14:00:43 <fdegir> #startmeeting Cross Community CI 14:00:43 <collabot> Meeting started Wed Nov 15 14:00:43 2017 UTC. The chair is fdegir. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:43 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic. 14:00:43 <collabot> The meeting name has been set to 'cross_community_ci' 14:00:48 <fdegir> #topic Rollcall 14:00:56 <fdegir> lets have a quick status update 14:01:15 <electrocucaracha> #info Victor Morales 14:01:17 <hw_wutianwei> #info tianwei Wu 14:01:19 <David_Orange> #info David Blaisonneau 14:01:19 <mbuil> #info Manuel Buil 14:01:29 <trinaths> #info Trinath Somanchi - NXP 14:01:35 <joekidder> #info Joe Kidder 14:01:40 <fdegir> the agenda is pretty same as the last few weeks 14:01:45 <fdegir> #link https://etherpad.opnfv.org/p/xci-meetings 14:02:01 <fdegir> starting with scenario status 14:02:12 <fdegir> #topic Scenario Status: os-odl-sfc 14:02:36 <fdegir> mbuil: any changes since the last time? 14:02:52 <mbuil> yes, should I info myself or better tell you and you do the filtering? 14:03:02 <fdegir> mardim: please info in 14:03:14 <fdegir> mbuil: ^ 14:03:39 <mbuil> fdegir 14:03:42 <mbuil> fdegir: ok 14:03:58 <ttallgren> #info TapioT 14:04:29 <epalper> #info Periyasamy Palanisamy 14:04:37 <mbuil> #info We are doing two things. 1st thing is upstreaming features to os_neutron 14:04:48 <mbuil> #link https://review.openstack.org/#/c/517259/ 14:05:01 <mbuil> #link https://review.openstack.org/#/c/510909/ 14:05:46 <mbuil> #info 2nd thing, we are trying to make it work with master. Unfortunately, we are finding several issues when trying xci with simple ODL to do L2 and L3 14:06:14 <fdegir> mbuil: any guess about where those issues are coming from? 14:06:21 <fdegir> mbuil: osa itself or xci stuff? 14:06:27 <mbuil> #info It deployes but things don't work. That is a consequence of not using tempest anymore :( 14:06:42 <mardim> #info Dimitrios Markou 14:07:38 <durschatz> #info Dave Urschatz 14:07:45 <mbuil> fdegir: one issue came from xci, three issues came from os_neutron and another issue I suspect is coming from glance openstack (not able to create a image ==> https://hastebin.com/ewufafehaq.vbs) 14:08:13 <fdegir> mbuil: ok 14:08:16 <fdegir> I'll info this in 14:08:23 <fdegir> #info one issue came from xci, three issues came from os_neutron and another issue I suspect is coming from glance openstack (not able to create a image ==> https://hastebin.com/ewufafehaq.vbs) 14:08:42 <mbuil> before we start running xci with scenarios we need to be careful because as we don't use tempest, there might be bugs when doing standard cloud operations 14:08:49 <fdegir> about tempest; we can enable it back 14:09:13 <hwoarang> #info Markos Chandras 14:09:22 <fdegir> it was excluded due to expiration of cirros dns record 14:09:47 <mbuil> fdegir: I would like that :) 14:09:58 <fdegir> actioning myself to try tempest locally and enable it if it works 14:10:11 <fdegir> #action fdegir to try tempest and enable it 14:10:20 <fdegir> mbuil: anything else? 14:10:24 <mardim> Also I want to add here that i get an error in ZUUl which I cannot replicate locally 14:10:54 <mardim> IS related to the linux headers which are essential for the ovs-nsh 14:10:57 <mardim> installation 14:11:03 <mbuil> fdegir: nothing else from my side 14:11:04 <fdegir> mardim: please paste link to that so we capture it 14:11:13 <mardim> if anyone has any idea why this is happenning 14:11:17 <mardim> please tell me 14:11:24 <mardim> #link https://hastebin.com/omonutovat.sm 14:11:42 <fdegir> mardim: we can take that after the meeting 14:11:52 <mardim> fdegir, sire thanks 14:11:55 <mardim> sure 14:12:03 <fdegir> moving on to the next scenario 14:12:15 <mardim> fdegir, Also I want to add something more 14:12:18 <fdegir> #topic Scenario Status: os-nosdn-ovs 14:12:39 <fdegir> mardim: please info in while we wait epalper electrocucaracha 14:12:50 <mardim> fdegir, I have also this patch for proper testing of ODL in OSA 14:13:11 <mardim> #link https://review.openstack.org/#/c/518964/ 14:13:15 <mardim> that's all 14:13:17 <mardim> thanks :) 14:13:18 <fdegir> thx mardim 14:13:30 <fdegir> electrocucaracha: epalper: anything to say about os-nosdn-ovs? 14:13:55 <fdegir> I see the patch is (almost) ready to go in 14:13:57 <fdegir> #link https://gerrit.opnfv.org/gerrit/#/c/43447/ 14:13:57 <electrocucaracha> fdegir: ? 14:14:08 <epalper> #info review for this scenario is at https://gerrit.opnfv.org/gerrit/#/c/43447/ 14:14:25 <fdegir> electrocucaracha: sorry - I thought you are also looking into ovs 14:14:34 <epalper> # there is a dependent review https://gerrit.opnfv.org/gerrit/#/c/46859/ 14:14:57 <epalper> #info i have tested this scenario locally and it works 14:15:08 <electrocucaracha> fdegir: yes, that's one thing that I have in my plate, but I'm still dealing with aio behind proxy :S 14:15:10 <fdegir> epalper: I see a comment from hw_wutianwei in first patch 14:15:36 <hw_wutianwei> fdegir: I just suggest to do that 14:15:47 <durschatz> #info there is a dependent review https://gerrit.opnfv.org/gerrit/#/c/46859/ for above info 14:15:47 <fdegir> epalper: you can perhaps address/respond to that 14:16:00 <epalper> fdegir: sure 14:16:32 <fdegir> thx for infoing in durschatz 14:16:44 <fdegir> before we move to the next scenario, I want to take a short discussion here 14:16:48 <fdegir> about ovs and ceph 14:17:11 <fdegir> ceph change has been merged 14:17:19 <fdegir> thanks to hw_wutianwei and anyone else contributed to that 14:17:29 <fdegir> the thing is, opnfv uses ovs and ceph by default 14:17:40 <fdegir> and these are part of os-nosdn-nofeature scenario 14:18:13 <fdegir> I think we need to combine these two and push it as os-nosdn-nofeature scenario as releng-xci/xci/scenarios/os-nosdn-nofeature 14:18:22 <fdegir> any comments/thoughts? 14:19:23 <fdegir> any objections? 14:19:30 <electrocucaracha> makes sense 14:19:33 <hw_wutianwei> after ovs finish, we can combine these to os-nosdn-nofeature scenario 14:19:53 <fdegir> hw_wutianwei: yes, that's the way but just want to ensure we are all on the same page 14:20:32 <fdegir> #info Once ovs integration is done, it will be combined together with ceph under the scenario os-nosdn-nofeature 14:20:49 <fdegir> we will still need vanilla osa for upstream verification which needs to be handled separately 14:20:54 <tinatsou> #info Tina Tsou 14:21:00 <fdegir> moving on to 14:21:05 <fdegir> #topic Scenario Status: os-odl-nofeature 14:21:14 <fdegir> epalper: is it you again? 14:22:13 <epalper> #info I'm testing https://gerrit.opnfv.org/gerrit/#/c/39239/ again to look for any changes required at openstack-user-config.yml file 14:22:56 <mbuil> epalper: when deploying xci + ODL in master, we have a problem because the ODL service in haproxy is conflicting with the repo service. I created this patch to fix it: https://review.openstack.org/#/c/519661/. There is one thing though which I don't get, why do we have two ports of ODL in haproxy? The 8080 I guess is used by neutron when using ODL in HA mode but the 8181? As far as I know, there is no service using that port, right? 14:22:56 <fdegir> thx epalper 14:24:09 <fdegir> #info when deploying xci + ODL in master, there is a problem because the ODL service in haproxy is conflicting with the repo service. A patch was proposed to fix it: https://review.openstack.org/#/c/519661/ 14:24:37 <fdegir> mbuil: you can perhaps take it after the meeting so we don't keep others waiting 14:24:52 <mbuil> fdegir: ok 14:24:54 <fdegir> #topic Scenario Status: kubernetes in XCI 14:25:03 <fdegir> hw_wutianwei: I think you took it over from s3wong 14:25:09 <fdegir> any updates? 14:25:10 <hw_wutianwei> fdegir: yep 14:25:34 <hw_wutianwei> i upload a patchset according stephen's, and it work now with aio. 14:26:01 <hw_wutianwei> #link https://gerrit.opnfv.org/gerrit/#/c/46153/ 14:26:11 <fdegir> #info k8s works with aio now, the patch is under review 14:26:39 <fdegir> glad to hear 14:26:39 <hw_wutianwei> fdegir: yep, I hope you can give more suggestion 14:26:57 <hw_wutianwei> and I will improve that 14:27:01 <fdegir> this brings a question now 14:27:12 <fdegir> our CI verification is for openstack and we have nothing for k8s 14:27:25 <fdegir> we need to add a mechanism to allow scenario based verification 14:27:34 <fdegir> rather than default/vanilla osa 14:27:53 <fdegir> hw_wutianwei: we can work on this together perhaps and adapt jobs accordingly 14:28:14 <hw_wutianwei> fdegir: ok 14:28:17 <fdegir> #info CI for XCI needs to be adapted, enabling scenario based patchset verification 14:28:37 <fdegir> we can talk about the details coming days 14:28:50 <fdegir> and everyone is welcomed to review/provide suggestions obviously 14:28:55 <hw_wutianwei> fdegir: and i aslo need add other scenarios such as ha noha in k8s later 14:29:05 <fdegir> hw_wutianwei: +1 14:29:37 <fdegir> #info ha and noha will be added for k8s later on 14:29:42 <fdegir> moving to congress 14:29:46 <hw_wutianwei> fdegir: I have one quetion 14:29:54 <fdegir> hw_wutianwei: go ahead 14:31:03 <hw_wutianwei> fdegir, David_Orange: I found there is a patch about install K8S with rancher 14:31:20 <hw_wutianwei> fdegir: and I use kubespray 14:31:37 <David_Orange> hw_wutianwei: yes, it the code i was talking last week 14:31:44 <hw_wutianwei> do we need support both of them 14:32:06 <David_Orange> hw_wutianwei: extracted from what we have here in Lannion. 14:32:42 <fdegir> maybe I can share what I think about this (and similar things) 14:32:53 <David_Orange> hw_wutianwei: no, as i said to you last week i share only of needed, i can close the patch, the important is to have something working 14:33:58 <hw_wutianwei> David_Orange: ok 14:34:05 <fdegir> when we talk about XCI, we mainly talk about providing feedback with the toolset we picked and framework we are developing 14:34:29 <fdegir> another important aspect of XCI is to give people opportunity to try things out, experiment with them and come up with different ways of doing things 14:35:21 <fdegir> we of course have a framework to fit in 14:35:37 <fdegir> and as long as those new things that come to XCI fulfil what is required by the framework, I am fine with it personally 14:36:24 <fdegir> but one thing to highlight here is that, whoever comes up with those, they need to ensure that stays in the framework and carried by them until it is accepted by the rest of the XCI and the OPNFV as a whole 14:36:44 <fdegir> so, please continue bringing in new things and share them with the rest 14:37:15 <hw_wutianwei> fdegir, David_Orange: thank you to make it clear 14:37:25 <fdegir> now congress 14:37:32 <fdegir> #topic Feature Status: Congress 14:37:36 <fdegir> Taseer: are you with us? 14:37:38 <Taseer> ye 14:37:40 <Taseer> yes 14:37:52 <Taseer> role has already been merged 14:38:02 <fdegir> #info Congress role has been merged upstream 14:38:16 <Taseer> #link github.com/openstack/openstack-ansible-os_congress 14:38:24 <Taseer> but the patch in OSA has not 14:38:44 <fdegir> Taseer: link please 14:39:01 <Taseer> okay 14:39:17 <Taseer> #link https://review.openstack.org/#/c/503971/ 14:39:52 <Taseer> evrardjp commented something about an experimental job. 14:40:08 <fdegir> Taseer: it looks good 14:40:11 <Taseer> but looks like he is not on work this week 14:40:37 <fdegir> I mean no objection to patch itself but rather having scenario and the job 14:40:52 <fdegir> Taseer: he was at openstack summit so it might take few days until he recovers 14:41:03 <Taseer> okay 14:41:06 <fdegir> thanks Taseer 14:41:15 <fdegir> hw_wutianwei: skipping ceph as it's done already 14:41:18 <Taseer> fdegir: you are welcome ! 14:41:22 <hw_wutianwei> fdegir: ok 14:41:33 <fdegir> #topic Improving Stability 14:41:52 <fdegir> so hwoarang and David_Orange have been talking about fixing stability issues 14:42:04 <fdegir> hwoarang: David_Orange: can you please summarize it for the rest? 14:42:34 <hwoarang> i will let David_Orange summarize since he took over stuff this week as i am busy with other things 14:43:01 <David_Orange> fdegir: stability you mean the bifrost from PDF ? 14:43:10 <fdegir> David_Orange: yes and moving bifrost into vm 14:43:24 <David_Orange> fdegir: ok 14:44:04 <David_Orange> i am reading the actual code to see how things works for now, to be sure i do not misse something 14:44:31 <David_Orange> i try to see how all environment variables works 14:45:08 <David_Orange> and will probably remove many of them 14:45:14 <mbuil> what means PDF in this context? 14:45:28 <David_Orange> but they will be set in the I/PDF 14:45:47 <David_Orange> mbuil: the description of the pod 14:45:50 <David_Orange> https://gerrit.opnfv.org/gerrit/#/c/46493/5/xci/file/ha/vpod-pdf.yaml 14:46:09 <mbuil> David_Orange: thanks 14:46:30 <David_Orange> to have one code to rule all cases: aio/ha/mini/noha + vm/baremetal 14:47:36 <fdegir> maybe I can add a more general summary to this 14:47:42 <David_Orange> i will merge my code and the actual to do as it is done now for OSA: generate config/generate bifrost inventory/run bifrost vanilla playbooks 14:48:05 <fdegir> #info What we mean with "improving stability" in this context is that, we want to isolate ourselves from the host as much as possible, ensuring when someone (users/developers/CI) attempts to use XCI, things work smoothly due to less number of things that might conflict with each other 14:48:56 <fdegir> #info Another aspect of this work is to increase the reuse by splitting node specific bifrost stuff from the actual stack/scenario installation so whatever you may want to install will work with the bifrost pieces we have 14:49:06 <David_Orange> so the idea is to push bifrost in a VM 14:49:53 <ttallgren> Container? 14:49:55 <fdegir> #info Incorporating PDF/IDF is important for ensuring we can run on any (type of) POD which has its PDF available 14:50:06 <mbuil> ok and use an extra level of nested virtualization 14:50:08 <electrocucaracha> is that going to prevent the installation of bifrost stuff in the host? 14:50:14 <hwoarang> hopefully 14:50:28 <fdegir> mbuil: no extra nesting 14:50:37 <David_Orange> ttallgren: it is made by kolla 14:50:46 <fdegir> node enrollment/deployment will be driven from opnfv vm rather than host 14:50:57 <hwoarang> besically bifrost will only take care of enrollement and provisioning. not with creating the VMs itself 14:51:02 <fdegir> and the target nodes will be on the same level as opnfv vm - same as today 14:51:16 <David_Orange> ttallgren: but for now, we reuse the opnfv host 14:51:17 <hwoarang> mbuil: no functional changes from where you stand 14:51:20 * electrocucaracha nice 14:51:35 <fdegir> ttallgren: we have plans to look at kolla and possibility to get rid of opnfv vm by moving those pieces into container 14:51:40 <fdegir> ttallgren: but one step at a time 14:51:41 <hwoarang> in the end you will get the same X VMs like today 14:51:47 <mbuil> I see, thanks 14:52:07 <fdegir> ttallgren: our concern today is isolation, splitting bifrost from overall process, and PDF/IDF 14:52:21 <electrocucaracha> fdegir: in that future, do we have plans to support kolla and osa? 14:52:34 <fdegir> electrocucaracha: that's a possibility 14:52:52 <fdegir> we need to move on 14:53:11 <fdegir> we can talk about this and other topics after the meeting 14:53:19 <fdegir> #topic Zuulv3 Prototype 14:53:57 <fdegir> #info We had a discussion with OpenStack Infra last week to start a prototype in OPNFV to try things that are not tried by OpenStack Infra 14:54:13 <fdegir> #info The prototype will be limited in scope initially and the details can be seen on the link below 14:54:21 <fdegir> #link https://etherpad.opnfv.org/p/opnfv-zuul-prototype 14:54:41 <fdegir> #info Everyone is welcomed to share their thoughts on the scope of the prototype and help us to get it work 14:55:09 <fdegir> #topic Hardware Availability 14:55:34 <fdegir> #info I have been told that we will not get second POD for XCI so we have an issue with hardware availability now 14:55:52 <fdegir> #info The only POD we have will be configured and mainly dedicated for CI 14:56:07 <fdegir> if anyone is in urgent need of hardware, please reach out to me and we see what we can do 14:56:52 <fdegir> and we appreciate if anyone has a machine to provide, especially to XCI developers 14:57:06 <fdegir> again, ping me if that's the case 14:57:10 <fdegir> the last topic is 14:57:22 <fdegir> #topic Working with other communities 14:57:41 <mbuil> are Intel PoDs back in service? 14:57:52 <fdegir> mbuil: pod20 has been retired 14:58:04 <fdegir> mbuil: pod16 is the only pod we have - which is the CI POD I mentioned 14:58:22 <fdegir> #info We are in contact with DPDK and CNCF regarding XCI and more info will be shared once we have it 14:58:22 <mbuil> ok, retired without replacement I suspect :( 14:58:40 <fdegir> #info We are also talking to OpenStack OpenLab people regarding possible collaboration 14:58:50 <fdegir> #info More info to follow when we have it 14:58:59 <fdegir> #topic AoB 14:59:08 <fdegir> so, 1 minute for any last minute topic 14:59:10 <durschatz> fdegir: I have another set of students coming in January and want to make sure I can successfully deploy os-nosdn-nofeature, have access to horizon and create VMs on that scenario. When is the best time for me to chime in and test the new stability changes? My hardware suply is a bit thin now also. 14:59:35 <fdegir> durschatz: once David_Orange fixes what he is working on, you can give it a try 14:59:44 <fdegir> durschatz: it will have pre-queens 14:59:46 <durschatz> :-) 15:00:03 <David_Orange> fdegir: about stability topic 15:00:22 <fdegir> David_Orange: let me end the meeting first 15:00:31 <fdegir> thank you all for joining and talk to you next week 15:00:33 <David_Orange> fdegir: sure :) 15:00:33 <fdegir> #endmeeting