13:57:46 <[1]JonasB> #startmeeting OPNFV Fuel status No 11 13:57:46 <collabot> Meeting started Tue May 5 13:57:46 2015 UTC. The chair is [1]JonasB. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:57:46 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic. 13:57:46 <collabot> The meeting name has been set to 'opnfv_fuel_status_no_11' 13:58:02 <[1]JonasB> #info Jonas Bjurel 13:58:12 <mskalski> #info Michal Skalski 13:59:00 <fdegir> #info Fatih Degirmenci 13:59:30 <zilard> #info Szilard Cserey 13:59:50 <jgasparakis> #info Joseph Gasparakis 14:01:53 <[1]JonasB> Lets see if we can trigger mr Berg Jabber allert with the golden word: stefan_berg 14:02:37 <stefan_berg> Yes! 14:02:40 <stefan_berg> Present! 14:02:44 <stefan_berg> Sorry, chatting with Dan. :) 14:02:51 <stefan_berg> #info Stefan Berg 14:02:55 <lmcdasm> #info daniel 14:03:09 <[1]JonasB> haha 14:03:26 <lmcdasm> sorry.. we was doing 'work' 14:03:50 <[1]JonasB> fdegir: do you have th goto meeting number for TSC, the one I've got doesnt seem to be the right one? 14:04:36 <fdegir> [1]JonasB: I use gotomeeting 14:04:52 <fdegir> https://global.gotomeeting.com/join/798898261 14:05:04 <[1]JonasB> fdegit: thx 14:05:12 <fdegir> np 14:05:15 <[1]JonasB> #topic Auto deployment 14:05:32 <[1]JonasB> So stefan, news a bout latest prototype? 14:05:37 <stefan_berg> Yes. 14:05:57 <stefan_berg> #info Hybrid VM Fuel/IPMI nodes DHA is completed. 14:06:14 <stefan_berg> #info Installation works, but failed due to changes needed to the network config. 14:06:31 <stefan_berg> #info Will retry again after Daniel did his magic just minutes ago. 14:06:58 <stefan_berg> #info Fun thing was that I just could use the same DEA.yaml and verify basic functionality with libvirt to make sure there was no fundamental issues. :) 14:07:18 <[1]JonasB> Cool 14:07:26 <stefan_berg> #info The IPMI DHA is in Gerrit. 14:07:39 <stefan_berg> #info Will add some documentation, but basically a go. 14:07:48 <[1]JonasB> stefan_berg: should it be merged? 14:07:55 <stefan_berg> #info Looking for volunteers to try it out. 14:08:10 <[1]JonasB> Ok 14:08:17 <stefan_berg> No, let's wait until I get proof that it installs clean with networking and get docs in place. 14:08:35 <[1]JonasB> lmcdasm: Any comments on the networking issue/resolution? 14:09:30 <lmcdasm> #info - issue was related to vlan misconfig 14:09:48 <[1]JonasB> lmcdasm: no bigg isue thoug? 14:09:51 <lmcdasm> #info in that i didnt assign VLANs for Stefan stuff (keep in mind we have about 10+ envs so the switch is a bit messy) 14:09:56 <lmcdasm> #info no "tech" issues nope 14:10:06 <[1]JonasB> Great! 14:10:43 <[1]JonasB> #info Let's move on to the status of Szilards Autodeployment production code 14:10:50 <lmcdasm> brb... 14:10:50 <zilard> #info I am refactoring the Python AutoDeploy according to prototype V2 14:11:01 <zilard> #info development still in progress 14:11:33 <zilard> #info that's all 14:11:47 <[1]JonasB> Zsilard: Thanks, keep on:-) 14:12:06 <zilard> thanks :) 14:12:33 <stefan_berg> We've also discussed that we need to do a few changes of the DEA.yaml to be more generic, we'll figure that out moving forward. 14:12:44 <lmcdasm> #info - my little piece 14:12:54 <[1]JonasB> lmcdasm: go on 14:13:25 <lmcdasm> #info - esxi adapter using OVFTOOL worked based on last weeks proto - once the merge is done ill refactor according to the updates (however, the IMS core takes precendence).. at the same time - ATT and others are starting to use Nested ESxi envs (docs) 14:13:35 <lmcdasm> #info - so there is some demand for it :) 14:13:38 <lmcdasm> (my two cents) 14:13:59 <[1]JonasB> lmcdasm: thanks 14:14:22 <zilard> thanks for info Daniel 14:14:31 <[1]JonasB> #topic LF Lab POD1 14:15:04 <jgasparakis> This is me I suppose? :-) 14:15:16 <[1]JonasB> #info So as we're in principal ready with the autodeploy, we will move on to LF-Lab 14:16:09 <[1]JonasB> #info Peter B and Joseph have kindly offered their hands to help 14:16:29 <[1]JonasB> #info We have a separate meeting to kick that off 14:17:55 <[1]JonasB> I have invited: joseph, stefan, Peter B, Fatih - anyone else that wants/needs to attend? 14:18:03 <zilard> me too :) 14:19:06 <[1]JonasB> zsilard: Ok 14:19:10 <zilard> thx 14:19:59 * stefan_berg wonders who stole szilards "s" 14:20:23 <zilard> :) I left it out, for simplicty 14:20:42 <stefan_berg> No theft involved, great! :) 14:20:49 <[1]JonasB> #topic Func test 14:21:07 <[1]JonasB> Jose or Morgan - any updates? 14:24:58 <morgan_orange> just joined 14:25:05 <morgan_orange> TSC meeting 14:26:22 <[1]JonasB> Two meetings in parallel :-) 14:27:30 <[1]JonasB> Lets move on 14:28:08 <[1]JonasB> #info No representative available to talk on Functest. 14:28:10 <morgan_orange> #info first tests performed on LF POD 2 5foreman) 14:28:27 <jose_lausuch> #info jose Lausuch 14:28:33 <morgan_orange> #link https://wiki.opnfv.org/opnfv_functional_testing table showing the status updated 14:28:34 <jose_lausuch> (just finished my other meeting) 14:29:31 <jose_lausuch> morgan go ahead:) 14:29:46 <morgan_orange> #info goal Tempest smoke tests shall be fully OK 14:30:21 <morgan_orange> #info jose_lausuch created a new script to pre deploy the tooling and the scenario, to be tested on LF POD 14:30:54 <[1]JonasB> Should we have have a focused investigation on the failing testcases? 14:31:23 <morgan_orange> jose_lausuch: and daniel are looking at it (on E/// POD 20 failures on 108 tests) 14:31:40 <[1]JonasB> super, can you info that 14:32:07 <morgan_orange> #info jose_lausuch and daniel investigation on Tempest failures on E/// POD (20 Ko on 108 tests) 14:32:22 <morgan_orange> #info BGS team to be involved once auto installation OK on LF PODs 14:32:24 <[1]JonasB> Anything more on functest for today? 14:32:33 <lmcdasm> sure.. i think you guys were doing stuff yesterday so im around today / whenever you want me to look at stuff 14:32:36 <lmcdasm> just ping me : 14:32:48 <[1]JonasB> lmcdasm: thanks! 14:32:58 <jose_lausuch> thanks 14:33:16 <morgan_orange> jose_lausuch: maybe we could save the json report somewhere to share the results 14:33:30 <morgan_orange> on LF POD, I mentioned the path to trozet but we could also put them on the wiki 14:33:30 <jose_lausuch> I can do that, I have all of them 14:33:34 <lmcdasm> sure.. you pastebin or you can drop it on any of the servers and i can fetch it 14:33:37 <lmcdasm> or you can put i tup on our public server 14:33:44 <lmcdasm> anyway you guys like 14:34:10 <jose_lausuch> I'll ping you later to do this 14:34:15 <trozet> morgan_orange: can we import them to jenkins 14:34:18 <lmcdasm> v.good 14:34:21 <trozet> they should be part of the results 14:34:43 <morgan_orange> we have to replug CI on functest on LF POD 14:35:01 <[1]JonasB> Okay can we continue? 14:35:51 <[1]JonasB> The only thing left for today was to list all things we know we need to do before release (apart from LF-lab and Functest). 14:36:12 <[1]JonasB> #topic remaining stuff needing attention before release 14:36:24 <[1]JonasB> #info Polishing documentation 14:36:32 <lmcdasm> #info - question - is it clear / solid the established test cases for passing release (from other meeting) 14:36:33 <lmcdasm> ? 14:36:52 <stefan_berg> #info Possibly some metadata adjustments of the DEA to support Ceph/Cinder nodes - currently not tested. Minor work. 14:36:53 <lmcdasm> i know we say "whatever functest has", but that could be big / small or somewhere in between.. 14:37:14 <[1]JonasB> lmcdasm: I think Morgan needs to answer that 14:37:17 <stefan_berg> #info More DEA: A more robust passing of the pre-deploy NTP config. 14:37:44 <[1]JonasB> #info changing default OpenStack uid/pwd 14:37:44 <morgan_orange> #info mty view http://artifacts.opnfv.org/functest/docs/functest.html was 14:37:54 <morgan_orange> #Rally Tempest smoke test fully OK 14:38:17 <morgan_orange> #Rally bench => as many tests as possible but can remove some 14:38:34 <morgan_orange> #ODL => pbandzi 14:38:42 <[1]JonasB> All: at current ODL is manually started after installation, should we stick with that for the release? 14:39:06 <lmcdasm> #info - yes, cause there is some ODL issue when you dont have routes out that will reqire some more work 14:39:10 <stefan_berg> Depends on how much extra configuration is needed to make it play. 14:39:26 <lmcdasm> #info - i have got the KARAF release guy to help us out - but i dont think we can a) do alot with ODL (since its just overlay) and its a visual check 14:39:46 <lmcdasm> #info - however, there are two simple tests (check for port listening) and check for login avbailable (docker interactive) that we could add 14:39:50 <lmcdasm> and that would simple test 14:39:55 <lmcdasm> what do you guys think? 14:39:57 <[1]JonasB> #info at current ODL is manually started after installation, we will stick with that for the release! 14:40:24 <stefan_berg> Let's discuss later on what ODL information needs to be persisted/configured for auto deployment, but that's another meeting. 14:40:28 <lmcdasm> v good 14:40:39 <[1]JonasB> Ok - agree 14:40:39 <lmcdasm> cause it needs a bit of work still \ 14:41:26 <morgan_orange> #info we have to ask pbandzi but Peter was able to start manually its ODL test suite on Intel POD, he planned to test them on LF POD today and was confident. 14:41:42 <pbandzi> hi 14:41:56 <pbandzi> odl can be started automaticaly also 14:41:58 <[1]JonasB> #info rebase to szilards auto-deployment code 14:42:48 <[1]JonasB> Anything more to add? 14:43:00 <pbandzi> right now i see that some test was not passing to me on lf pod2 but that might be caused by wrong paramters poassed by me to robotsuite 14:43:34 <[1]JonasB> We should soon start discussion on a SR1, SR2 releases! 14:44:26 <[1]JonasB> Anything more? 14:44:46 <[1]JonasB> 1 14:44:54 <jgasparakis> I have a question 14:44:58 <[1]JonasB> Yep 14:45:14 <jgasparakis> Has anybody tried to build the fuel iso behind a proxy? 14:45:33 <stefan_berg> jgasparakis: I don't think so. 14:45:49 <stefan_berg> There's no proxy support in there afaik. 14:45:58 <jgasparakis> Yeah, I know :-) 14:46:06 <jgasparakis> I must be the first then... 14:46:22 <jgasparakis> I tried it yesterday and failed miserably. 14:46:28 <[1]JonasB> jgasparakis: Shouldnt be a problem, sinse technically an installation doesnt need internet the way Foreman do 14:46:40 <stefan_berg> Tried setting http_proxy and https_proxy without luck? 14:46:50 <jgasparakis> Yes 14:46:55 <[1]JonasB> jgasparakis: How did it fail? 14:47:07 <lmcdasm> sorry - question? 14:47:18 <lmcdasm> the Foreman doesnt need internet anymore (that was corrected)? 14:47:55 <lmcdasm> (hey Joseph) - i have done a build behind proxy and it really doesnt work without a couple port forwards (depending on your firewall) 14:48:04 <lmcdasm> im happy to give you an hand hacking around if you need 14:48:04 <lmcdasm> :) 14:48:35 <jgasparakis> Ok, several points of input and some questions, let me try to answer a few... 14:49:41 <jgasparakis> JonasB: It was failing inside the containers when it was doing apt-get's, wget's and git clone's. And some other network related stuff... 14:50:09 <[1]JonasB> jgasparakis: But that is build, not deploy 14:50:14 <jgasparakis> I had to add some ENVs and RUNs in the dockerfiles 14:50:47 <stefan_berg> Aaah, makes sense, proxy settings wouldn't be propagated to the container to start with. 14:50:52 <jgasparakis> JonasB: yes, I understand. I wanted to build the iso in order to do the deployment. Isn't this the right way to start 14:51:06 <jgasparakis> stefan_berg: exactly! 14:51:21 <[1]JonasB> jgasparakis: Yes, that is valid too, just to be on the same page:-) 14:52:12 <jgasparakis> So a question to this audience: do we care to fix this for Arno? If not I will find another way to get the iso. 14:52:15 <[1]JonasB> stefan_berg: SHould we try to pass proxy settings to docker? 14:53:03 <stefan_berg> OK, two questions there. Personally I don't think this has top priority for Arno (as compared to auto deployment and testing) but it would be good to have in there if possible. 14:53:13 <[1]JonasB> #info If possible, fix build issues if behind a proxy 14:53:40 <stefan_berg> But yes, we definately should make sure that all proxy settings set at the host level are propagated to the container, and fix any remaining proxy issues. 14:54:07 <stefan_berg> Hopefully not that hard. 14:54:08 <jgasparakis> Ok, I have hacked some thigs around and the build progresses has moved ahead quite a bit. 14:54:28 <jgasparakis> But there are still some issues and I need to make priority call as to where I will focus. 14:54:33 <stefan_berg> jgasparakis: I usually start the container interactively and then do a "make iso" to test things out. 14:55:17 <stefan_berg> On my small priority list for the near future is: 14:55:30 <stefan_berg> 1. Get manual deployment done in the LF lab. 14:55:37 <[1]JonasB> jgasparakis: What Stefan just mentioned is a great way to penetrate the darknes of the build inside the container 14:55:42 <stefan_berg> 2. Verify sanity (of deployment!). :) 14:55:58 <jgasparakis> stefan_berg: Yes, I was doing this, in order to add things to Dockerfiles. 14:56:15 <stefan_berg> 3. Create a template for the deployment (it's a tool in the auto deploy suite) 14:56:30 <stefan_berg> 4. Verify auto deploy with these settings. 14:56:42 <jgasparakis> Maybe we should capture these priorities in the minutes? 14:56:47 <stefan_berg> Then I'm happy to also jump onto fixing any remaining proxy issues with building. 14:57:18 <[1]JonasB> Lets discuss that separately in the other call we have 14:57:20 <jgasparakis> Ok, cool. Let's talk about this more if needed in the autodeploy meeting. 14:57:26 <[1]JonasB> So we need to wrap up now 14:57:27 * fdegir stefan_berg, 5th step is jenkins I suppose 14:57:46 <[1]JonasB> Going once 14:57:58 <[1]JonasB> Going twice 14:57:58 <stefan_berg> fdegir: Yes! Jenkins is always #1 in my book, but #5 in this list. :) 14:58:02 <fdegir> :D 14:58:05 <fdegir> thanks 14:58:21 <[1]JonasB> Thanks everyone, sametime tomorrow! 14:58:27 <[1]JonasB> #endmeeting