14:02:26 <fdegir> #startmeeting Cross Community CI 14:02:26 <collabot> Meeting started Wed Jan 11 14:02:26 2017 UTC. The chair is fdegir. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:26 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic. 14:02:26 <collabot> The meeting name has been set to 'cross_community_ci' 14:02:37 <fdegir> #topic Rollcall 14:03:06 <fdegir> please type in your name 14:03:10 <fdegir> #info Fatih Degirmenci 14:03:14 <hwoarang> #info Markos Chandras 14:03:20 <s_berg> #info Stefan Berg 14:03:32 <hw_wutianwei> #info Tianwei Wu 14:04:01 <fdegir> I hope yolanda can join as well 14:04:12 <yolanda> #info Yolanda Robla 14:04:13 <fdegir> first of all, welcome everyone and happy new year to you all 14:04:14 <yolanda> sorry 14:04:22 <hwoarang> happy new year 14:04:33 <hw_wutianwei> happy new year 14:04:40 <fdegir> #topic General Updates 14:05:08 <Julien-zte> #info Julien 14:05:23 <fdegir> as I mentioned 14:05:24 <fdegir> #info We have renamed this activity and expanded the scope to work on all Cross Community CI Activities together 14:05:29 <Julien-zte> happy new year 14:06:08 <fdegir> #info This means that OpenStack, ODL and future Cross Community CI activities will be driven under this as we can share stuff between these activities 14:06:24 <hwoarang> sounds good 14:06:41 <fdegir> and finally, I want to welcome s_berg to the team! 14:06:42 <Julien-zte> I have find several jobs related this have been created in releng 14:06:57 * s_berg waves 14:07:08 <fdegir> moving on to the next topic 14:07:10 <s_berg> Thanks Fatih. 14:07:20 <fdegir> #topic OpenStack Cross Community CI 14:07:44 <fdegir> #info Bifrost verification jobs were disabled during Intel lab migration 14:07:54 <fdegir> #info They've been enabled back yesterday 14:08:18 <hw_wutianwei> I have deployed the openstack using Openstack-ansble. but it is unstable 14:08:19 <fdegir> hwoarang: do you want to give info about the bifrost log uploads? 14:08:23 <Julien-zte> good news 14:08:36 <hwoarang> so yeah i have posted a patchset to upload logs to artifacts.opnfv.org 14:08:48 <hwoarang> this was requested during a review in the upstream openstack gerrit 14:08:55 <hw_wutianwei> you need try several times 14:08:59 <hwoarang> so please go and review it if you have a couple of minutes to spare 14:09:02 <hw_wutianwei> I also submit the patch in gerrit. people can use this patch to test. it is necessary to improve. 14:09:02 <fdegir> #info Patch to upload bifrost logs to artifacts.opnfv.org is up for review 14:09:17 <fdegir> #link https://gerrit.opnfv.org/gerrit/#/c/26831/ 14:09:51 <fdegir> #info Once we fix this part, I'll ping Bifrost PTL and try to get us voting rights for bifrost on OpenStack Gerrit 14:10:11 <fdegir> they already discussed this in ironic meeting but there were some questions from the rest of the OpenStack people 14:10:13 <jmorgan1> #info Jack Morgan 14:10:18 <hwoarang> sounds good 14:10:32 <fdegir> anyone wants to give an update about openstack-ansible? 14:10:40 <fdegir> perhaps hw_wutianwei ? 14:11:01 <hwoarang> personally, i have started looking at it, but suse support is somewhat missing so i guess i will have to do the suse port first :) 14:11:10 <hw_wutianwei> Do we have a plan when we can deploy the Openstack？ 14:11:21 <yolanda> i started to test yesterday, but i had problems because my host was centos 14:11:27 <yolanda> so had to adjust package install, network 14:11:33 <yolanda> still on the process of testing it 14:11:41 <fdegir> #info hwoarang started looking into openstack-ansible, but suse support is somewhat missing so he will have to do the suse port first 14:11:41 <hw_wutianwei> I just test the ubuntu 14:11:43 <Julien-zte> hi fdegir, we use try to deploy vms by bifrost for openstack-ansilbe 14:12:18 <Julien-zte> the builderimage only support 14.04 ? and openstack-ansible requires 16.04 for ubuntu? 14:12:21 <fdegir> hw_wutianwei: I ask the same quesiton to you as you made some progress I suppose 14:12:32 <Julien-zte> and the bifrost only deploy one nics for the vms. 14:12:53 <Julien-zte> shall we improve these settings with multiple nics? 14:12:55 <yolanda> Julien-zte, so bifrost works fine with xenial noe 14:12:56 <yolanda> now 14:12:57 <s_berg> I've tried successfully (so far) with Ubuntu 16.04 and the Newton build, looked to be a supported combo. 14:13:18 <fdegir> s_berg: I suppose it is virtual? 14:13:28 <hw_wutianwei> bifrost suppot the 16.04 14:13:34 <Julien-zte> ok, good news 14:13:50 <Julien-zte> any try for multiple nics with bifrost? 14:13:51 <s_berg> fdegir: Yes, virtual all-in-one-host. 14:14:14 <s_berg> Would love to try it out in a distributed setting (will look at doing that virtually as well to try it out) 14:14:18 <yolanda> Julien-zte, no support yet for multiple nics, but TheJulia (bifrost PTL) suggested we could create a raw network_info.json and we could pass it 14:14:25 <fdegir> #info s_berg tried openstack-ansible virtual all-in-one-host with Ubuntu 16.04 and the Newton build 14:14:44 <yolanda> as an alternative, hw_wutianwei was configuring all the networking using some ansible playbooks after deploying bifrost... 14:15:06 <Julien-zte> that's what we want to try. 14:15:24 <Julien-zte> hw_wutianwei, any progress has achieved? 14:15:30 <hw_wutianwei> it is seem ok, when I try. 14:16:04 <fdegir> hw_wutianwei: I suppose the patch you sent to releng is for running openstack-ansible? 14:16:06 <fdegir> https://gerrit.opnfv.org/gerrit/#/c/26865/ 14:16:18 <hw_wutianwei> yep 14:16:21 <fdegir> (haven't had time to look at the details) 14:16:30 <fdegir> great hw_wutianwei 14:16:39 <hw_wutianwei> and configuring all the networking of vms 14:16:55 <Julien-zte> that's great 14:16:57 <fdegir> I hope to give a try and send comments if I find any 14:17:08 <Julien-zte> I will pay more time in codes in releng about bifrost 14:17:15 <fdegir> #info hw_wutianwei sent a patch to releng for openstack-ansible. Please try and comment back 14:17:18 <fdegir> #link https://gerrit.opnfv.org/gerrit/#/c/26865/ 14:17:20 <Julien-zte> it will save our time 14:17:53 <fdegir> I think we are having some progress with openstack-ansible 14:18:03 <fdegir> which is really good 14:18:16 <yolanda> so far the comments i had for that review, same as other, is that it shall be isolated in a different folder 14:18:20 <yolanda> and don't break our current jobs 14:18:50 <yolanda> also support for centos/suse shall be needed, as we test on these hosts as well. Even if the vms are xenial, the deployment hosts can be other 14:19:03 <hwoarang> ideally it should be a separate job for now 14:19:08 <Julien-zte> the network settings' patch will be upstreamed in openstack? 14:19:20 <fdegir> yes, our aim was, is, and will be supporting the 3 OSes: ubuntu, centos, suse 14:19:25 <fdegir> like we are doing with bifrost 14:19:59 <fdegir> hw_wutianwei: I will put this as a comment but just to mention here as well 14:19:59 <Julien-zte> if new job required, add it in releng:) 14:20:15 <fdegir> hw_wutianwei: it would be good to have simple readme accompanying the code like bifrost 14:20:31 <hw_wutianwei> ok, i will add this 14:20:38 <fdegir> hw_wutianwei: so we can understand what it is the osa is doing 14:20:55 <Julien-zte> good 14:21:03 <fdegir> any more info/comment about openstack-ansible? 14:21:16 <hwoarang> i do 14:21:26 <fdegir> please go ahead hwoarang 14:21:31 <hw_wutianwei> i also will add a few comments to this script 14:21:35 <hwoarang> so what's the plan for puppet-infracloud? Yolada, are we still looking at both of them? 14:21:48 <hwoarang> *Yolanda 14:21:52 <fdegir> yolanda: ^ 14:22:22 <yolanda> infracloud is quite stopped at the moment 14:22:48 <yolanda> i guess first we need to evaluate if OSA is a valid option and we could go with that instead 14:23:05 <hwoarang> ok but i guess the final goal is to use one of them and not provide code for both 14:23:35 <yolanda> yes, i'd say so 14:23:38 <hwoarang> ok 14:23:40 <hwoarang> second question 14:23:51 <hwoarang> so we want to use OSA on top of bifrost 14:24:20 <hwoarang> how are we going to handle this in the CI? say upstream bifrost makes a change and our CI kicks in. Are we going to do a full bifrost+OSA cycle and provide feedback for that? 14:24:34 <hwoarang> if one of the two components is broken, then our feedback is somewhat useless 14:24:54 <yolanda> it will be same situation as bifrost + puppet-infracloud 14:25:03 <fdegir> what I think is 14:25:22 <fdegir> there are 2 types of CI activities here 14:25:39 <fdegir> first one is having bifrost + osa working for opnfv to deploy openstack from master 14:25:53 <fdegir> this will be done using the proven/working versions of bifrost + osa 14:26:07 <fdegir> second one is providing feedback to openstack bifrost + osa 14:26:12 <fdegir> and this could be done in a way that 14:26:31 <fdegir> if we get bifrost change, we should test whole chain with bifrost change + proven/working version of osa 14:26:33 <fdegir> and vice versa 14:26:38 <Julien-zte> yes, the openstack version is verified earlier than opnfv now does. 14:26:42 <fdegir> so we only change one of them at any given time 14:27:02 <fdegir> we need to keep track of what we verified in order to lock versions both for opnfv purposes 14:27:09 <fdegir> and for upstream patch verification 14:27:14 <hwoarang> oh so we plan to provide feedback for the combined solution 14:27:15 <hwoarang> ok 14:27:18 <Julien-zte> what will we do to "change one of them at any given time" 14:27:21 <fdegir> so we can always have working baseline for ourselves 14:27:38 <fdegir> and latest + working baseline for upstream verification 14:27:43 <fdegir> I hope this does make sense 14:27:59 <fdegir> perhaps I shouldn't guess to much but 14:28:13 <fdegir> if we imagine a job that does bifrost and osa in serial order 14:28:20 <fdegir> and if we are verifying bifrost only 14:28:24 <fdegir> osa one shouldn't vote 14:28:32 <fdegir> to upstream 14:28:59 <fdegir> but we will run it for our own purposes to see if that bifrost patch breaks our bifrost + osa working baseline if we intend to move to it 14:29:31 <hwoarang> yes but we could also provide feedback for the individual components too. so the bifrost jobs provides feedback, then we test osa on top and we provide feedback for that too. 14:29:49 <fdegir> hw_wutianwei: but our verificaiton is based on patchset-submitted 14:29:53 <hwoarang> anyway i was wondering because i guess it would be helpful especially for hw_wutianwei to structure the code accordingly 14:29:56 <fdegir> hwoarang: ^ 14:30:18 <fdegir> we will find it out soon 14:30:21 <hwoarang> ok 14:30:30 <fdegir> once hw_wutianwei patch gets in, we should directly create jobs for it 14:30:33 <fdegir> and make it run 14:31:05 <fdegir> and I still have the ap to bring up baremetal daily jobs for bifrost 14:31:11 <fdegir> which I need to look into 14:31:12 <Julien-zte> hi fdegir, if we want to work as a 3rd ci for openstack, I think we will prepare multiple slaves. for new in opnfv only is working? 14:31:26 <Julien-zte> too many updates in openstack 14:31:31 <fdegir> Julien-zte: we run verification for both for openstack and opnfv 14:31:42 <fdegir> Julien-zte: for bifrost 14:31:44 <fdegir> https://build.opnfv.org/ci/view/3rd%20Party%20CI/ 14:31:56 <fdegir> Julien-zte: when we come to the point where we have things up and running 14:32:00 <Julien-zte> I mean multiple slaves for the same job 14:32:03 <fdegir> and if we experience resource shortages 14:32:13 <fdegir> I know who to talk to 14:32:16 <fdegir> jmorgan1: ^ 14:32:27 <Julien-zte> ^_^ 14:32:34 <fdegir> Julien-zte: we can improve that by using VMs rather than baremetal for verification 14:32:42 <fdegir> as hwoarang does for suse 14:32:54 <fdegir> as we need to verify it on all OSes we support 14:33:01 <hwoarang> a somewhat big-ish VM should be able to run the bifrost job just fine 14:33:05 <Julien-zte> yah 14:33:18 <fdegir> hwoarang: can you type in the specs of the vm you created for suse? 14:33:22 <Julien-zte> you mentioned, intel pods come back now, fgedir 14:33:27 <fdegir> I might try doing that 14:33:39 <fdegir> for ubuntu and then centos perhaps 14:34:03 <fdegir> Julien-zte: right - we use intel-pod4 which has been back since last week 14:34:10 <hwoarang> 16 vcpus, 120G disk, 16G RAM 14:34:40 <fdegir> #info a vm with 16 vcpus, 120G disk, 16G RAM should be able to run bifrost 14:34:50 <fdegir> should we move on? 14:35:04 <hw_wutianwei> huawei also can provide virtual pod now 14:35:15 <fdegir> hw_wutianwei: moarrr hw 14:35:34 <fdegir> thanks hw_wutianwei, I'll contact you offline 14:35:50 <hw_wutianwei> ok 14:35:57 <hwoarang> lol 14:36:09 <fdegir> #topic ODL Cross Community CI status 14:36:21 <Julien-zte> hi fdegir, if please send me gpg key, I will send you a openvpn account for the test the timedelay 14:36:24 <fdegir> #info We also started running patchset verification for ODL netvirt 14:37:11 <fdegir> #info How this works is that; when a new patch is sent for ODL netvirt, we bring up OpenStack from snapshots in about 10 minutes, install netvirt, and run functest against it 14:37:37 <fdegir> #info Currently trozet is working on creating apex snapshots which will be incorporated into the job we created for this purpose 14:37:52 <fdegir> #link https://build.opnfv.org/ci/view/3rd%20Party%20CI/job/odl-netvirt-verify-virtual-master/ 14:38:13 <fdegir> we will have more progress with this coming days/weeks 14:38:15 <jmorgan1> fdegir: yes? 14:38:28 <fdegir> jmorgan1: we put an ap on you 14:38:48 <fdegir> that was it about odl patch verification 14:38:57 <fdegir> last topic is 14:39:02 <fdegir> #topic Infra needs/updates 14:39:17 <fdegir> #info The Cross Community CI resources came back online 14:39:23 <fdegir> #info Thanks jmorgan1 for this 14:39:31 <fdegir> #info The page is updated with latest assignments 14:39:40 <fdegir> #link https://wiki.opnfv.org/display/pharos/Intel+Pod4 14:39:58 <fdegir> I suppose everyone has somewhere to work on 14:40:05 <jmorgan1> fdegir: we don't have any more resources right now in our lab 14:40:07 <fdegir> if not, please contact me so I try to find machines for you 14:40:17 <fdegir> jmorgan1: I was joking :) 14:40:42 <fdegir> #action fdegir to contact hw_wutianwei for vpods from Huawei 14:40:49 <fdegir> #topic AOB 14:40:56 <fdegir> anyone wants to bring up anything? 14:41:43 <fdegir> I suppose not 14:41:53 <fdegir> thank you all and talk to you in 2 weeks 14:41:57 <fdegir> #endmeeting