13:57:33 <[1]JonasB> #startmeeting BGS Fuel status meeting
13:57:33 <collabot> Meeting started Thu May 28 13:57:33 2015 UTC.  The chair is [1]JonasB. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:57:33 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
13:57:33 <collabot> The meeting name has been set to 'bgs_fuel_status_meeting'
13:57:42 <[1]JonasB> #info Jonas Bjurel
13:57:47 <lmcdasm> #info Daniel Smith
13:58:11 <jose_lausuch> parallel weekly functest meeting
13:58:14 <jose_lausuch> #info Jose Lausuch
13:58:29 <mskalski> #info Michal Skalski
13:58:36 <szilard> #info Szilard Cserey
13:59:55 <[1]JonasB> So I guess the ODL testing is the topic for today, but since Jose is in a parallell meeting maybe it is not worth while?
14:00:17 <[1]JonasB> Jose what do you think?
14:00:25 <stefan_berg> #info Stefan Berg
14:00:34 <jose_lausuch> I can try
14:01:03 <[1]JonasB> Jose: a short summary?
14:01:10 <jose_lausuch> difficult :)
14:01:23 <jose_lausuch> its a bit of a messy situation right now
14:01:26 <jose_lausuch> lets see
14:02:29 <jose_lausuch> #info we are getting to know now how foreman implements ODL, so that we can do the same things for Fuel deployments
14:03:09 <jose_lausuch> #info Foreman deployment currently ongoing, later we will check what is needed to understand how it is done
14:03:33 <jose_lausuch> #info troubles about stablishing priorities
14:03:41 <jose_lausuch> #info discussions ongoing
14:03:50 <lmcdasm> info from Chris (Same thing Jose and I asked Dan) - is that FOREMAN deployed Openstack with ODL config and then switches as well
14:04:14 <jose_lausuch> #info focus for Fuel deployment:
14:04:15 <lmcdasm> still not clear when / where they are testing and how that is done (i.e. test Openstack and then do a switch to ODL mode and run the "same" tests again).
14:04:27 <[1]JonasB> Ok, Daniel do you have the credentials to log into POD 2 and fish the bridge config?
14:04:35 <jose_lausuch> I have credentials
14:04:37 <lmcdasm> no.. i dont, i mean i can go looking
14:04:38 <jose_lausuch> I can help on that
14:05:06 <jose_lausuch> #info   1)      Deploy  2)      Run functes SET1  3)      Switch to ODL + reboot  (if nonHA deployment)  4)      Run functest SET2            (if nonHA deployment)
14:05:08 <lmcdasm> but really - four days to hear - i can go looking myself? - not very collaborative.. where is the other people with Jose and i working on this for a week reelase?
14:05:21 <jose_lausuch> $info   SET1 = Tempest + vPing  + Rally Bench,  SET2 = ODL single test using robot that Peter wrote (creating neutron net/subnet/port and checking the result also in ODL rest api, similar to your curl approach)
14:05:35 <lmcdasm> thanks Jose - is that what is being done on FOREMAN then?
14:05:43 <jose_lausuch> No
14:05:47 <jose_lausuch> foreman currently does:
14:05:54 <[1]JonasB> jose: Can you help Daniel to get the config he wants to look at.
14:05:57 <jose_lausuch> 1)      Deploy  2)      Run functes SET1+SET2
14:06:02 <jose_lausuch> yes
14:06:09 <lmcdasm> How??  that doesnt make sense
14:06:27 <lmcdasm> Dan just informed Chris that FOREMAN doesnt come up in ODL mode
14:06:31 <lmcdasm> so SET2 then always fails
14:06:40 <lmcdasm> unless they do the same switch that we are outlining for FUEL
14:06:44 <jose_lausuch> yes, it fails and asked for help to troubleshoot
14:06:57 <jose_lausuch> well, it was like this for Fuel as well, before knowing about that ODL impacts O/S tests
14:07:00 <lmcdasm> SET2 is ALWAYS GOING TO FAIL cause the opentack is not in ODL mode
14:07:01 <jose_lausuch> eactly
14:07:06 <[1]JonasB> jose: what fails - POD2?
14:07:06 <jose_lausuch> thats the info we are waiting for, or?
14:07:07 <aricg> can someone merge this:  https://gerrit.opnfv.org/gerrit/#/c/687/
14:07:16 <jose_lausuch> ODL test on POD2, I saw a failure yesterday
14:07:17 <aricg> Found the line that was killing rst2pdf
14:07:20 <lmcdasm> (yes aric)
14:07:22 <lmcdasm> i will now
14:07:45 <lmcdasm> msg aricg done
14:07:49 <aricg> thx
14:07:59 <lmcdasm> ok - so that its clear
14:08:06 <lmcdasm> POD2 is doing the same semantrics we are doing
14:08:11 <lmcdasm> your SET2 will alays fail
14:08:18 <lmcdasm> cause you havent set POD2 (FORMEAN) into ODL mode
14:08:24 <jose_lausuch> #info need to modify the Jenkins job to trigger ODL test when it makes sense
14:08:27 <lmcdasm> they are doing (as i have come to piuck apart)
14:08:32 <lmcdasm> the same thing we are
14:08:36 <jose_lausuch> ok
14:08:38 <lmcdasm> now - my questino still stands
14:08:43 <lmcdasm> when you put the system in ODL mode
14:08:48 <lmcdasm> are you only doing SET2
14:08:49 <jose_lausuch> taht is the information we missed (I had no clue about) and we need to integrate
14:08:53 <lmcdasm> or are you doing all the tests again?
14:08:55 <jose_lausuch> no
14:08:59 <jose_lausuch> the idea would be
14:09:07 <jose_lausuch> correct me if Im wrong or if it doesnt make sense
14:09:10 <jose_lausuch> the same as Fuel
14:09:23 <jose_lausuch> 1)      Deploy  2)      Run functes SET1  3)      Switch to ODL   4)      Run functest SET2
14:09:31 <[1]JonasB> jose: sound as the right thing to do
14:09:38 <jose_lausuch> ok
14:09:50 <jose_lausuch> but still need info about the procedure of how to do step 3
14:10:01 <[1]JonasB> lmcdasm: Do you agree?
14:10:27 <lmcdasm> well.. if that is what we want to do then fine
14:10:33 <lmcdasm> shouldnt we be doing the same things on both systems
14:10:34 <jose_lausuch> sorry for the ignorance, but I dont know about how ODL is integrated and switched on/off and this things
14:10:38 <lmcdasm> since they are both deploying the same way?
14:10:46 <aricg> This needs to be rebased now, and it will pass as well. https://gerrit.opnfv.org/gerrit/#/c/673/
14:10:51 <jose_lausuch> that is what we have to figure out now
14:10:55 <jose_lausuch> what makes sense
14:11:14 <jose_lausuch> what would you say it makes sense Daniel?
14:11:18 <lmcdasm> i cant tell you
14:11:20 <aricg> ChrisPriceAB: I guess that one was for you ^ :)
14:11:23 <lmcdasm> sorry.. i dont set the release spec
14:11:31 <lmcdasm> this is defined from the PTO and the TSC
14:11:57 <lmcdasm> we have asked for this infomation weekly (the whole picture) for about a month
14:12:07 <lmcdasm> and get little pieces.. so i think that the testing team shoudl sit down
14:12:15 <[1]JonasB> lmcdasm: I think we need to be pragmatic here and now and propose the way forward
14:12:16 <lmcdasm> and come up with the plan and ask the two deploers how they are doing things
14:12:19 <jose_lausuch> wait
14:12:29 <jose_lausuch> I got this info right now from Morgan (didnt know)
14:12:40 <[1]JonasB> lets hear
14:13:06 <jose_lausuch> yesterday, 6 ODL tests were OK 12 failed
14:13:32 <lmcdasm> can you give a bit more context?
14:13:53 <lmcdasm> on which POD.. and since there are only 10 cases, im not sure how that can be (reading from Peters website).
14:14:17 <jose_lausuch> ok, forget that, the errors are due to some other things
14:14:20 <jose_lausuch> need to troubleshoot
14:14:25 <jose_lausuch> sorry, forget what I have said
14:14:35 <lmcdasm> https://wiki.opnfv.org/r1_odl_suite
14:14:40 <lmcdasm> this is what i understand we are testing again
14:14:47 <lmcdasm> is that the complete test case for ODL?
14:15:09 <jose_lausuch> yes
14:15:14 <lmcdasm> on the table it says that all 16 cases have passed for POD2
14:15:23 <lmcdasm> so there is a discrpancy..
14:15:24 <jose_lausuch> manually yes
14:15:27 <jose_lausuch> from jenkins not yet
14:15:41 <jose_lausuch> I havent done myself on POD2, I think Peter will know
14:15:50 <lmcdasm> ok.. so then the FUEL side (since you did it on Monday can be updated as well)(
14:15:51 <lmcdasm> ?
14:15:52 <jose_lausuch> I managed to run it on POD1 manually succesfully as well
14:15:52 <stefan_berg> /window show 5
14:16:10 <jose_lausuch> yes, but doing some tweaks
14:16:19 <[1]JonasB> jose: Can you then update the Wiki page
14:16:21 <lmcdasm> what tweeks?
14:16:33 <jose_lausuch> bugs in the script which work manually but not from Jenkins
14:16:34 <ChrisPriceAB> on it aricg, will report back
14:16:35 <lmcdasm> it should be the same manualy tests that peter ran no?
14:16:39 <jose_lausuch> the goal is to make it ran from jenkins
14:16:51 <jose_lausuch> yes, and I should tell him
14:17:03 <jose_lausuch> that it is not acceptable to have it PASSED if its not like that from Jenkins
14:17:18 <jose_lausuch> and currently they are not PASSED (at least not all of them) in last functest job last night
14:17:33 <jose_lausuch> I wil update the wiki
14:17:34 <lmcdasm> sure - sounds good.. im still super confused that we dont have a solid test specifictaion
14:17:36 <jose_lausuch> this is a bit old info
14:17:42 <lmcdasm> is it 12 cases, I count 16,
14:17:59 <jose_lausuch> forget about the 12
14:18:02 <lmcdasm> and sure we can say " in Fuel - we are doing this", but then we arent comparing apples to apples if the setup for FOREMAN is way different
14:18:11 <lmcdasm> but seems like maybe im the only one screaming about this.
14:18:25 <jose_lausuch> I am also screaming now :)
14:18:42 <jose_lausuch> what do you suggest?
14:18:50 <jose_lausuch> to have different steps for Fuel and Foreman?
14:19:07 <lmcdasm> well.. i think a slower approach is needed
14:19:15 <lmcdasm> first - are we clear how to set the FOREMAN machine into ODL mode?
14:19:26 <lmcdasm> second- we make the semantic in FUEL the same
14:19:34 <jose_lausuch> for 1) Im not, who is?
14:19:45 <jose_lausuch> aricg: ping
14:19:48 <jose_lausuch> sorry
14:19:49 <[1]JonasB> jose: pbanzi must know
14:19:51 <jose_lausuch> fdegir: ping
14:19:59 <lmcdasm> third - the testing team releases a clear ppt / or one pager that shows the tesing scope and how the test are run - just like the 5+ charts differenet people have given
14:20:11 <lmcdasm> hold on a second here
14:20:17 <fdegir> jose_lausuch: yes
14:20:22 <jose_lausuch> can we have 2 json jobs, 1 for foreman and one for fuel if we have different approach about how to swithc to ODL?
14:20:27 <lmcdasm> how can the testing reports be coming in at all if the testing team doesnt know how to set it into ODL?
14:20:38 <lmcdasm> so that means really that no testing on ODL (from functest) has been done ?
14:21:19 <fdegir> jose_lausuch: you mean 2 different jobs - one for fuel and the other for foreman
14:21:19 <jose_lausuch> we assumed that was transparent for the tests
14:21:23 <jose_lausuch> yes fdegir
14:21:26 <fdegir> if so, yes
14:21:46 <lmcdasm> ok.. well. hmm. then i would suggest that we take a step back and get familliar with what we are testing.
14:21:58 <jose_lausuch> the switching you are talking about is something new, we assumed ODL was available from scratch (wrong assumption, but nobody said the opposite)
14:22:03 <lmcdasm> it does no good to saY " i got a passed" if we dont understand (at least a bit) what we are actually testing
14:22:30 <jose_lausuch> I think we need peter for this discussion
14:22:44 <lmcdasm> ODL implementation creates a overlay so changes all the networking - which is why im asking if you should not be doing the whole testing again - since the ODL tests now just show a simple subset, but we know that when you try to do other Openstack things, it breaks,
14:22:46 <jose_lausuch> thanks fdegir
14:23:17 <jose_lausuch> so then I would like to know exactly what you asked:   -          How are we deploying things (single deploy, double deploy, deploy then alter?).
14:23:21 <lmcdasm> anyway - its moot
14:23:28 <[1]JonasB> ping trozet
14:23:36 <lmcdasm> if the methods for how to change the mode to ODL arent in place in the testing machine yet
14:23:39 <lmcdasm> that is your first step
14:23:53 <lmcdasm> (after you read the overview section from the opendaylight site that is :) )
14:23:59 <trozet> hi [1]JonasB
14:24:26 <jose_lausuch> I dont want to spend 1 day understanding that, I asked for help about who could answer taht quickly
14:24:32 <jose_lausuch> we dont have time to build up competence right now
14:24:36 <[1]JonasB> trozet: Can you share som light on if ODL is automatically activated when you deploy Forman
14:24:44 <jose_lausuch> after Arno we have to for sure
14:24:48 <jose_lausuch> but we have to get things done
14:25:40 <lmcdasm> that doesnt make sense.
14:25:50 <lmcdasm> going back to the point of testing, if you are just looking for a greenlight
14:25:54 <lmcdasm> then how do you know its acutally working
14:26:36 <lmcdasm> if you dont take 10 minutes to read the page then it falls to others to take that time to give you the explanation.. anyway - doesnt matter, but the picture of what ODL is doing speaks a thousand words
14:26:51 <lmcdasm> and would have avoided the assumption that a) tthere is no difference between when the system is in ODL mode (vxlan) or not
14:27:06 <trozet> [1]JonasB: ODL is configured if the odl_flag is set to true
14:27:07 <lmcdasm> b) undestand the IRP - integration reference points - so that the test cases have some relevance.
14:27:07 <jose_lausuch> if its 10 minutes then its ok, I was expecting hundres of pages of documentation
14:27:16 <trozet> in our yaml inventory file
14:27:37 <lmcdasm> hey Tim - so can you outline the steps that are needed to switcvh to ODL mode once the system is deployed?
14:27:43 <[1]JonasB> trozet: DO you know if that is default?
14:27:44 <lmcdasm> or do you only deploy one way or the other?
14:28:04 <jose_lausuch> these are the key questions
14:28:11 <trozet> [1]JonasB: it is the default
14:28:18 <trozet> hi lmcdasm
14:28:29 <jose_lausuch> so we are deploying every day with ODL mode
14:28:37 <trozet> yes
14:28:40 <lmcdasm> and the vxlan tunnel are coming up with each deployment
14:28:47 <lmcdasm> and integrated to ODL
14:28:49 <lmcdasm> ?
14:28:49 <trozet> yes
14:28:59 <trozet> ODL running on controller1
14:29:05 * ChrisPriceAB It should deploy with ODL and be validated in that state. That's our target
14:29:07 <jose_lausuch> next: does it make sense to test Tempest in that setup?
14:29:12 <trozet> all OVS switches (5 of them) are connected to ODL
14:29:23 <trozet> all OVS switches meaning br-int
14:29:25 <jose_lausuch> as far as I understand from Daniel, It doesnt
14:29:38 <trozet> yes it does
14:29:55 <trozet> thats the point of Arno is ODL + O/S integration testing
14:29:58 <trozet> since we are the first to do it
14:30:00 <lmcdasm> well. on FUEL not yet.. i need to get in (it would seem to login myself) to see how they are doing things to bring it up iun the same fashion
14:30:03 <trozet> we need to shake those bugs out
14:30:14 * ChrisPriceAB aricg the rebase worked fine.
14:30:18 <lmcdasm> however, this means that when testing has been running, its been doing the full suite in ODL/VXLAN mode right?
14:30:26 <trozet> right
14:30:30 <jose_lausuch> thats what I assume yes
14:30:40 <lmcdasm> so then Jose, our approach has been wrong
14:30:50 <trozet> and there are bugs in ODL not setting up vxlan tunnels correctly
14:30:53 <lmcdasm> if the goal is to have semblance between the two
14:30:58 <jose_lausuch> Im just wondering if some of the tempest test cases have failed because of having ODL (hopefully not because of that)
14:31:16 <lmcdasm> yes Jose - thats why the tempatest have been failing moresore on POD2 than on POD1
14:31:25 <lmcdasm> cause on POD1, you are doing it in VLAN segregation mode
14:31:29 <lmcdasm> not with VXLAN tunnel setup at all
14:31:43 <lmcdasm> and for FUEL, im telling you that when you go full funtest in ODL/VXLAN mode its going to be worse
14:31:44 <jose_lausuch> ok
14:31:50 <lmcdasm> we are using a really really old OVS
14:31:54 <jose_lausuch> yes, I understood that
14:31:58 <lmcdasm> so ..  we need to reset the dial on all tests
14:31:59 <lmcdasm> i think
14:32:00 <jose_lausuch> thats why I proposed having 4 steps
14:32:07 <lmcdasm> cause there is no clarity on the setups you have been testing
14:32:17 <lmcdasm> you have already been testing FULL TEST with VXLAN/ODL
14:32:17 <jose_lausuch> 1)      Deploy  2)      Run functes SET1  3)      Switch to ODL + reboot  (if nonHA deployment)  4)      Run functest SET2            (if nonHA deployment)
14:32:20 <lmcdasm> no
14:32:27 <lmcdasm> what is the point of switching
14:32:28 <lmcdasm> ?
14:32:39 <jose_lausuch> mmm
14:32:42 <lmcdasm> CHris has stated that the result is Opesntack with ODL(VXLAN)
14:32:52 * ChrisPriceAB can I have a BGS committer push the amended patch for https://gerrit.opnfv.org/gerrit/#/c/673/. Only change since last commit is the rebase.
14:32:55 <trozet> +1 theres no point in doing anything non ODL
14:32:57 <jose_lausuch> right
14:33:14 <jose_lausuch> but that is why you said it would take 3 weeks work?
14:33:28 <jose_lausuch> or it is something quicker to fix?
14:33:30 <lmcdasm> correct.. if you want the full test (which is what i understand you want)
14:33:44 <lmcdasm> i dont know Jose, how goes the fixing of the bugs on FOREMAN?
14:33:49 <ChrisPriceAB> jose_lausuch, lmcdasm: Isn't it best to run the tests on the deployed state.  Otherwise we don't know the state of the tempest tests after the ODL integration...
14:33:50 <jose_lausuch> Daniel, I dont want anything, I just want to get done what is required by the project
14:33:50 <lmcdasm> that is the question
14:34:10 <lmcdasm> ok.. i dont know what to say
14:34:11 <trozet> we file the bugs with the appropriate software
14:34:15 <trozet> ive filed a bug on ODL
14:34:20 <trozet> for O/S integration already
14:34:21 <ChrisPriceAB> 1) Deploy 2) Switch to ODL + reboot (if nonHA deployment) 3) Run functest all (if nonHA deployment)
14:34:22 <trozet> we need to file more
14:34:33 <lmcdasm> Tim - go you guys have HA working with ODL at all?
14:34:39 <lmcdasm> if not, then we can stop talking about HA and non-HA
14:34:49 <ChrisPriceAB> HA ODL does not work in Helium
14:34:55 <trozet> O/S HA + non-HA ODL (HA ODL will only be available in Lithium)
14:35:01 <trozet> but let me clarify something
14:35:06 <jose_lausuch> are we talking now the whole time about Arno?
14:35:09 <lmcdasm> thanks Chris- however, im talking about at the Ostack level with HAProxy
14:35:11 <ChrisPriceAB> yes
14:35:22 <lmcdasm> not about the native ODL support in ODL
14:35:41 <trozet> Fuel is running non-HA O/S + ODL, when you do that you are going to have a lot more passes than HA O/S+ODL
14:35:46 <trozet> as far as tempest testing
14:35:54 <trozet> so its not apples to apples to compare foreman and fuel results
14:35:55 <lmcdasm> no we arent
14:36:03 <lmcdasm> we havent run any tempets with ODL turned on
14:36:08 <trozet> I
14:36:08 <jose_lausuch> ChrisPriceAB: step 3 of your statement is not clear
14:36:09 <lmcdasm> its only been VLAN Seg deployment
14:36:16 <trozet> lmcdasm: I'm saying if you do run
14:36:19 <jose_lausuch> if its a HA deployment we dont run functest at all?
14:36:23 <trozet> with the identical setup to us
14:36:30 <trozet> except non -HA O/S
14:36:35 <trozet> you will have more passes
14:36:40 <trozet> the reason is
14:36:58 <trozet> because ODL has trouble figuring out where vxlan endpoints should be, when you have that many OVS instances
14:37:18 <trozet> 2 compute ovs instances + 1ovs instance (NON HA) vs 2 compute ovs instances and 3 ovs instances on controllers (HA)
14:37:33 <trozet> so folks shouldnt compare results from pod 2 to pod 1
14:37:34 * ChrisPriceAB jose_lausuch: we just need to run the same suite against POD1 and POD2. (if we don't have OS HA on POD1 the results may differ but that's OK=)
14:37:47 <trozet> right
14:37:49 <jose_lausuch> ok
14:37:56 <trozet> then dig into test cases
14:37:58 <trozet> figure out why they fail
14:37:58 <jose_lausuch> as long it is documented, right?
14:37:59 <trozet> file bugs
14:38:04 <trozet> yeah
14:38:07 <trozet> we need to know why they fail
14:38:15 <trozet> it cant be because an installer installed wrong
14:38:20 <trozet> it needs to be a real bug
14:38:25 <jose_lausuch> then I would correct it to: 1) Deploy 2) Switch to ODL + reboot (if nonHA deployment) 3) Run functest all
14:38:35 <lmcdasm> nope Jose
14:38:40 <lmcdasm> you need two different approaches
14:38:45 <lmcdasm> for FOREMAN ODL is deployed
14:38:49 <lmcdasm> for FUEL you will need to switch
14:39:00 <jose_lausuch> yes, step2
14:39:05 * ChrisPriceAB yes but before the tests are run
14:39:18 <lmcdasm> are we all agreed that we will run no TESTING again anything without ODL / VXLAN turned on?
14:39:20 * ChrisPriceAB that's part of the deploy flow
14:39:35 <lmcdasm> no Jose - FOREMAN deploys ODL nativelly. so you do 1) Deploy and 3) Run test all
14:39:37 <lmcdasm> no switching
14:39:46 <jose_lausuch> Im talking about fuel only
14:39:52 <jose_lausuch> I know there is no switch in Foreman
14:39:56 <jose_lausuch> sorry for not being clear
14:40:12 <jose_lausuch> FUEL --> 1) Deploy 2) Switch to ODL + reboot (if nonHA deployment) 3) Run functest all
14:40:25 <jose_lausuch> FOREMAN --> 1) Deploy 2) Run functest all
14:40:25 <lmcdasm> well.. now it begs a real questino
14:40:29 <jose_lausuch> exactly
14:40:34 <jose_lausuch> that comes your concern
14:40:39 <lmcdasm> then why not change FUEL puppet to simply deploy in ODL mode then
14:40:52 <lmcdasm> cause seems to me that we have now done weeks of testing on an object that we dont want
14:41:03 <jose_lausuch> having Fuel with nonHA+ODL, you said Temepst will not work
14:41:08 <lmcdasm> nope
14:41:09 <jose_lausuch> so what is the poing of that?
14:41:19 <jose_lausuch> point*
14:41:20 <lmcdasm> it needs work - wont work as it is today
14:41:25 <jose_lausuch> excactly
14:41:36 <jose_lausuch> so ChrisPriceAB how do we handle that?
14:41:38 <lmcdasm> so.. if you want to run the switch
14:41:40 <jose_lausuch> what is the focus for Arno?
14:41:40 <lmcdasm> then thats fine.
14:41:59 * ChrisPriceAB can you state the question concisely?
14:42:00 <lmcdasm> if the goal is testing results for Ostack with ODL then you need to reset the counters on all your FUEL testing
14:42:22 <trozet> why does ChrisPriceAB always use /me ? :)
14:42:33 * ChrisPriceAB is very shy
14:42:41 <trozet> haha
14:42:49 <trozet> but /me is for 3rd person context
14:42:52 <trozet> like this
14:42:54 * trozet runs to the store
14:43:14 <ChrisPriceAB> yeah looks like I ask myself questions...
14:43:21 <trozet> haha
14:43:45 <lmcdasm> so, Jose i think CHris is waiting for you to re-ask the question
14:44:00 <trozet> jose_lausuch: focuse for Arno is OPNFV which at the moment is ODL + O/S, and eventuall service function changing (SFC) enabled in ODL
14:44:01 <jose_lausuch> Solution A) 1) Deploy 2) Functest without ODL  3) Switch to ODL 4) test ODL
14:44:01 <jose_lausuch> Solution B) 1 ) Deploy 2) Switch to ODL) Test full Functest
14:44:12 <jose_lausuch> Solution B needs more work and cannot make it for Arno
14:44:25 <lmcdasm> * caveat -
14:44:50 <[1]JonasB> So realistically I think we have a situation for Fuel that we either run tempest with non ODL configuration + a limited ODL test and fix the rest for SR1, or we do not release fuel the 4'th.
14:45:04 <ChrisPriceAB> Can I ask why we can't run the tests the same way on both PODs?
14:45:19 <jose_lausuch> Daniel, shoot
14:45:24 <lmcdasm> my final comment  there is no defined exit criteria for relase - with this approach, i can say sure its done and then mark things as "bugs"
14:45:41 <lmcdasm> when / how are you drawing the line on what is acceptable for release and not?
14:45:56 <jose_lausuch> me?
14:46:01 <lmcdasm> Answer Chris - see above Tim explained it
14:46:18 <lmcdasm> as well that is what i have been trying to get infomraino for so I can make it the same and we can compare apples to apples
14:46:32 <lmcdasm> but seems this is not going to be possible.
14:47:27 <ChrisPriceAB> Ok, so tim explained that running OS non-HA will result in better results on POD1.  Is that a blocking issue?
14:48:12 <jose_lausuch> So we dont focus on testing Fuel HA?
14:48:24 <lmcdasm> well there was a bit more than that.. bsaically about the vxlan tunnel
14:48:37 * ChrisPriceAB is very confused because he understood we could not get Fuel HA working with ODL.
14:48:46 <lmcdasm> anyway - its moot.. the point being how you want to proceed at this point.  we have a month of testing on FUEL where we havent been testing the right objec
14:48:54 <lmcdasm> thats not true Chris -
14:49:16 <lmcdasm> so.. we can try to set the FUEL to the same configuration as FOREMAN
14:49:25 <lmcdasm> (once the POD2 is up and i can go pick the pieces that we are looking for)
14:49:27 * ChrisPriceAB by next week, but that we could get Fuel single instance & ODL working by next week.
14:49:35 <lmcdasm> well - please define working?
14:49:40 <lmcdasm> since we had it working for 4 months
14:49:54 <[1]JonasB> lmcdasm: we havent had month of testing - comfused :-/
14:49:55 <lmcdasm> where the scope wsa simply to create netowrking and see them reflected in ODL gui
14:50:00 <lmcdasm> now the scope had changed
14:50:11 <jose_lausuch> btw, deployment is completed
14:50:49 <jose_lausuch> lets talk after the meeting Daniel to get the needed information
14:50:52 <[1]JonasB> I need to run for a meeting
14:51:05 <[1]JonasB> #endmeeting