
08:27:49 <sorantis> there will be two jenkins jobs deploying two VIMs separately
08:28:08 <sorantis> then the third job will make sure one keystone is used
08:28:34 <joehuang> yes. Dimitri, could you publish this in multisite wiki?
08:28:35 <SAshish> and the third job is your script?
08:29:03 <sorantis> yes
08:29:18 <sorantis> several scripts
08:29:18 <SAshish> fine
08:29:31 <SAshish> yeah yeah, I meant this will be used there. Got it
08:31:38 <joehuang> ok. continue the MR auto deployment environment in OPNFV
08:31:59 <sorantis> what’s our decision regarding the failing jobs?
08:32:17 <SAshish> I dnt think restart the services will work
08:32:33 <SAshish> am not sure which service we have to restart
08:33:00 <SAshish> there is memory issue on jumpshot itself
08:35:36 <SAshish> jumphost*
08:36:08 <sorantis> it can be restarted right?
08:37:11 <SAshish> the machine itself?
08:38:07 <joehuang> sorry, my network was broken
08:38:08 <sorantis> how does jumphost affect the jenkins job? it doesn’t play any role, does it?
08:39:01 <SAshish> If these fuel nodes are created on jumphost?
08:40:12 <SAshish> there is libvirt running on jumphost
08:40:16 <joehuang> I forget the topology
08:40:18 <sorantis> they were created once, initially
08:40:19 <SAshish> Joe do you have access to lab right now?
08:40:30 <SAshish> just wanted to verify virsh list
08:40:33 <SAshish> on jumphost
08:40:43 <joehuang> my network was just broken, will try
08:44:05 <joehuang> sorry, I can't establish VPN connection now
08:44:51 <SAshish> okay, I will check and tell you
08:44:53 <SAshish> later
08:45:19 <SAshish> if we are the only ones who are using this POD then we can restart
08:45:19 <joehuang> thank you
08:45:26 <SAshish> or else we cant rigjt
08:45:27 <SAshish> right
08:46:00 <joehuang> need to send msg in the mail-list, to see if there are other ones are using the environment, especially for jumphost
08:47:51 <SAshish> okay
08:47:54 <sorantis> I have to switch to another meeting now unfortunately
08:47:58 <joehuang> If the environment can't be rescued by restarting, we may leave it like this for a while
08:48:01 <joehuang> ok
08:48:05 <joehuang> to Dimitri
08:48:26 <joehuang> and wait for the MR deployement scripts
08:48:28 <sorantis> I’ll continue with the centralized jobs, hopefully will get some results soon
08:48:41 <SAshish> fine
08:48:56 <joehuang> enviroment maintainence needs too much time
08:49:03 <SAshish> arey pharos guys responsible for maintainence?
08:49:15 <SAshish> or it is the team to which they allocate?
08:49:32 <joehuang> there is a team, pharos
08:49:47 <sorantis> ok, thanks for the meeting. bye
08:49:55 <SAshish> Good day Dimitri
08:49:56 <joehuang> Thank you dimitri, bye
08:50:20 <SAshish> so are they responsible for maintainence?
08:50:21 <SAshish> or WE?
08:50:31 <joehuang> To Ashish, is there any challenge for the feature implementation
08:50:55 <SAshish> no nothing as of now
08:51:06 <joehuang> the node allocated to multi-site should be our responsible to maintain it, but for the jump host, should be pharos
08:51:20 <SAshish> okays
08:51:30 <SAshish> so can we raise a jira on pharos?
08:51:42 <joehuang> of course we can if we find some issue
08:51:51 <Goutham> hi joe
08:51:58 <joehuang> Hi, Goutham
08:52:05 <Goutham> http://hastebin.com/azikulawiy.hs
08:52:14 <Goutham> this is the kingbird cli
08:52:56 <joehuang> very good!
08:53:06 <Goutham> i will send this over document we will discuss this over mail
08:53:17 <SAshish> yeah, better
08:53:25 <joehuang> you can use source openrc admin admin to specify user, role in environment
08:53:33 <joehuang> Great job
08:53:48 <joehuang> let's discuss in mail-list
08:54:11 <joehuang> only source openrc is not enough
08:54:56 <SAshish> he might be doing that in openrc file it self
08:55:04 <SAshish> anyways the structure looks fine
08:55:10 <SAshish> Good job Goutham
08:55:11 <joehuang> yes
08:55:44 <joehuang> openstack client has some guideline for command line
08:56:32 <joehuang> #link http://docs.openstack.org/developer/python-openstackclient/
08:57:29 <joehuang> It's good to align with openstack client. They have lots of good practice
08:58:48 <Goutham> yea i will see this
08:58:51 <joehuang> #link https://www.youtube.com/watch?v=EMy9IsRHY-o
08:59:29 <joehuang> there is one video in Barcelona summit about openstackclient, much better compaired to single service client
09:00:21 <joehuang> ok, time is up now, let's continue development, and have meeting next week
09:01:30 <joehuang> #endmeeting