13:00:12 <fdegir> #startmeeting Cross Community CI 13:00:12 <collabot> Meeting started Wed Oct 4 13:00:12 2017 UTC. The chair is fdegir. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:12 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic. 13:00:12 <collabot> The meeting name has been set to 'cross_community_ci' 13:00:22 <fdegir> anyone for XCI meeting? 13:00:59 <hwoarang> im here 13:01:10 <mbuil> fdegir: are there cookies? 13:01:16 <fdegir> in fact I do 13:01:18 <epalper> I'm in :) 13:01:23 <fdegir> we have kanelbulle day in Sweden 13:01:39 <mbuil> fdegir: ooooh!!! 13:01:55 <mbuil> nice 13:02:00 <mbuil> #info Manuel Buil 13:02:02 <fdegir> if you ever visit here (again), I'll buy you some 13:02:30 <fdegir> please info in your names 13:02:52 <fdegir> while you do that, put the link to the agenda which is very similar to last week's 13:03:01 <fdegir> #link https://etherpad.opnfv.org/p/xci-meetings 13:03:07 <epalper> #info Periyasamy Palanisamy 13:03:07 <OPNFV-Gerrit-Bot> Markos Chandras proposed releng-xci: xci: Bump bifrost SHA https://gerrit.opnfv.org/gerrit/44021 13:03:17 <Taseer> #info Taseer 13:03:28 <hwoarang> #info Markos Chandras 13:03:34 <fdegir> first topic is 13:03:37 <fdegir> #topic Multi-distro support 13:03:59 <fdegir> #info Ubuntu and OpenSUSE work fine and work with CentOS is progressing 13:04:08 <fdegir> anything to add hwoarang ? 13:04:10 <hwoarang> nope 13:04:16 <durschatz> #info Dave Urschatz 13:04:47 <fdegir> #topic CI for XCI 13:05:00 <tinatsou> #info Tina Tsou 13:05:04 <fdegir> #info xci-verify got some cool stuff thanks to hwoarang 13:05:05 <ttallgren> #info Tapio Tallgren 13:05:22 <fdegir> #info The jobs run in clean VMs to ensure we don't get strange results 13:05:22 <David_Orange> #info David Blaisonneau 13:05:33 <fdegir> #info Ubuntu and OpenSUSE jobs are voting 13:05:46 <fdegir> ttallgren: you want to say something about Centos? 13:05:56 <ttallgren> Yes 13:06:00 <fdegir> please go ahead 13:06:00 <hwoarang> one patch needs review https://gerrit.opnfv.org/gerrit/#/c/44049/ which basically uses pre-built OS for the clean VM instead of building new images on every run 13:06:07 <hwoarang> make the job 10 minutes faster ;p 13:06:10 <hwoarang> *makes 13:06:33 <ttallgren> I got to network configuration, made a silly mistake, and spent a lot of time trying to find a trivial mistake 13:06:59 <fdegir> #info ttallgren is working on network configuration for Centos 13:07:37 <ttallgren> I could not take the network config from Ubuntu or CentOS directly, so I rewrote them from scratch 13:07:37 <fdegir> ttallgren: anything else to add? 13:08:02 <hwoarang> ttallgren: the suse one shouldnt be so different compare to centos 13:08:08 <hwoarang> at least the ifcfg parts 13:08:20 <ttallgren> Just a question: why do we have only one interface for the VMs? 13:08:48 <fdegir> ttallgren: the sandbox is based on upstream spec for "test" deployments 13:09:00 <fdegir> ttallgren: having proper VMs with multiple network interfaces and so on is in the backlog 13:09:24 <ttallgren> Ok. Would be easier to just add interfaces, rather than adding VLAN interfaces 13:09:35 <fdegir> ttallgren: we also need to switch to our own VM creation as well since current VMs are created by bifrost as "test" VMs 13:09:47 <fdegir> so multiple things to fix in order to have more proper setup 13:10:16 <timirnich> #info Tim Irnich 13:10:25 <ttallgren> Ok. I have been wondering when the real VMs come up 13:10:45 <hwoarang> i am not sure i follow 13:10:48 <fdegir> ttallgren: right after the release so we don't break it totally 13:10:52 <hwoarang> what are the real VMs? 13:11:17 <ttallgren> hwoarang: fdegir is referring to "test" VMs 13:11:19 <fdegir> hwoarang: the VMs we use now has only 1 interface etc and the deployment is based on test spec from osa 13:11:40 <fdegir> hwoarang: and these VMs are created using bifrost test script so they are not as good as they can be 13:11:48 <hwoarang> do you know what's missing? 13:11:53 <hwoarang> apart from the single NIC? 13:12:05 <fdegir> we can't control the spec per VM 13:12:13 <hwoarang> anything else? 13:12:14 <fdegir> they all get same cpu/ram/disk 13:12:29 <fdegir> need to look at what I've done for this part 13:12:35 <hwoarang> ok 13:12:52 <fdegir> ttallgren: so, some things are already known and will be fixed 13:13:07 <fdegir> but until that happens, what we need to do is to be done based on what we have in the repo 13:13:55 <fdegir> #info A patch is waiting to be reviewed which uses pre-built OS for the clean VM instead of building new images on every run, cuttine the time by 10 minutes 13:14:03 <fdegir> #link https://gerrit.opnfv.org/gerrit/#/c/44049/ 13:14:08 <fdegir> #topic Testing 13:14:18 <fdegir> #info prepare-functest role has been merged 13:14:23 <fdegir> mbuil: what about run-tests? 13:14:41 <mbuil> fdegir: Regarding functest integration, I have two patches 13:14:50 <mbuil> fdegir: The 1st patch fixes a missing dependency and I think is ready to be reviewed: https://gerrit.opnfv.org/gerrit/#/c/44101/ 13:15:00 <fdegir> #info A new playbook is currently being written for running tests; starting with functest 13:15:12 <fdegir> #info The patch fixes a missing dependency and ready for review 13:15:17 <fdegir> #ljnk https://gerrit.opnfv.org/gerrit/#/c/44101/ 13:15:41 <mbuil> fdegir: The 2nd patch sets everything up for functest and runs the functest container: https://gerrit.opnfv.org/gerrit/#/c/43635/ However, after discussing with ollivier and howarang I think it might be easier to create and run the docker container through the docker CLI instead of using ansible 13:15:55 <ttallgren> hwoarang: The CentOS networking configuration is complicated, since by default it uses NetworkManager and no one likes it. NetworkManager supports the ifcfg scripts and if there is no ifcfg, then it uses the "traditional" way. So I followed this: https://major.io/2017/04/13/openstack-ansible-on-centos-7-with-systemd-networkd/ 13:16:14 <fdegir> #info The 2nd patch sets everything up for functest and runs the functest container 13:16:20 <fdegir> #link https://gerrit.opnfv.org/gerrit/#/c/43635/ 13:16:39 <mbuil> fdegir: the reason is that the functest container does not upload functest logs and that will make things difficult if the test fails because we might not see the problem 13:17:09 <hwoarang> ttallgren: that's strange because upstream OSA supports regular ifcf networking on CentOS. if you use systemd that means centos will do thing completely different than the rest of the distros which will be hard to simplify the role 13:17:28 <fdegir> #info We will run the testing using docker CLI to ensure we can make what went wrong visible and upload the test logs 13:17:46 <fdegir> thx mbuil - we will talk about sfc in its own topic 13:17:50 <mbuil> fdegir: so for the time being, I suggest using docker CLI "docker run...". That means the playbook sets everything up for functest and then we run the functest container with normal CLI. It is ok from jjob perspective or? 13:18:17 <fdegir> mbuil: please fix it in a simple way and we get it into jjb 13:18:29 <mbuil> fdegir: yes, sir :) 13:18:40 <fdegir> mbuil: as I mentioned, this is probably temporary and we need to revisit this after release 13:18:58 <fdegir> #topic Documentation 13:19:09 <fdegir> #info Started moving documentation to docs.opnfv.org 13:19:23 <fdegir> #info User guide is available: docs.opnfv.org/en/latest/submodules/releng-xci/docs/xci-user-guide.html 13:19:42 <hwoarang> one question 13:19:53 <fdegir> #info Work with Overview, Developer Guide, and XCI way of working is in progress 13:19:55 <hwoarang> how does this documentation relates to the doc patches you submit to releng-xci ? 13:20:10 <fdegir> hwoarang: this documentation is generated from those patches 13:20:14 <hwoarang> ok 13:20:27 <fdegir> hwoarang: but the docs are available on docs.opnfv.org post-merge 13:20:41 <hwoarang> is there a way to generate such documentation locally to check how our patch would look like? 13:20:52 <fdegir> hwoarang: yes, there is 13:20:58 <fdegir> let me put the link 13:21:02 <hwoarang> ok thank you 13:21:36 <fdegir> #info You can test your doc patches locally and the guide below explains how it can be done 13:21:38 <fdegir> #link http://docs.opnfv.org/en/latest/how-to-use-docs/include-documentation.html#testing-build-documentation-locally 13:22:16 <fdegir> #info Also, jenkins posts comments to the change pointing the generated documents such as "Patch Set 2: 13:22:16 <fdegir> Document link(s): 13:22:16 <fdegir> http://artifacts.opnfv.org/releng-xci/review/43955/index.html" 13:22:51 <fdegir> but please note that the toolchain that generates the documents for non-merged patches is different from the one that generates after merge 13:23:15 <fdegir> so expect small differences and it is better to test documents locally 13:23:28 <fdegir> #topic Bifrost jobs status 13:23:38 <mbuil> fdegir: I was planning to add a guide in SFC about how to deploy the SFC scenario with XCI but it might be worth trying to create one general "how to run scenarios" guide in xci and then I could just write the SFC details and refer to that releng-xci wiki page 13:23:58 <fdegir> mbuil: that will be part of developer and way of working guide 13:24:05 <fdegir> can the person who added bifrost topic talk about it? 13:24:08 <hwoarang> yes 13:24:40 <hwoarang> so right now bifrost jobs are queuing but nothing executes them 13:24:41 <fdegir> hwoarang: please go ahead 13:24:41 <hwoarang> so 13:24:47 <hwoarang> 1) do we still want them? 13:25:11 <hwoarang> that's only for testing upstream bifrost patches. internal bifrost changes are being tested in the regular XCI jobs 13:25:40 <hwoarang> so the opnfv-bifrost-* jobs can be deleted right now. I want to know what happens to the openstack-bifrost-* ones 13:25:55 <fdegir> this depends on the hw availability so if we have enough machines, I think we should continue running openstack-bifrost-* ones 13:25:59 <hwoarang> we can resume them, and use the clean VM approache. 13:26:02 <hwoarang> *approach 13:26:13 <fdegir> as I know the Bifrost PTL is looking into feedback we give - when we give them 13:26:19 <hwoarang> ie just the first bit of the XCI job 13:26:49 <hwoarang> ok since we still need them I will work on enable these back 13:27:04 <hwoarang> *enabling 13:27:09 <hwoarang> can't type today again 13:27:23 <fdegir> #action hwoarang to remove opnfv-bifrost-* jobs, update openstack-bifrost-* jobs to use clean-VM approach and enable them when we have the machines 13:27:44 <hwoarang> i'm done 13:27:48 <fdegir> thx 13:28:01 <fdegir> #topic Release Readiness 13:28:09 <fdegir> #info Release has been delayed by 2 weeks so skipping the topic 13:28:56 <fdegir> #topic Scenario/Feature Status: os-odl-nofeature 13:29:13 <fdegir> epalper: I think the status is same as last week - ie change is still pending review? 13:29:26 <epalper> Now moved code changes related to os-odl-nofeature scenario from releng-xci into ansible-opendaylight repo 13:29:38 <epalper> #link https://git.opendaylight.org/gerrit/#/c/63938/ 13:29:58 <fdegir> #info epalper moved changes related to the scenario from releng-xci into ansible-opendaylight repo 13:30:03 <epalper> will raise another patch on top of mbuil's releng-xci review for the integration. 13:30:06 <fdegir> epalper: that's cool! 13:30:14 <epalper> #link https://gerrit.opnfv.org/gerrit/#/c/43469/ 13:30:19 <fdegir> epalper: but 13:30:31 <fdegir> epalper: I'm not sure if ODL will accept this type of stuff there 13:31:07 <fdegir> epalper: I mean having xci and/or osa things there 13:31:26 <epalper> I just raised the review. Let us wait for their review comments 13:31:40 <epalper> even sfc is from ODL :) 13:32:00 <fdegir> epalper: right - I will put this as review there since I know their ansible role is a proper/standard one 13:32:16 <fdegir> thx epalper - will come to ovs after sfc 13:32:21 <mbuil> epalper: I am also not sure because the SFC scenario includes non-ODL stuff: openstack, OVS... 13:32:26 <fdegir> #topic #topic Scenario/Feature Status: os-odl-sfc 13:32:41 <fdegir> mbuil: sfc simply works, isn't it? 13:33:07 <mbuil> fdegir: almost. I was having an issue with the credentials because of a problem with user_Variables.yml 13:33:07 <fdegir> mbuil: anything else to mention? 13:33:16 <mbuil> fdegir: The first sfc scenario patch is merged: https://gerrit.opnfv.org/gerrit/#/c/43165/ 13:33:28 <mbuil> fdegir: I have a second one in the oven. I am deploying with it right now to test if it works (after the user_variables.yml discussion we had): https://gerrit.opnfv.org/gerrit/#/c/44031/ 13:33:36 <fdegir> #info First sfc scenario patch is merged: https://gerrit.opnfv.org/gerrit/#/c/43165/ 13:34:12 <fdegir> #info Work with another patch is going on: https://gerrit.opnfv.org/gerrit/#/c/43165/ 13:34:28 <mbuil> fdegir: you wrote the same link twice 13:34:39 <fdegir> #undo 13:34:39 <collabot> Removing item from minutes: <MeetBot.ircmeeting.items.Info object at 0x33a7690> 13:34:50 <fdegir> #info Work with another patch is going on: https://gerrit.opnfv.org/gerrit/#/c/44031/ 13:34:52 <fdegir> thx mbuil 13:35:35 <fdegir> mbuil: I'm not adding user_variables.yml details as of yet until you verify the approach 13:35:49 <mbuil> fdegir: ok 13:36:20 <fdegir> #topic Scenario/Feature Status: os-nosdn-ovs 13:36:26 <fdegir> epalper: how about this scenario? 13:36:45 <epalper> raised a review in releng-xci: https://gerrit.opnfv.org/gerrit/#/c/43447/ and testing it now. 13:37:09 <epalper> might need to right repo to move this scenario 13:37:19 <epalper> need to find* 13:37:24 <fdegir> #info A change has been sent for review and currently being tested locally: https://gerrit.opnfv.org/gerrit/#/c/43447/ 13:37:43 <fdegir> epalper: until we find that, we can keep it next to os-odl-nofeature in releng-xci 13:37:59 <epalper> ok, sure 13:38:19 <fdegir> we can perhaps talk to ovs team once we have something and they take it if they want to be part of XCI 13:38:25 <fdegir> thx epalper 13:38:46 <fdegir> #topic Scenario/Feature Status: k8s-nosdn-nofeature 13:39:06 <fdegir> Stephen is not with us today due to Chinese public holidays so here is a short update I got from him 13:39:32 <fdegir> #info s3wong is evaluating kubespray to see what needs to be done to setup a simple k8s cluster using it 13:39:57 <fdegir> #info He added some details to the etherpad https://etherpad.opnfv.org/p/xci-k8s 13:40:26 <fdegir> #info Patches will be sent once he is back from holiday and done with first steps 13:40:36 <fdegir> David_Orange: anything from your side for kubernetes? 13:40:38 <OPNFV-Gerrit-Bot> Manuel Buil proposed releng-xci: Create the run-tests playbook https://gerrit.opnfv.org/gerrit/43635 13:41:13 <David_Orange> fdegir: nothing from now, i was waiting fir s3wong input/first tests 13:41:31 <fdegir> David_Orange: right 13:41:52 <David_Orange> fdegir: but i can put it in the todo list, to test kubspray too 13:41:53 <fdegir> David_Orange: his notes give some idea about what he is doing so you can perhaps take a look and add what you think directly there 13:42:14 <fdegir> thx David_Orange 13:42:15 <David_Orange> fdegir: ok, i will 13:42:28 <fdegir> btw kubespray is just one option but a good looking one 13:42:41 <fdegir> if someone else has better idea, please share it on the etherpad directly 13:42:58 <fdegir> #topic Scenario/Feature Status: Congress 13:43:05 <fdegir> Taseer: your turn 13:43:41 <fdegir> moving to ceph 13:43:45 <Taseer> yes, I am able to setup a congress container, and run through the pip install path 13:44:06 <fdegir> Taseer: good - you made it before I change the topic 13:44:16 <fdegir> #info Taseer is able to setup a congress container, and run through the pip install path 13:44:36 <Taseer> failing on adding the service to keystone 13:44:48 <fdegir> Taseer: ok 13:45:00 <fdegir> Taseer: have you looked at the code bryan_att provided? 13:45:07 <fdegir> Taseer: I suppose he's doing similar things there 13:45:45 <fdegir> #info Failure while adding the service to keystone - investigation ongoing 13:45:54 <Taseer> fdegir: yes, but he is doing using the bash commands, I need to follow the OSA convention 13:46:09 <fdegir> Taseer: ok 13:46:46 <fdegir> #topic Scenario/Feature Status: Ceph 13:47:04 <fdegir> Tianwei is also on holidays so here is a short update 13:47:35 <fdegir> #info Tianwei is working on integrating ceph: https://gerrit.opnfv.org/gerrit/#/c/42503/ 13:47:48 <fdegir> #topic Baremetal status 13:47:57 <fdegir> David_Orange: I think you are still waiting some more reviews... 13:48:00 <David_Orange> testing on ONAP OpenLab: playbooks runs are all green but glance can not write on nfs mount point due to local rights (770 on this pod, 777 on pod1). Debug is ongoing 13:48:24 <David_Orange> fdegir: yes, but this can wait for a few days$ 13:48:24 <fdegir> #info David_Orange is testing it on ONAP OpenLab 13:48:40 <David_Orange> fdegir: i am cleaning the code to push playbooks tasks to ansible roles 13:48:40 <fdegir> #info Playbook runs are all green but glance can not write on nfs mount point due to local rights (770 on this pod, 777 on pod1). Debug is ongoing 13:49:07 <fdegir> thx for that too and again sorry for lack of reviews 13:49:13 <David_Orange> fdegir: so it should be push in a few days 13:49:19 <fdegir> the release stuff messed up things for everyone 13:49:21 <David_Orange> a new deployment is planed on Orange Pod2 community lab (PDF/IDF writting is ongoing) 13:49:27 <David_Orange> fdegir: np 13:49:43 <fdegir> #info The code is being cleaned up and roles will be created 13:49:51 <fdegir> #info A new deployment is planed on Orange Pod2 community lab (PDF/IDF writting is ongoing) 13:50:09 <fdegir> David_Orange: anything else? 13:50:09 <David_Orange> this will give me another source for testing ONAP open lab issue 13:50:22 <David_Orange> ys, i am also preparing new openstack_user_config for nonha deployments using 'test' template 13:50:39 <fdegir> #info Testing it on Orange POD2 will give chance to check the issue happening in ONAP OpenLab 13:50:52 <David_Orange> for those that wana play with it with less servers 13:51:07 <fdegir> David_Orange: for baremetal you mean? 13:51:18 <David_Orange> yes, 1 controller, n computes 13:51:31 <fdegir> David_Orange: ok - so you are creating flavors for baremetal 13:51:37 <fdegir> David_Orange: which is good 13:51:46 <David_Orange> fdegir: yes :) 13:52:01 <durschatz> David_Orange: are there multiple nics in bare metal deployments 13:52:13 <David_Orange> fdegir: if you have some recommendation on that i will take 13:52:16 <fdegir> David_Orange: perhaps we can have aio, mini, noha, and ha so no matter if it is virtual or baremetal we have the same flavors 13:52:32 <tinatsou> #link https://wiki.opnfv.org/display/AUTO 13:52:45 <David_Orange> durschatz: yes, it is based on PDF/IDF, so for actual testes there is 3 interfaces 13:52:58 <David_Orange> durschatz: including vlan management 13:53:14 <durschatz> sweet! 13:53:50 <fdegir> #info Different flavors for baremetal will be available similar to the ones with virtual machines 13:53:56 <fdegir> thx David_Orange! 13:54:11 <fdegir> tinatsou: what about the link you put? 13:54:14 <David_Orange> i will have a few refactor to do with PDF/IDF updates, but it will be simple 13:54:52 <fdegir> moving to the last topic 13:54:54 <fdegir> #topic AoB 13:55:03 <fdegir> #info We are waiting for new machines to become available 13:55:05 <fdegir> jmorgan1: ^ 13:55:40 <fdegir> #info Workaround for Jenkins slave connection from Intel lab works fine and it will be used until the firewall is configured by Intel corp IT 13:55:49 <fdegir> anyone has anything to add? 13:56:12 <fdegir> durschatz: any issues with the sandbox? 13:56:28 <durschatz> sandbox is working fine 13:56:30 <durschatz> also 13:56:50 <durschatz> I have need to deploy ODL on bare metal pod soon 13:57:01 <fdegir> durschatz: how soon? 13:57:20 <durschatz> may also approach Dave to test ONAP deployments 13:57:54 <fdegir> durschatz: as you can guess, we first have to get odl done on virtual machines 13:58:05 <durschatz> next week so I may just use apex unless something solid comes up with XCI 13:58:11 <durschatz> yes 13:58:27 <fdegir> David_Orange is our baremetal expert so it should just work fine 13:58:45 <fdegir> and one final note 13:58:59 <fdegir> #info Bifrost and OSA shas might be bumped soon 13:59:03 <durschatz> seriously though, shuould I use apex for now? 13:59:07 <fdegir> thank you all! 13:59:12 <fdegir> #endmeeting