17:02:08 <narindergupta> #startmeeting OPNFV joid
17:02:08 <collabot`> Meeting started Wed Dec 23 17:02:08 2015 UTC.  The chair is narindergupta. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:02:08 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:02:08 <collabot`> The meeting name has been set to 'opnfv_joid'
17:02:14 <ibenrodriguez> can you info in to the meting?
17:02:24 <ibenrodriguez> #info Iben Rodriguez
17:02:26 <narindergupta> #info Narinder Gupta
17:02:44 <narindergupta> #topic
17:14:44 <ibenrodriguez> we are chatting with ashyoung about integrating ONOS with JOID - the systems they have that can run ubuntu 14 are virtual the physical need ubuntu 15
17:15:24 <ibenrodriguez> ashyoung: it’s recommended to setup a jenkins slave and then have the jenkins jobs do the mass and juju JOID setup and ONOS installation -
17:15:43 <ibenrodriguez> maas can also drive other installers like fuel, apex, and compass
17:16:14 <narindergupta> #info https://wiki.opnfv.org/get_started_juju_deploy
17:16:46 <narindergupta> #info https://wiki.opnfv.org/joid/get_started
17:16:47 <ibenrodriguez> the san jose lab is available for other teams to use
17:24:02 <narindergupta> https://wiki.opnfv.org/pharos/maas_getting_started_guide
17:24:26 <ashyoung> cool
17:27:49 <ibenrodriguez> now we’re talking about juniper lab setup - and jjb scripts made by david at orange
17:28:23 <ibenrodriguez> there was a topic before about a script narindergupta made for jump host vm reset - it’s not documented or working with jenkins yet but that’s coming
17:29:30 <ibenrodriguez> https://docs.google.com/presentation/d/1agyVSzmg4BIfU87S7a1HnscjYGZi8NY96TIN9sUxx00/edit#slide=id.gf4cec6a78_0_23
01:31:22 <yuanyou> Hi all ,i can use the maas to power on the node ,but the node can't commisioning sucessfully?
17:12:14 <narindergupta> bryan_att, hi Bryan happy new year
17:12:25 <narindergupta> bryan_att, how are you and things are?
17:23:24 <narindergupta> yuanyou, hi
01:32:01 <yuanyou> Hi,Narindergupta.Happy new year
15:52:54 <narindergupta> yuanyou, hi
04:08:40 <narindergupta> yuanyou, hi
05:14:13 <yuanyou> hi narinder
06:50:06 <narindergupta> yuanyou, i have made changes and checkin the code to include eth1 in compute
06:50:22 <narindergupta> yuanyou, do you have any other question on joid?
06:51:28 <narindergupta> yuanyou, also if you can provide us the charm diagram and how networking works in case onosfw will be great? Also if onosfw charm is complete then please test it on bare metal in any of your labs?
07:24:56 <yuanyou> ok,thanks
07:27:16 <yuanyou> i will send you the networking detail info.
07:35:12 <yuanyou> and my bare metal is limited,can you supply a lab to test the onosfw charm.
07:41:41 <narindergupta> unfronately i also have limited lab access as intel pod5 is already configured with CI now
07:42:01 <narindergupta> and i have pod6 working for my development work. Does it installs in VM?
07:53:01 <yuanyou> yes,I have installed it in VM
07:56:38 <yuanyou> I have a question. when we use CI to test our charms, we have 2 steps: massdeploy and deploy.sh. If we should all excute the two steps.
07:57:53 <narindergupta> maasdeploy should be done one time
07:58:09 <narindergupta> then you cn keep repeating deploy.sh
08:02:51 <yuanyou> The CI lab should have excute the maasdeply.sh. And how many labs do we have?
15:11:53 <fdegir> narindergupta: ping
15:12:21 <David_Orange> fdegir: hi
15:12:30 <fdegir> hi David_Orange
15:12:56 <David_Orange> fdegir: happy new year to you
15:13:10 <fdegir> thx David_Orange, same to you
15:13:29 <David_Orange> fdegir: do you have a few minutes to discuss about joid jobs ?
15:13:52 <fdegir> David_Orange: can do that
15:14:06 <fdegir> David_Orange: before we go into details, I must say that those jobs will probably be updated
15:14:15 <fdegir> David_Orange: once the scenario definitions become available
15:14:32 <fdegir> David_Orange: so my proposal would be to hold on with the changes for a while
15:14:42 <David_Orange> fdegir: ok, good to know i am working on the to merge pods workers
15:15:15 <fdegir> David_Orange: you mean builders for the pods
15:15:28 <David_Orange> fdegir: yes builders
15:15:40 <fdegir> David_Orange: that's good
15:15:58 <David_Orange> fdegir: to merge them to have to avoir problem for functest
15:16:08 <fdegir> David_Orange: yes
15:16:11 <fdegir> David_Orange: but the thing is
15:16:21 <fdegir> David_Orange: you had some local files on Orange POD
15:16:34 <David_Orange> yes, but it is optionnal
15:16:39 <fdegir> David_Orange: ok
15:16:52 <fdegir> so builders should be same then
15:16:52 <David_Orange> and it is a part of the things i want to discuss
15:17:27 <fdegir> ok
15:17:44 <David_Orange> fdegir: to avoid this config file for ci i need to add ceph config in pods params
15:18:13 <David_Orange> but shall we do it now or wait for the end of the code freeze
15:18:47 <fdegir> no need to wait
15:18:52 <fdegir> releng will not have code freeze
15:19:03 <David_Orange> the optional config file must be only used to set specific passwords, not pod params
15:19:23 <fdegir> can you send a patch so I can see what this really means
15:19:38 <David_Orange> ok, good for me, i do that
15:20:15 <fdegir> looking forward to 2 patches then
15:20:19 <fdegir> anything else?
15:20:50 <narindergupta> fdegir,
15:20:56 <fdegir> hi narindergupta
15:21:10 <fdegir> narindergupta: I wanted to check with you the status of POD5 & POD6
15:21:14 <fdegir> narindergupta: how it is going?
15:22:34 <narindergupta> fdegir, hi
15:22:46 <narindergupta> fdegir, both are ready
15:23:05 <fdegir> narindergupta: does this mean you deploy 3 controllers + 2 computes?
15:23:06 <David_Orange> fdegir: yes, is there the priority on params, and if yes, how it works. If i set a param on project, and the same in pod-defaults, which one will be used ? (eg: [project] CEPH_DISK=/srv [pod] CEPH_DISK=/dev/sdb)
15:23:27 <fdegir> David_Orange: the ones that is lower in parameter list
15:23:31 <narindergupta> fdegir, correct
15:23:42 <fdegir> narindergupta: cool
15:23:49 <fdegir> narindergupta: can you send me the ip of pod6 jumphost?
15:23:52 <narindergupta> fdegir, yes can you restart the pod
15:24:23 <fdegir> narindergupta: the job you mean?
15:24:39 <narindergupta> fdegir, yes also how can i get access to restart the job?
15:24:54 <narindergupta> fdegir, i wanted to restart yesterday once i completed the work but could not
15:24:54 <fdegir> narindergupta: please send an email to opnfv-helpdesk
15:25:00 <David_Orange> fdegir: ok, thx. Is it compatible with the zen coding rules of jenkins or a usage to avoid ?
15:25:24 <fdegir> narindergupta: just to remind what we will do with these pods
15:25:35 <fdegir> narindergupta: pod5 will run against stable/brahmaputra branch
15:25:48 <narindergupta> fdegir, ok
15:25:49 <fdegir> narindergupta: pod6 will be connected to jenkins and run against master branch
15:26:07 <narindergupta> fdegir, gotcha you that will be helpful.
15:26:15 <fdegir> narindergupta: so it is important for you to check jenkins to see if any job is running on any of the pods
15:26:22 <fdegir> narindergupta: if so, hands off please
15:26:33 <narindergupta> fdegir, also is it possible after running with old we can run job with open contrail and onofw as well?
15:26:44 <fdegir> narindergupta: if the trigger time is approaching, again please refrain from doing manual work on the pods
15:27:05 <narindergupta> fdegir, gotcha you yeah i can see trigger time is 3:00 AM
15:27:05 <fdegir> narindergupta: onos + opencontrail will be enabled once odl is verified to be working on both pods
15:27:14 <narindergupta> fdegir, ok thanks
15:27:19 <fdegir> narindergupta: please note that those times are in UTC
15:27:48 <narindergupta> fdegir, for funkiest they need the credential files in /var/lib/jenkins and for that job need some modification
15:28:37 <fdegir> then copying the credential files to the folder can be done within the builder
15:28:57 <fdegir> so everytime the job starts, it puts whatever file needed to whatever path
15:31:14 <fdegir> narindergupta: running a standalone deploy now: https://build.opnfv.org/ci/view/joid/job/joid-deploy-intel-pod5-daily-master/16/console
15:31:27 <fdegir> narindergupta: once this ends, will run full one with functest
15:31:39 <fdegir> narindergupta: and will hook in yardstick later on - probably tomorrow
15:31:54 <narindergupta> fdegir, ok sounds good
15:34:43 <fdegir> narindergupta: I need to kill the run
15:34:55 <fdegir> missed setting controller to odl
15:35:02 <fdegir> restarting full run instead
15:35:06 <David_Orange> narindergupta: by default for CI do we need to reinstall MAAS and Bootstrap ?
15:35:09 <narindergupta> Iok
15:35:24 <narindergupta> David maas reinstall is not needed
15:35:39 <narindergupta> maas can be installed once and can be used forever
15:35:48 <David_Orange> narindergupta: so i set maas reinstall to false by default
15:35:52 <fdegir> narindergupta: https://build.opnfv.org/ci/view/Brahmaputra%20CI/job/joid-odl-intel-pod5-daily-master/18/console
15:36:34 <narindergupta> David_Orange, but if there are changes done in maas deployments.yaml then i would prefer to install maas as well to test
15:36:47 <David_Orange> narindergupta: yes, i know, but from a CI point of view, install time will be reduced
15:37:03 <narindergupta> David_Orange, agreed
15:37:23 <David_Orange> narindergupta: i prepare the jenkins param, so it can be changed
15:37:37 <narindergupta> David_Orange, that is helpful thanks.
15:41:37 <David_Orange> narindergupta: i have a param named INSTALLER_VERSION in jenkins orange pod default set to 'latest', and i can not find one in installed default, do you manage the absence of this param in your script ?
15:42:31 <narindergupta> installer we use ppa/stable or ppa/next
15:42:47 <fdegir> David_Orange: that's ci parameter
15:43:16 <narindergupta> idea is when ppa/next will be released then i will copy the same to opnfv ppa so we have constant version.
15:43:16 <David_Orange> so it is needed by jenkins
15:45:04 <fdegir> David_Orange: not utilized yet but could be
15:45:13 <fdegir> David_Orange: so please leave it as it is
15:45:24 <David_Orange> ok
15:46:14 <fdegir> narindergupta: what is the ip of pod6 jumphost?
15:46:48 <narindergupta> 10.4.1.1
15:47:00 <narindergupta> public is 10.2.66.2
15:47:19 <narindergupta> i have created the jenkins user there
15:47:46 <fdegir> can you send me the password?
15:48:41 <fdegir> or add my pubkey to authorized keys
15:48:45 <fdegir> narindergupta: ^
15:48:45 <narindergupta> for jenkins user it is jenkins
15:48:54 <narindergupta> but i need to add your ssh keys
15:49:02 <fdegir> narindergupta: passwd auth is not available
15:49:03 <fdegir> yep
15:49:28 <narindergupta> will you send me your ssh keys?
15:49:53 <narindergupta> if you have launchpad account then give me launchpad id i can import from their as well
15:50:12 <fdegir> narindergupta: sent it
15:52:05 <narindergupta> fdegir, done please try now?
15:52:46 <fdegir> narindergupta: works, thanks
15:52:57 <fdegir> narindergupta: will fix jenkins side of things
15:53:02 <fdegir> probably tomorrow
15:53:33 <narindergupta> fdegir, no problem i clone the releng and try to modify the jobs
15:53:45 <narindergupta> you can find releng directory there
15:54:13 <fdegir> narindergupta: releng repo is only needed for connecting node to jenkins
15:54:22 <narindergupta> oh ok
15:54:26 <fdegir> narindergupta: jobs will end up on jenkins
15:54:31 <fdegir> narindergupta: please leave the rest to me
15:54:34 <narindergupta> let me know if i can learn
15:54:43 <fdegir> narindergupta: will do
15:54:51 <narindergupta> fdegir, thanks
15:54:55 <fdegir> narindergupta: apart from this
15:55:07 <fdegir> narindergupta: can you please update the wiki with correct info
15:55:09 <fdegir> narindergupta: https://wiki.opnfv.org/get_started/pod_5_-_specs
15:55:13 <fdegir> narindergupta: https://wiki.opnfv.org/get_started/pod_6_-_specs
15:55:22 <narindergupta> fdegir, yes i will do it today.
15:55:24 <fdegir> since we now use node1s as jumphosts
15:55:25 <David_Orange> fdegir: about openrc file that need to be shared between jenkins jobs, i set the param relative to $HOME (jenkins home) but if you prefer something else let me know
15:55:37 <narindergupta> fdegir, also virtual install also working with joid
15:55:44 <fdegir> narindergupta: cool!
15:55:52 <fdegir> we need a machine for that one to test it then
15:55:59 <fdegir> David_Orange: how does that openrc file ends up there?
15:56:31 <David_Orange> the script executed by the builder
15:56:38 <fdegir> fetch-os-creds?
15:56:43 <narindergupta> to do install run ./02-maasdeploy.sh then run ./deploy.sh -o odl
15:57:18 <David_Orange> fetch-os-creds is the script executed by functest to get creds, but with joid we share it with functest docker
15:57:36 <fdegir> David_Orange: ok
15:57:46 <fdegir> David_Orange: but the openrc file is not a static file, right?
15:57:56 <fdegir> David_Orange: you always create new one when the job starts
15:58:02 <David_Orange> no, it is written at each install
15:58:39 <David_Orange> at each joid_deploy
15:59:18 <fdegir> David_Orange: isn't is same with what I said above?
16:00:51 <David_Orange> if you means "job starts"="joid_deploy job starts" yes, if it is "joid/functest/... job starts" it is false
16:02:02 <fdegir> David_Orange: I really don't care which job puts it there as long as it is generated automatically
16:02:04 <fdegir> :)
16:02:37 <fdegir> can fix it later on to make it common though - if possible
16:04:12 <David_Orange> fdegir: ok :)
16:12:45 <David_Orange> fdegir: today we have 2 multijob builders one for orange-pode2 and one for intel-pod5, but we can merge it i thinks. Orange is not a CI pod but it will run functest AND yardstick to at the end of the deploy. Are you ok to merge them or do you prefer to keep them seperate ?
16:13:11 <fdegir> David_Orange: please keep them separate for the timebeing
16:13:18 <fdegir> David_Orange: I'll be adding intelpod6
16:13:28 <fdegir> David_Orange: once I verify the jobs for CI pods
16:13:33 <fdegir> David_Orange: then we can look at merging them
16:14:08 <David_Orange> ok, i just merge deploy builders
16:14:16 <fdegir> yep
16:14:26 <fdegir> David_Orange: so pod5 & pod6 use the same multijob at the start
16:14:46 <fdegir> David_Orange: and then orange can join them after we're sure ci pods work
16:14:52 <David_Orange> ok
16:15:23 <David_Orange> fdegir: about, triggers, will jenkins fails if there is no triggers declared for a pod ?
16:15:53 <fdegir> David_Orange: nope - lack of trigger prevents jenkins from triggering the job
16:16:00 <fdegir> automatically I mean
16:16:06 <fdegir> so no harm
16:16:26 <David_Orange> fdegir: so i can remove orange-pod2 trigger for a better lisibility
16:24:36 <fdegir> David_Orange: that will be needed when we switch to 1 multijob for all pods
16:24:53 <fdegir> David_Orange: up to you
16:24:58 <fdegir> for current version
16:25:49 <David_Orange> ok, so i keep it, to prevent futur problems
16:25:53 <David_Orange> thx
16:29:12 <fdegir> np
16:38:04 <fdegir> narindergupta: ping
16:39:09 <fdegir> narindergupta: I noticed the joid deploy got stuck for 6 days while I'm on vacation
16:39:11 <narindergupta> fdegir, in meeting
16:39:12 <fdegir> narindergupta: https://build.opnfv.org/ci/job/joid-deploy-intel-pod5-daily-master/12//console
16:39:26 <fdegir> narindergupta: ok, please take a look at above log
16:39:36 <narindergupta> fdegir, sure
16:39:43 <fdegir> narindergupta: and let me know a timeout value (3 hours etc.) so we kill it if it hangs
16:39:49 <fdegir> narindergupta: in order not to block CI totally
16:43:11 <narindergupta> fdegir, do not know the reason but yes 2 hrs timeout is good enough. If we you can put timeout for different scripts then may be 1 hr timeout for maas deployment should be ok.
16:44:37 <fdegir> narindergupta: we can have timeout per job
16:44:51 <fdegir> narindergupta: meaning whatever the deploymnent jenkins job executes
16:44:58 <fdegir> narindergupta: total time please - 3 hours?
16:45:11 <narindergupta> total time 2 hrs.
16:45:24 <fdegir> narindergupta: better to have offset
16:45:44 <fdegir> I generally put 50% extra
16:45:46 <narindergupta> fdegir, as we are seeing normally  1 hr 28  minutes yeah 3 hrs could be best
16:45:51 <fdegir> ok
16:45:58 <narindergupta> fdegir, gotcha you will work then 3 hrs.
16:46:04 <fdegir> yep
16:58:48 <David_Orange> narindergupta: just to be sure, if your script returns 0, it is success
16:59:02 <narindergupta> David_Orange, correct
16:59:06 <David_Orange> ok
17:02:05 <dduffey> Hi
17:02:10 <catbus1> o/
17:02:20 <arturt_> #meetingstart join weekly
17:02:31 <arturt_> #staetmeeting join weekly join weekly
17:02:40 <collabot`> arturt_: Error: Can't start another meeting, one is in progress.  Use #endmeeting first.
17:02:52 <arturt_> #endmeeting