15:04:53 <pbandzi> #startmeeting BGS weekly team meeting
15:04:53 <collabot> Meeting started Mon Jun  1 15:04:53 2015 UTC.  The chair is pbandzi. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:04:53 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:04:53 <collabot> The meeting name has been set to 'bgs_weekly_team_meeting'
15:05:22 <pbandzi> Frank will not join today
15:05:30 <pbandzi> #info Peter Bandzi
15:05:44 <[1]JonasB> #info Jonas Bjurel
15:05:49 <fdegir> #info Fatih Degirmenci
15:06:16 <jose_lausuch> #info Jose Lausuch
15:06:22 <pbandzi> agenda Laudry list updates since last meeting : https://wiki.opnfv.org/meetings/bgs#may292015
15:07:08 <pbandzi> lets start from point 1.
15:07:12 <morgan_orange> #info Morgan Richomme
15:07:14 <pbandzi> #topic  POD1 auto deploy with ODL and automated testing.
15:07:49 <pbandzi> jose_lausuch have you had a time to look at pod1 or busy with troubleshooting pod2?
15:07:53 <lmcdasm> #info Daniel Smith
15:08:23 <[1]JonasB> #info I have no information, been sick in bed all weekend and still am
15:08:24 <lmcdasm> #info JOse provided me feedback on POD1 this am, I generated a new patch and it is deployign now
15:08:47 <dfarrell07> #info Daniel Farrell
15:08:48 <lmcdasm> #info Jose is working / has attention on POD2 - so Fatih is showing me how to do deploys to POD1 so i can work on this
15:09:00 * dfarrell07 is here to help with ODL stuff if needed
15:09:01 <jose_lausuch> thanks Imcdasm
15:09:01 <lmcdasm> #info My LF credentials are not working on the Jenkins page though
15:09:17 <jose_lausuch> #info on Foreman: issues with dhcp service assigning IPs to the VMs --> vPing doesnt work. Troubleshooting ongoing.
15:09:28 <pbandzi> lmcdasm thanks so now is deploying with ODL and we are waiting for results. correct?
15:10:01 <jose_lausuch> deployment of pod1 will take 1 more hour
15:10:28 <pbandzi> ok thanks, lets go to point 2
15:10:37 <pbandzi> #topic bootable ISO for Foreman
15:10:45 <pbandzi> radez any news?
15:11:36 <pbandzi> seems not here now. anyone else know something about ISO for Foreman?
15:12:08 <radez> pbandzi: made some progress this morning
15:12:34 <lmcdasm> yes Peter - that is correct. .but it will require some more work - the initial estimate (with testing) was three weeks -
15:12:34 <radez> pbandzi: nothing new to report yet but I have gotten past the block I had last week and have tests launching
15:12:41 <lmcdasm> so we are trying to do it in 48 hours.. no easy
15:13:30 <pbandzi> lmcdasm yes understand that
15:13:48 <pbandzi> radez: thanks any estimation when this can be done?
15:14:23 <radez> pbandzi: not as of now... I should have enough of a test by COB today to report something more concrete tomorrow
15:14:37 <pbandzi> ok thank you radez
15:14:55 <pbandzi> lets move to next topic
15:15:05 <pbandzi> #topic documentation of failing tests (
15:15:18 <pbandzi> jose_lausuch I think it's your turn :)
15:15:25 <morgan_orange> I can give a status
15:15:45 <morgan_orange> #info we have currently several errors on Tempest
15:16:07 <morgan_orange> #info some are already documented in the wiki, we need to move it to the rst
15:16:29 <morgan_orange> but we are troubleshooting network issues on POD2 that could help fixing some of them
15:16:47 <morgan_orange> additionnaly there are false errors due to bad configuration of tempest (auto generated file)
15:16:54 <pbandzi> thank you morgan_orange
15:16:59 <morgan_orange> push request done towards Rally but it is not the rigth procedure
15:17:05 <morgan_orange> so trying to resubmit something
15:17:13 <morgan_orange> using the right way
15:17:21 <morgan_orange> it will fix all the invalid Creds errors
15:17:35 <MatthewL3> I am familiar with rally and tempest
15:17:40 <morgan_orange> so basically I think we shall be able to say which test is not running and why
15:17:58 <MatthewL3> if there need some help I can check that
15:18:11 <morgan_orange> assuming that vPing and ODL are successfull (which is not the case) it should be OK...
15:18:21 <morgan_orange> ok thanks MatthewL3
15:18:34 <morgan_orange> we were discussing soem stuff on opnfv-testperf
15:18:50 <pbandzi> MatthewL3 you can sync with jose_lausuch and morgan_orange
15:19:09 <morgan_orange> for tempest you may have a look at https://wiki.opnfv.org/r1_tempest
15:19:10 <MatthewL3> pbandzi: ok
15:19:52 <morgan_orange> 2 main errors remaining: Miltiple possible networks (conf workaround in progress - known issue) + networks issues (post commit on update/create network/subnet/port/...)
15:20:27 <pbandzi> thanks morgan_orange for detailed statyus
15:20:42 <morgan_orange> last run on POD2 this afternoon lead to 15 errors (but to be retested...to see if reproducible)
15:20:51 <MatthewL3> yes we can discuss offline maybe..
15:20:59 <morgan_orange> yep
15:21:35 <pbandzi> #topic finalize vPing
15:22:01 <pbandzi> I think no news because of network issue on pod2 correct?
15:22:27 <morgan_orange> yes it seems that the second VM does not get IP ..troubleshooting in progress
15:22:41 <jose_lausuch> exactly
15:22:48 <jose_lausuch> this is highes prio
15:23:00 <jose_lausuch> tim helping out
15:23:28 <pbandzi> #info no news because of network issue on pod2 - troubleshooting of network issues on POD2 in progress - trzoet is helping
15:23:38 <pbandzi> #undo
15:23:38 <collabot> Removing item from minutes: <MeetBot.ircmeeting.items.Info object at 0x1ddc8d0>
15:23:46 <pbandzi> #info no news because of network issue on pod2 - troubleshooting of network issues on POD2 in progress - trozet is helping
15:24:04 <pbandzi> #topic  ODL tests ok on both PODs
15:24:50 <pbandzi> #info POD1 deployment in progress. then will try odl tests on it
15:25:06 <jose_lausuch> what does the last sentence mean?
15:25:13 <jose_lausuch> they are not ok yet :)
15:25:39 <pbandzi> i mean i will try them when POD1 is redeployed
15:25:51 <jose_lausuch> ok
15:26:11 <jose_lausuch> but please sync with Imcdasm, because he will work on POD1 to get ODL things done
15:26:23 <pbandzi> ok
15:26:25 <lmcdasm> Peter - do you know how to setup ODL on POD1
15:26:27 <lmcdasm> ?
15:26:37 <lmcdasm> seen the scripts / undestand the steps / etc?
15:26:48 <pbandzi> not yet I thought it will be deployd by script
15:26:52 <lmcdasm> well.. no
15:27:04 <lmcdasm> thats just it - it doesnt come deployed (from jenkins) with ODL running
15:27:16 <lmcdasm> you need to login and run the scripts to turn on the ODL and set the networks
15:27:25 <jose_lausuch> we assume for Arno that we do it manually
15:27:32 <pbandzi> aha ok
15:27:36 <lmcdasm> you have to cause of the way FUEL deploys
15:27:46 <lmcdasm> to re-iterate, we didnt change FUEL upstream
15:27:56 <lmcdasm> so we let fuel deploy in VLAN seg mode and then do a transform afterwards
15:28:07 <lmcdasm> via the two scripts (stage_odl.sh - starts the ODL container and sets it up)
15:28:17 <lmcdasm> config_net_odl.sh changes the networking to the way we want
15:28:26 <lmcdasm> you will still need to remove any networks/ routers, etc that are their prior
15:28:34 <lmcdasm> (not scripted yet - if we have time maybe we can do it).
15:29:06 <pbandzi> lmcdasm we can sync offline
15:29:29 <lmcdasm> ok - but since we are in the meeting, lets coordinate now so that we are aware of your plans
15:29:35 <pbandzi> I will ping you Daniel to ask which script i shopuld check
15:29:51 <lmcdasm> well.. ok then -= so then i wont work more on POD1
15:29:55 <pbandzi> right after this meeting
15:29:57 <lmcdasm> as planned and you can
15:30:11 <jose_lausuch> btw, I started working on this
15:30:27 <lmcdasm> as outlined at the start of the meeting in the info section
15:30:38 <lmcdasm> so now im confused - are we going with the plan Jose adn I stated
15:30:43 <lmcdasm> or are we passing it to peter now?
15:30:53 <pbandzi> jose you can continue to work on it
15:31:05 <jose_lausuch> sorry, wrong place
15:31:13 <lmcdasm> Jose is owrking on POD2 with Tim, Peter.
15:31:28 <lmcdasm> im waiting for POD1 to finish deploy and test the uypdates i merged this morning
15:31:37 <lmcdasm> however, im fine to pass it ot you and you can test what you like as well
15:31:38 <pbandzi> I maybe missunderstand that. I thought that I need to start some script after deployment but seems that you are just working on this script.
15:31:44 <jose_lausuch> anyway, I did this small test yesterday
15:31:45 <jose_lausuch> http://pastebin.com/b3i8qAcJ
15:31:54 <jose_lausuch> just in case we want to even automate if the manual steps works
15:32:06 <jose_lausuch> needs to be corrected though
15:32:50 <lmcdasm> i have already put most of that into a new stage_odl.sh but we can look at it
15:33:07 <jose_lausuch> ah ok
15:33:08 <lmcdasm> for now, the issue is that br-int is not (again) being created when you hook it up to ODL
15:33:10 <jose_lausuch> then perfect
15:33:16 <lmcdasm> but they are good edits and i will incorporate them
15:33:36 <jose_lausuch> ok
15:33:49 <jose_lausuch> fdegir: ping
15:33:57 <jose_lausuch> I thinking about the job
15:33:58 <fdegir> jose_lausuch: yes
15:34:10 <jose_lausuch> if we have to activate ODL manually
15:34:23 <jose_lausuch> we mihgt need to modify fuel-daily right?
15:34:28 <fdegir> yes
15:34:41 <fdegir> or you run stuff manually one-by-one
15:34:52 <lmcdasm> perhaps we can get it working
15:35:00 <lmcdasm> and then we can figure out how to call the two scripts from Jenkins?
15:35:12 <jose_lausuch> if you manage, Daniel, we could create a new job to be ran before functest or something
15:35:17 <lmcdasm> a simple /etc/rc3.d script at the end of deployment (S99-start.sh) would do it
15:35:40 <fdegir> but don't we need to run non-odl tests before enabling odl?
15:35:41 <jose_lausuch> something like that
15:35:47 <lmcdasm> hmmm.
15:35:54 <lmcdasm> well..  can we please come to a consesus
15:35:58 <lmcdasm> we have back and forht on this
15:36:07 <fdegir> ok - donät know the details
15:36:07 <lmcdasm> Morgan has stated the System under test is ODL + OPENSTACK
15:36:13 <fdegir> just asking   - don't take my word
15:36:21 <lmcdasm> there are no reasons (unless we put it here) that we want to do NON-ODL integratedd tests
15:36:27 <lmcdasm> or have we gone on this back this?
15:36:29 <fdegir> if you agreed something - please forget me silly question
15:36:34 <lmcdasm> its ok Fatih
15:36:40 <jose_lausuch> ja, lets do the same for both pods
15:36:40 <lmcdasm> others asked the same thing this morning
15:36:45 <lmcdasm> so I would like to get consensus
15:36:51 <jose_lausuch> lets stick to OS+ODL
15:37:21 <jose_lausuch> but I would like to get the +1 from ChrisPriceAB in this regard
15:37:37 <ChrisPriceAB> oh wait what? who?
15:37:48 <jose_lausuch> :)
15:37:56 <jose_lausuch> about the target fuel test env
15:38:08 <ChrisPriceAB> we should be aiming for os/odl testing right?
15:38:12 <jose_lausuch> O/S+ODL right? to run all the tests (including vping, tesmpest, etc)
15:38:18 <ChrisPriceAB> yerp
15:38:19 <jose_lausuch> ok
15:38:21 <jose_lausuch> agreed
15:39:01 <jose_lausuch> thanks ChrisPriceAB
15:39:09 <ChrisPriceAB> np
15:39:21 <jose_lausuch> so, what I would do fdegir
15:39:32 <jose_lausuch> create a new job called: run_odl or something like that
15:39:44 <jose_lausuch> and in the fuel-daily, put it right after deployment and before functest
15:39:48 <lmcdasm> question - maybe silly
15:39:56 <lmcdasm> but dont you alrady have this ? since you are running this on POD2?
15:40:01 <lmcdasm> and POD2 comes deployed with ODL
15:40:04 <lmcdasm> so its the same thing no?
15:40:09 <jose_lausuch> mmm
15:40:12 <jose_lausuch> exactly
15:40:13 <lmcdasm> its not OS + ODL
15:40:17 <jose_lausuch> pod 2 comes deployed with odl
15:40:18 <jose_lausuch> pod1 doesnt
15:40:22 <lmcdasm> right
15:40:35 <lmcdasm> again..  please try not to think about automation until its working
15:40:41 <lmcdasm> you can deploy POD1 now
15:40:41 <jose_lausuch> ok ok
15:40:47 <lmcdasm> manually start ODL and change the nets
15:40:56 <lmcdasm> then kick off the same job you sent to POD2
15:40:57 <lmcdasm> no?
15:41:22 <jose_lausuch> why do you need to do that for pod2 if odl is already there?
15:41:25 <jose_lausuch> I dont get it
15:41:31 <lmcdasm> the same NAMED JOB as POD2
15:41:44 <pbandzi> lmcdasm, jose_lausuch do we have some steps documented we need to follow in order to make ODL run on pod1?
15:41:47 <lmcdasm> i.e. the same thing you send to POD2 now.. you would send to POD1 after thge manual steps no?
15:41:58 <lmcdasm> yes peter, its in the patch id and in the README
15:42:14 <pbandzi> ok thans
15:42:26 <lmcdasm> Jose - to try this again..  we deploy a System ..
15:42:32 <lmcdasm> we then change it to ODL mode
15:42:38 <lmcdasm> then you kick off functest
15:42:45 <lmcdasm> which "should" be the same for both PODs
15:42:47 <lmcdasm> no?
15:42:47 <jose_lausuch> but the fuel-daily job contains 3 jobs (build, deploy, functest), the same as foreman-daily
15:42:49 <jose_lausuch> aha
15:43:08 <jose_lausuch> there is no "change to ODL mode" in POD2
15:43:13 <lmcdasm> i didnt say there was
15:43:38 <jose_lausuch> mmm
15:43:45 <jose_lausuch> you said: then we change it to odl
15:44:02 <jose_lausuch> that makes sense for fuel, but not for foreman, which comes already with odl, or?
15:44:05 <jose_lausuch> or what am I missing?
15:44:08 <lmcdasm> its the same thing
15:44:16 <lmcdasm> you are doing the same thing for both PODs
15:44:20 <lmcdasm> you do build, deploy and functest
15:44:25 <jose_lausuch> yep
15:44:27 <lmcdasm> but for POD1, before you kick off functest
15:44:31 <lmcdasm> there is a manual step
15:44:33 <jose_lausuch> yes
15:44:50 <lmcdasm> the functest job for POD1 (FUEL) should be identical to the one for FOREMAN - correct?
15:44:56 <jose_lausuch> yes
15:45:23 <lmcdasm> ok.. so...  are we clear then or?
15:45:31 <jose_lausuch> yes
15:45:45 <jose_lausuch> what I ment is that we cannot just run functest after deploy for Fuel
15:45:49 <jose_lausuch> because ODL will not be in place
15:46:19 <jose_lausuch> we need to do it manually
15:46:23 <jose_lausuch> and then run functest
15:46:30 <jose_lausuch> so we need to modify fuel-daily-master job
15:46:35 <jose_lausuch> to not execute functest
15:46:54 <jose_lausuch> (it is not the same daily job for both pods)
15:47:01 <jose_lausuch> fdegir: right?
15:47:06 <fdegir> correct
15:47:22 <jose_lausuch> so we can remove functest job inside fuel-daily
15:47:29 <jose_lausuch> and run it manually when ODL is in place
15:47:48 <fdegir> yes
15:47:54 <jose_lausuch> ok
15:48:03 <jose_lausuch> Daniel, do you agree?
15:48:22 <lmcdasm> sure...
15:48:57 <lmcdasm> and then we will put it back when we have it working and then you can have some small thing that at the start of functest calls the change.
15:49:06 <lmcdasm> or whatever (again - cross that when we come to it).
15:49:14 <jose_lausuch> ja, thats an option
15:49:22 <jose_lausuch> although I wouldnt include in functest
15:49:26 <jose_lausuch> but will see
15:49:31 <jose_lausuch> lets get first thing first
15:49:54 <pbandzi> #info in order to allow add ODL manualy we remove functest job inside fuel-daily, and run it manually when ODL is in place
15:51:24 <jose_lausuch> yep
15:51:46 <pbandzi> Do we want to discuss anything else or we can end meeting now?
15:51:56 <jose_lausuch> nothing from my side
15:52:40 <pbandzi> Thank you all ...
15:52:53 <pbandzi> #endmeeting