15:59:34 <frankbrockners> #startmeeting BGS daily release readiness status meeting
15:59:34 <collabot> Meeting started Fri Apr 24 15:59:34 2015 UTC.  The chair is frankbrockners. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:59:34 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:59:34 <collabot> The meeting name has been set to 'bgs_daily_release_readiness_status_meeting'
16:00:02 <frankbrockners> #info Frank Brockners
16:00:17 <pbandzi> #info pbandzi
16:01:04 <[1]JonasB> #info Jonas Bjurel
16:01:49 <HKirksey> #info Heather Kirksey
16:03:01 * frankbrockners am pinging a couple of folks
16:03:35 <bryan_att> #info Bryan Sullivan
16:03:39 <trozet> #info Tim Rozet
16:04:26 <frankbrockners> ok - looks like we have a quorum
16:05:23 <frankbrockners> #topic Updates on BGS status - updating https://wiki.opnfv.org/releases/arno/releasereadiness
16:05:42 <frankbrockners> Jonas, Tim could you give us a brief update?
16:05:55 <trozet> I can go
16:06:08 <frankbrockners> go ahead Tim
16:06:42 <trozet> #info workaround added to deploy so that vbox VM is now working on jumphost.  Still not sure what the root cause is, but we suspect it is UCS adding a VLAN tag egress to the host
16:07:18 <trozet> #info VLAN header with VID 0 from UCS is due to no VLAN being configured from UCS-> Fabric Interconnect, Frank to talk about changes to the LF setup
16:07:39 <trozet> #info IPMI now working after LF changes to the firewall, and verified Foreman can control the nodes
16:07:46 <[1]JonasB> vbox?
16:08:00 <trozet> #info testing out building a node right now, will try to deploy nodes today
16:08:19 <trozet> vbox == virtualbox
16:08:52 <[1]JonasB> trozet: Yes I know, thought we used KVM and libvirt on the jumphost?
16:09:10 <trozet> thats the end of my status
16:09:38 <trozet> libvirt doesnt have great support with Vagrant.  It's a 3rd party plugin and I had issues trying to use it with CentOS7
16:10:12 <trozet> virtual box seems to work just fine, and its under GPL license
16:10:17 <[1]JonasB> trozet: Does that pe�revent us to install fuel on KVM/Libvirt?
16:11:09 <trozet> there are some issues running vbox kernel modules and kvm kernel mods at the same time
16:11:16 <trozet> in the cleanup script we should each remove the modules
16:11:23 <trozet> from the kernel
16:11:37 <trozet> (once we do multi installers on the same jumphost)
16:11:52 <[1]JonasB> 3trozet: Cause we're depending on Libvirt for our fuel installation orchestration:-(
16:12:02 <trozet> yeah thats fine
16:12:02 <frankbrockners> let's #info that
16:12:43 <trozet> #info each installer should cleanup the kernel mods it is using after a job, so that hypervisors dont conflict.  Example: kvm modules colliding with vboxdrv
16:12:53 <frankbrockners> thanks tim
16:13:06 <frankbrockners> Jonas - could you give a brief status update?
16:13:34 <[1]JonasB> Just to make sure we're talking about the same thing, were talking about the jump host right?
16:13:58 <frankbrockners> [1]JonasB: yes
16:14:13 <[1]JonasB> #info documentation is 100% ready for review in git
16:14:52 <[1]JonasB> #info Were finalizing the HW adapter for Eri lab and hope to autodeploy on Mon-Tue
16:15:17 <[1]JonasB> #info Started to work on HW adapter for LF-lab
16:16:05 <[1]JonasB> #info Centos support in, a small issue with HA for centos, troubleshooting ongoing.
16:16:52 <[1]JonasB> #info: Jose is working with Tempest tests for Fuel), should be 100% reusable for Foreman
16:16:58 <[1]JonasB> Thats all from me
16:17:06 <frankbrockners> thanks jonas
16:17:32 <frankbrockners> quick discussion on testing - given that there was some debate on the alias
16:17:52 <frankbrockners> #topic Testing for BGS
16:19:00 <frankbrockners> #info Current understanding is: Functest will supply a set of tests (Rally/Tempest and Robot). Functest will also take care of installing Robot and Tempest/Rally on the jumphosts of POD1 and POD2.
16:19:40 <fdegir> rally & robot have been installed on jhs
16:19:44 <frankbrockners> #info BGS is not immediately involved with testing - only as that we have to make sure their tests exec successfully
16:19:58 <frankbrockners> is this common understanding?
16:20:23 <frankbrockners> #info rally & robot have been installed on jhs (per fdegir)
16:20:52 <frankbrockners> can someone confirm or correct the above understanding?
16:20:56 <[1]JonasB> Formally yes, but I think we need to engage, we will fail if there is no functest in place
16:22:15 <frankbrockners> [1]JonasB: "engage" means engaging with the FuncTest team?
16:22:26 <[1]JonasB> Yes
16:23:05 <frankbrockners> 100% agreed - will try to have Morgan join all of our synch meetings moving forward.
16:23:46 <frankbrockners> ok - let's move to the next topic: LF HW config
16:23:59 <frankbrockners> #topic Updates to LF HW config
16:25:03 <trozet> #info One thing related to LF.  The admin network will need to be routable to the public network so that the PXE nodes can get out to the internet.
16:25:05 <frankbrockners> #info For proper operation, there is a need to have all traffic tagged *within* the fabric (you won't notice on the vNIC). Right now, we don't do this in all cases.
16:26:17 <frankbrockners> #info E.g. ...9.64/26 isn't mapped to a VLAN in the fabric.
16:26:30 <[1]JonasB> Dont understand, that is normal practice! There is normaly never untagged trafic in a fabric
16:27:10 <frankbrockners> #info In addition, there is a need to fully isolate PODs (so that we don't confuse each other) - that would mean subnetting the current .8.0/24 and .9.0/24 further.
16:27:45 <frankbrockners> [1]JonasB: Agreed. We're trying to get to a clean config here
16:28:46 <frankbrockners> #info Recommendation from my end: We gather all re-config requirements on a wiki or etherpad and once complete - we do the change in a single maintenance window - because it will impact connectivity to the jumphosts etc.
16:29:19 <frankbrockners> Do folks see this as workable?
16:29:43 <[1]JonasB> That last thing I didnt get, how does it impact the GWs?
16:30:50 <frankbrockners> right now .9.0/24 is a single VLAN (411). If we subnet it further, it'll affect the gateway - you need a different default GW etc.
16:31:36 <[1]JonasB> Ok so for the subnetting, correct. I thought it was due to the tags inside the fabric - right - agree!
16:32:03 <frankbrockners> we could add a section on https://wiki.opnfv.org/get_started/lflab_hosting - gather, review, and once everyone is happy - exec
16:32:41 <frankbrockners> ok ?
16:32:48 <trozet> yeah lets just not take too long :)
16:32:48 <[1]JonasB> I think it is time to instead collect store this stuff in git
16:33:24 <[1]JonasB> There is little tracability on the Wiki
16:34:18 <frankbrockners> [1]JonasB: Agreed. But to gather the changes I'm good with something "quick and dirty" - etherpad might be good enough. Once done - proper documentation needs to go into git.
16:35:16 <frankbrockners> any other opinions?
16:35:55 <[1]JonasB> frankbrockners: Not other than those expressed :-)
16:36:08 <frankbrockners> Let's start with a section on wiki - we can remove it post the change - and once the setup is stable - move to git
16:36:53 <frankbrockners> #info HW config changes required to be gathered in a section on the wiki https://wiki.opnfv.org/get_started/lflab_hosting - once stable, we'll move the entire page to git (needed for Arno anyway)
16:37:05 <frankbrockners> anything else? or are we done for today?
16:37:42 * frankbrockners taking silence as a no...
16:37:45 <[1]JonasB> Have a great week-end
16:37:59 <frankbrockners> Everyone have a great weekend!
16:38:06 <frankbrockners> #endmeeting