16:00:02 <frankbrockners> #startmeeting OPNFV BGS daily release readiness synch
16:00:02 <collabot> Meeting started Wed May 20 16:00:02 2015 UTC.  The chair is frankbrockners. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:02 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:00:02 <collabot> The meeting name has been set to 'opnfv_bgs_daily_release_readiness_synch'
16:00:08 <frankbrockners> #info Frank Brockners
16:00:15 <trozet> #info Tim Rozet
16:00:48 <pbandzi> #info Peter Bandzi
16:02:26 <jose_lausuch> #info Jose Lausuch
16:04:44 <[1]JonasB> #info Jonas Bjurel
16:07:57 <ChrisPriceAB> #info chris price
16:09:22 <pbandzi> who
16:09:26 <pbandzi> sorry
16:10:18 <pbandzi> differnt window
16:10:35 <ChrisPriceAB> Hehe
16:12:04 <jose_lausuch> shall we start? :)
16:13:05 <pbandzi> I think i can provide brief updat for ODL test
16:13:33 <jose_lausuch> ok
16:14:14 <pbandzi> #info ODL tests were PASSED on POD2, but need to add cleaning after them, because probably it cause troube to other functests
16:14:26 <pbandzi> #info going to try them on POD1
16:14:29 <pbandzi> end
16:14:38 <jose_lausuch> pbandzi: what kind of clean do you need?
16:14:40 <jose_lausuch> neutron networks?
16:14:48 <jose_lausuch> instances?
16:15:27 <pbandzi> yes i create networks and subnets and porst and i have already created other test which delete these but they are not yet in repo so I am going to push them
16:15:36 <jose_lausuch> ok
16:15:57 <jose_lausuch> anyway, ODL tests are ran after rally
16:16:05 <[1]JonasB> pbandzi: They will fail on POD1 with current deployment, we need to prepare a non HA deployment for the ODL tests
16:16:12 <trozet> yeah but on the last functest run they were ran while rally was running
16:16:18 <trozet> which might have choked up the rally run
16:16:45 <jose_lausuch> ah ok
16:16:47 <jose_lausuch> thats strange
16:16:50 <jose_lausuch> it should be sequential
16:17:00 <frankbrockners> hmm ... isn't is that ODL is in non-clustered mode anyway?
16:17:13 <frankbrockners> so single instance of ODL in all cases - or not?
16:17:17 <pbandzi> [1]JonasB: ok. is Stefan working on POD1 now? I can synchronize with him
16:17:19 <trozet> pbandzi triggered the ODL run manually cause we wanted to test something
16:17:42 <trozet> ODL is not clustered, single instance
16:18:03 <trozet> oh you were asking about POD1, sorry
16:18:06 <frankbrockners> ... so that should mean we have the same setup for ODL across both PODs
16:18:19 <frankbrockners> i.e. testing for ODL with Robot should be the same
16:19:43 <[1]JonasB> As said, for POD1 we need to set-up a non HA deployment for ODLL testing
16:20:22 * ChrisPriceAB Q? we are not running odl in ha for Arno in either deploy?
16:20:29 <frankbrockners> [1]JonasB: Am a bit confused: ODL is always non-HA
16:20:39 <trozet> yeah ODL doesnt support HA in helium
16:20:58 * ChrisPriceAB ok cool, I'll stop telling lies.
16:21:05 <trozet> :)
16:21:48 <[1]JonasB> Yes, but as I have explained - we (fuel) doesnt support combbbination of OS HA and ODL.
16:22:26 <[1]JonasB> That     is unfortunate, but still the case, we're working on it for SR1
16:22:55 <frankbrockners> ah ok - it is HA in Openstack and not ODL
16:23:15 <frankbrockners> that means the main setup we'd test for Fuel would be a non-HA OS with ODL - correct?
16:23:33 <[1]JonasB> Yep, combination OS HA and Standard neutrom ML2 - OK
16:23:45 <[1]JonasB> But not OS HA + ODL
16:24:59 <frankbrockners> ok - let's focus testing on OS + ODL without HA then for Fuel.
16:25:22 <frankbrockners> IMHO the native neutron networking is mostly a nice to have
16:26:35 <[1]JonasB> I need to jump off, just one update
16:26:41 <frankbrockners> [1]JonasB: What is the default deploy that happens nightly right now?
16:26:58 <frankbrockners> is that with or without ODL?
16:27:08 <frankbrockners> (I think it misses ODL right now)
16:27:48 <[1]JonasB> #info Four consecutive successfull jenkins automated deploys since yesturday.     We're running every 4 hours now
16:28:41 <frankbrockners> [1]JonasB: Per the question above: Is this a OS + ODL deploy - or OS standalone?
16:29:22 * frankbrockners Jonas seems to have dropped... - Let's table the POD1 questions for now
16:29:36 <frankbrockners> let's move to updates on POD2
16:29:43 <frankbrockners> #info updates on POD2
16:29:52 <frankbrockners> trozet: Quick update from your end?
16:30:22 <trozet> #info still working on the external net patch.  No real updates.  Trying to fix one last issue before I commit
16:30:38 <jose_lausuch> #info latest deployments on POD1 succeded https://build.opnfv.org/ci/job/genesis-fuel-deploy/lastBuild/console
16:30:46 <trozet> #info Fatih has disabled the every 6 hour runs until I am done debugging this
16:30:50 <trozet> #info then we will turn it back on
16:30:57 <frankbrockners> trozet: Any hope to have the patch by tomorrow?
16:31:32 <frankbrockners> jose_lausuch: Do you know what is being deployed? Is this OS+ODL - or is this plain OS (with HA)?
16:31:40 <trozet> i hope so.  Changing linux interface config has big conflicts with puppet.  Trying to fix it
16:31:52 <jose_lausuch> I think its HA, but would need confirmation
16:32:00 <jose_lausuch> it is HA because we got 3 controllers yes
16:32:03 <jose_lausuch> I checked that
16:32:24 <frankbrockners> jose_lausuch: Could we switch to using ODL?
16:32:36 <jose_lausuch> yes
16:32:42 <jose_lausuch> but I dont know the impacts on the tests
16:32:46 <jose_lausuch> hopefully none
16:32:56 <frankbrockners> that way we can at least run the ODL tests on POD1
16:33:06 <jose_lausuch> yes, proceed that way
16:33:09 <jose_lausuch> it should be fine
16:33:11 <frankbrockners> immediate impact would be: We can test ODL
16:33:17 <jose_lausuch> that way we can also see how our tests behave
16:33:22 <jose_lausuch> yep
16:33:24 <jose_lausuch> :)
16:33:26 <frankbrockners> let's info this in...
16:34:05 <frankbrockners> #info Auto-deploy on POD1 to switch to OpenStack + ODL deployment (i.e. enable ODL): That way ODL Robot tests can run on POD1 as well
16:34:20 <frankbrockners> pbandzi: Do we have Robot ready on POD1 as well?
16:35:22 <pbandzi> I can prepare it today before go home
16:35:47 <frankbrockners> ok - that way we might be able to run it on the next deploy -
16:36:04 <frankbrockners> I'll drop an email to Fatih
16:36:10 <jose_lausuch> POD1 is still set to run every 6 hours, right?
16:37:52 * frankbrockners seems that noone has an immediate answer
16:38:17 <jose_lausuch> ok:)
16:38:26 <frankbrockners> jose_lausuch: Any additional updates on testing?
16:38:45 <jose_lausuch> yes
16:38:50 <jose_lausuch> #info functest jenkins job failed because of 1)bad default credentials, 2) failed command in the job.
16:38:50 <frankbrockners> did we make any progress on fixing some of the failed tests etc.?
16:39:07 <jose_lausuch> not really, troubleshooting the jenkins problems
16:39:08 <jose_lausuch> #info 1) the OS_AUTH_URL is set now to http://172.30.9.70:5000/v2.0. It should be the same from now on. It was a different one before triggering the jenkins automatic deployments
16:39:16 <jose_lausuch> #info 2) I improved the jenkins job by removing config_functest.py script and downloading it again every time to have always the latest from master, the command rm $HOME/functest/config_functest.py 2&>/dev/null fails when ran from jenkins. Need to check with Fatih
16:39:16 <jose_lausuch> https://build.opnfv.org/ci/view/functest/job/functest-opnfv-jump-1/lastBuild/console .
16:39:55 <jose_lausuch> for 1) I contacted Stefan and we agreed (for now) to have that URL, which is fixed
16:40:04 <jose_lausuch> but it is not an elegant solution (not portable)
16:40:21 <jose_lausuch> for 2) I need to talk to Fatih or Aric
16:40:26 <jose_lausuch> not to be done here right now though
16:40:55 <jose_lausuch> I had also problems deploying a VM in POD2 today, but havent had the time to troubleshoot why yet
16:41:23 <jose_lausuch> thats it
16:42:54 <frankbrockners> thanks jose_lausuch
16:43:09 <jose_lausuch> #info btw, Aric installed the jenkins plugin showin the "Next executions" or scheduled jobs
16:43:24 * fdegir have a question
16:43:29 <jose_lausuch> so you can see what will be executed next
16:43:48 <jose_lausuch> shoot
16:44:09 <fdegir> I've seen functest job hanging number couple of times
16:44:23 <fdegir> this blocks triggering of the next job
16:44:27 <jose_lausuch> in which pod?
16:44:32 <fdegir> we can introduce timeout
16:44:38 <jose_lausuch> yes, Im doing it now :)
16:44:40 <fdegir> I think it was on pod2
16:44:43 <jose_lausuch> it was vPing actually
16:44:53 <jose_lausuch> waiting forever for a VM to come up
16:44:58 <fdegir> and one more yesterday which I can't remember which pod was it
16:45:03 <fdegir> yep
16:45:03 <jose_lausuch> I Will push a patch later to cope with that
16:45:12 <jose_lausuch> with a timeout
16:45:20 <fdegir> ok
16:45:29 <fdegir> then I don't add timeout in jenkins
16:45:36 <jose_lausuch> there is a timeout for the ping, but there was not for the VM build process
16:45:40 <jose_lausuch> no need
16:45:43 <jose_lausuch> easier in the script
16:45:50 <fdegir> ok
16:46:00 <jose_lausuch> I need your help later :)
16:46:06 <fdegir> yep, we talk later
16:46:08 <jose_lausuch> for the job .yaml
16:46:09 <jose_lausuch> ok
16:46:28 <frankbrockners> fdegir: Different topic. Now that you're here, could you switch autodeploy on POD1 to deploy a O/S + ODL setup (rather than O/S standalone) - so we can test ODL?
16:47:01 <fdegir> I can if someone tells me what I need to do
16:47:11 <fdegir> who should I talk to?
16:47:25 <jose_lausuch> stefan_berg: ping ?
16:47:27 <frankbrockners> Probably check with Jonas or Daniel Smith
16:47:31 <fdegir> ok
16:47:33 <frankbrockners> or Stefan right
16:47:35 <fdegir> will contact them
16:47:39 <frankbrockners> many thanks
16:47:43 <fdegir> no
16:47:45 <fdegir> np
16:47:50 <frankbrockners> will also send a quick email
16:48:00 <frankbrockners> anything else to cover today?
16:48:46 <jose_lausuch> not from my side
16:49:02 <frankbrockners> ok... - looks like we're done
16:49:11 <frankbrockners> thanks everyone...
16:49:14 <frankbrockners> #endmeeting