08:06:36 <joehuang> #startmeeting multisite
08:06:36 <collabot`> Meeting started Thu Nov  3 08:06:36 2016 UTC.  The chair is joehuang. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:06:36 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
08:06:36 <collabot`> The meeting name has been set to 'multisite'
08:06:51 <joehuang> welcome Goutham, new contributor
08:06:57 <GouthamPratapa> thanks guys :)
08:07:01 <joehuang> #info rollcall
08:07:07 <joehuang> #topic rollcall
08:07:12 <joehuang> #info joehuang
08:07:14 <SAshish> #info Ashish
08:07:21 <GouthamPratapa> #info GouthamPratapa
08:07:27 <sorantis> #info dimitri
08:08:15 <joehuang> #topic summit short summary
08:08:15 <SAshish> so I have incorporated Dimtri's latest comments in the doc
08:08:19 <SAshish> sure please
08:09:57 <joehuang> #info cloud operators have lots of concerns on multi-region cloud which part of them could be addressed by Kingbird
08:10:02 <joehuang> #link https://etherpad.openstack.org/p/BCN-ops-control-plane-design
08:11:11 <joehuang> It will be good to make kingbird more visible in OpenStack operators and developers
08:11:27 <joehuang> these concerns are arising.
08:12:51 <SAshish> is the awareness for kingbird better now?
08:13:04 <SAshish> I mean when compared to earlier sumit
08:13:06 <SAshish> summit*
08:13:39 <sorantis> No it’s not. Our presentation was alas rejected, so we couldn’t show off anything
08:13:52 <joehuang> During the summit, also talked to David about our plan, he is not very happy if we skip some milestones and define our own milestones
08:14:24 <joehuang> we may prompt it in the mail-list
08:14:26 <SAshish> okay, which milestones we skipped?
08:14:39 <sorantis> we can have those milestones for multisite
08:14:47 <sorantis> but run kingbird our way
08:15:07 <joehuang> installer intergation etc
08:15:12 <joehuang> #link https://wiki.opnfv.org/display/multisite/Multisite+Release+D+Planning
08:15:16 <sorantis> there’s also a thread on ML regarding other use cases that touch on multisite
08:15:48 <joehuang> MS2 ~ MS5
08:16:23 <joehuang> to Dimitri, yes, some other projects also want multi-site CI/test bed infrastructure
08:17:02 <joehuang> Dimitri proposed "Automated multi-site deployment via Jenkins"
08:17:10 <joehuang> we can talk about this topic
08:17:49 <sorantis> Fatih has proposed the following alternative to setting up MR environment
08:17:49 <sorantis> https://docs.google.com/presentation/d/1BXa14pEYacxVx6zNPD5bDBl97XVORRzA5jAyV8leR08/edit?usp=sharing
08:18:16 <joehuang> and OPNFV is also discussing to integrate Ocata or Newton, so I think the plan may be changed after the decision
08:18:35 <sorantis> as you can see, there’s no work needed in any installer. It’s jenkins who will automatically deploy 2 regions, and other jobs to reconfigure them for centralized keystone
08:19:55 <sorantis> I’ve added only one step, that configures centralized keystone
08:20:02 <SAshish> who will perfrom "configure centralized keystone"?
08:20:03 <joehuang> Whether Fuel support post-reconfiguration
08:20:12 <sorantis> Fatih also said we can have one more node, for a real world setup
08:20:15 <joehuang> and service restart automaticaly
08:20:28 <sorantis> don’t think so. It should be scripted
08:20:31 <SAshish> ansible?? not sure
08:20:36 <sorantis> I’ve started looking into it
08:21:03 <sorantis> Will not use federation though
08:21:15 <SAshish> okay, the script you started already right
08:21:18 <joehuang> But fuel will be changed to using container, will the work wasted to setup multi-site environment based on current implementation
08:21:24 <sorantis> I think for our case it’s best to have just one keystone
08:21:26 <SAshish> I remember you showed me a script
08:21:45 <sorantis> joe, that’s why we won’t do it in fuel
08:21:48 <joehuang> to Dimitri, +1 not using federation now
08:22:09 <sorantis> we’ll make a script that will work with an already installed system
08:22:30 <sorantis> fuel or not, the /etc/… config files should be there anyway
08:22:30 <SAshish> then we need two nodes for two fuel installations
08:22:37 <sorantis> and we will get them
08:22:48 <sorantis> refer to the diagram
08:22:59 <sorantis> it considers two nodes, with two parallel installations
08:23:14 <joehuang> you mean using SSH to directly change the etc/ files
08:23:33 <SAshish> yeah. I meant one more node we should ask to pharos
08:24:35 <joehuang> it will be quite difficult if the openstack installed with HA mode, non-HA mode fuel installation will be better to change its configuration
08:24:38 <sorantis> joe, either a plain hardcore scripting or something else
08:24:46 <sorantis> I don’t have strong preferences tbh
08:24:53 <sorantis> +1
08:25:04 <sorantis> absolutely totally 100% agree
08:25:23 <sorantis> the HA mode caused only more troubles while debugging
08:25:31 <joehuang> #info non-HA fuel installation for two sites
08:25:44 <SAshish> the port issue
08:25:54 <SAshish> for running tempest from container
08:26:30 <joehuang> Good, I think in this way the effort is more acceptable
08:27:09 <sorantis> yes, and it reuses existing functions
08:27:33 <sorantis> so what’s missing from the diagram is the blue-box job
08:27:55 <sorantis> also jenkins should publish some information after the two parallel jobs complete
08:28:04 <sorantis> such as IP addresses
08:28:28 <joehuang> Multisite parent job is not ready too
08:28:30 <sorantis> we’re also discussing usage of DNS
08:29:15 <sorantis> to avoid hassle with ip addresses. We’ll see if it’s possible in the intel Lab
08:30:25 <sorantis> any objections? thoughts?
08:30:56 <joehuang> no objections.
08:30:57 <sorantis> I think Fatih came up with it’s a good way for multisite deployment automation.
08:31:08 <SAshish> fine
08:31:27 <joehuang> Yes, we need help and colloboration from other teams
08:31:52 <sorantis> that we won’t get :)
08:33:07 <joehuang> Ok, go ahead
08:34:39 <GouthamPratapa> hi
08:35:08 <joehuang> Hi, Please
08:35:11 <GouthamPratapa> i need a small clarification on my gerrit comit
08:35:25 <GouthamPratapa> regardingh micro and mini versions
08:35:42 <GouthamPratapa> are u referring to this "https://specs.openstack.org/openstack/nova-specs/specs/kilo/implemented/api-microversions.html"
08:36:33 <joehuang> yes
08:36:59 <joehuang> Nova already support microversion
08:37:35 <GouthamPratapa> ok
08:38:32 <joehuang> For the day-light-saving time was ended, so we'll start the meeting at the time as today, how do you think about this
08:38:46 <GouthamPratapa> so the idea will be like using the latest version of nova
08:39:14 <GouthamPratapa> and making it compatable to for the respective micro versions right?
08:39:52 <GouthamPratapa> by which i mean based upon the request the respective micro version
08:40:02 <GouthamPratapa> must be used.
08:40:43 <SAshish> I think: novaclient object should be created with the major version, then api specific micro versions can be used
08:41:37 <joehuang> Client in Kingbird should specify the version it supports, and then query nova to see what's version range it support, and match the maximum version for client and Nova both support. otherwise fallback to min version for both
08:43:02 <joehuang> version negotiation and decide the highest matched version to use
08:44:45 <GouthamPratapa> ok fine
08:44:49 <GouthamPratapa> thanks
08:46:12 <joehuang> For D release planning of multisite, we may keep it aligned with OPNFV template, but for some milestones which are not applicable to multisite, we just set finished when needed.
08:46:31 <joehuang> David don't like customization.
08:47:09 <SAshish> yes
08:47:48 <sorantis> fine
08:48:07 <joehuang> good, thanks
08:48:19 <SAshish> Guys, then is the document okay ?
08:48:31 <SAshish> Please give confirmation
08:48:36 <sorantis> this however doesn’t apply to kingbird development
08:49:35 <joehuang> The document is good enough for code implementation, and update it after patches merged to reflect the final implementation
08:50:13 <joehuang> yes, multisite plan doesn't apply to Kingbird development
08:50:33 <sorantis> +1
08:50:40 <joehuang> Kingbird can keep the milstone aligned with Ocata release
08:50:49 <joehuang> it's one of OpenStack project
08:51:58 <SAshish> cool
08:52:24 <joehuang> How about next topic "Revisit Geographical Redundancy"?
08:52:38 <joehuang> or discuss it next week?
08:52:47 <sorantis> yes please
08:52:55 <sorantis> there are many things on the table already
08:52:56 <SAshish> Starving to death :)
08:53:22 <joehuang> Ashish need to have launch, so let's discuss it next time
08:53:52 <SAshish> thanks
08:54:17 <sorantis> thanks for the meeting
08:54:19 <joehuang> thank you for attending the meeting, see you next time
08:54:24 <sorantis> bye
08:54:24 <joehuang> #endmeeting