18:15:56 <edwarnicke> #startmeeting Armoury Weekly Meeting
18:15:56 <odl_meetbot> Meeting started Mon Sep 21 18:15:56 2015 UTC.  The chair is edwarnicke. Information about MeetBot at http://ci.openstack.org/meetbot.html.
18:15:56 <odl_meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
18:15:56 <odl_meetbot> The meeting name has been set to 'armoury_weekly_meeting'
18:15:58 <mmarsale1> #info mmarsale
18:16:02 <edwarnicke> #topic Rollcall
18:16:08 <adetalhouet> #info adetalhouet
18:16:14 <yapeng> #info yapeng
18:16:14 <edwarnicke> Please #info in (again) now that I properly started the meeting :)
18:16:16 <edwarnicke> #info edwarnicke
18:16:25 <subh> #info subh
18:16:28 <mmarsale1> #info mmarsale
18:16:52 <edwarnicke> ariel_noy: You still with us :)
18:17:01 <ariel_noy> yes
18:17:15 <edwarnicke> ariel_noy: Could you #info in again so it shows up in the meeting minutes (I borked starting the meeting)
18:17:26 <ariel_noy> #info ariel_noy
18:18:13 <edwarnicke> Last call for roll call
18:18:21 <edwarnicke> Going once, going twice...
18:18:26 <edwarnicke> #topic Agenda Bashing
18:18:59 <edwarnicke> Speak up for things you'd like to discuss today :)
18:19:24 <mmarsale1> edwarnicke: PTL election ?
18:19:30 <edwarnicke> mmarsale1: Definitely on the list :)
18:19:40 <edwarnicke> Others?
18:19:47 <adetalhouet> review the workload manager patch from ariel_noy
18:19:56 * edwarnicke suspects we should probably finish up the Release Plan...
18:20:16 * edwarnicke and get some folks working on trello cards to get things going forward
18:20:26 <adetalhouet> fair enough
18:20:42 <edwarnicke> yapeng: adetalhouet ariel_noy ijw subh Any other thoughts?
18:20:55 <edwarnicke> alagalah: ?
18:21:00 * alagalah thinking
18:21:18 <yapeng> agreed
18:21:22 <adetalhouet> edwarnicke: not really
18:21:29 <edwarnicke> adetalhouet: OK...
18:21:34 <ariel_noy> ok
18:21:42 * ijw tries not to think on Mondays, it's ageing
18:21:43 <edwarnicke> So I think we have: PTL elections, Release Planning
18:21:56 * edwarnicke tries not to think at all ;)
18:22:02 <edwarnicke> #topic PTL elections
18:22:03 * alagalah pulling up wiki
18:22:15 <adetalhouet> https://wiki.opendaylight.org/view/Armoury#Agenda_for_next_meeting
18:22:37 <edwarnicke> adetalhouet: :)
18:22:46 <adetalhouet> oops, there is nothing there
18:22:49 <edwarnicke> #link https://lists.opendaylight.org/pipermail/armoury-dev/2015-September/000007.html <- PTL election kickoff email
18:23:02 <edwarnicke> So far no one has responded self nominating...
18:23:51 <mmarsale1> edwarnicke: so whats the plan in such situations ? :)
18:24:02 <edwarnicke> Well... someone needs to be PTL ;)
18:24:11 <alagalah> mmarsale1, Else we can't be in Be
18:24:11 <adetalhouet> could someone explain me what is the role of a project's PTL?
18:24:20 <edwarnicke> adetalhouet: Sure :)
18:24:27 <edwarnicke> adetalhouet: It comes in a few steps
18:24:52 <edwarnicke> adetalhouet: At the top level, there's the one specified in our governance, which basically means that should Armoury become a core project, the PTL would get a seat on the TSC (probably a bit off)
18:24:53 <adetalhouet> edwarnicke: please enumerate the steps so I can be fully aware of what is implicates :)
18:25:27 <edwarnicke> adetalhouet: Also, the PTL is the one responsible for trimming the committer rolls in the event that a committer has been idle more than 6 months (they don't *have* to, but at six months idle they can)
18:25:48 <edwarnicke> adetalhouet: And then traditionally they also are the point of contact for the Simultaneous Release stuff
18:26:19 <edwarnicke> adetalhouet: They also tend to be the ones calling and leading the project meetings, though that doesn't strictly have to be that way
18:26:25 <edwarnicke> adetalhouet: Does that answer your question?
18:26:40 <adetalhouet> edwarnicke: ok I understand - thanks for the explanation
18:26:45 <adetalhouet> edwarnicke: yes it does
18:27:02 <edwarnicke> Anyone else have questions?
18:27:36 <anipbu> PTL should send the milestone emails
18:27:44 <edwarnicke> anipbu: Also a good point :)
18:28:31 <edwarnicke> So... someone needs to step forward
18:28:33 <adetalhouet> Ok. Well ... we need someone ... so let's say it: I'm sure not being the most qualify for the role, but I could present myself
18:28:45 <edwarnicke> adetalhouet: That would be most appreciated :)
18:29:14 <adetalhouet> edwarnicke: yes I bet is it :)
18:29:27 <adetalhouet> I'll respond to the mail
18:29:32 <edwarnicke> adetalhouet: Many thanks :)
18:29:52 <edwarnicke> Also... folks, if you haven't subscribed to: https://lists.opendaylight.org/mailman/listinfo/armoury-dev please do
18:30:03 <edwarnicke> And committers please chime in one way or the other on the PTL election
18:30:06 <alagalah> adetalhouet, yay!
18:30:49 <edwarnicke> Anything else on PTL elections before we move on?
18:30:49 <ariel_noy> Congratulation
18:31:02 <anipbu> adetalhouet: yay!
18:31:15 <edwarnicke> #topic Release Plan
18:31:17 <adetalhouet> alagalah: anipbu ariel_noy : Thanks
18:31:22 <alagalah> So with 12 committers, he will need 6 votes
18:31:29 <alagalah> adetalhouet, he has mine :D
18:31:36 <edwarnicke> https://wiki.opendaylight.org/view/Armoury/Beryllium_Release_Plan
18:31:36 <anipbu> #link https://wiki.opendaylight.org/view/Armoury/Beryllium_Release_Plan <- release plan
18:31:43 <ariel_noy> how do we vote?
18:31:46 <edwarnicke> anipbu: Thank you for grabbing that :)
18:31:49 <alagalah> ariel_noy, respond to the email
18:31:58 <adetalhouet> alagalah: thank you for the support
18:31:59 <edwarnicke> ariel_noy: Traditionally we vote on the mailing list, so there's a record in the mailing lists archives
18:32:02 <alagalah> ariel_noy, Traditionally "+1, 0, -1"
18:32:09 <edwarnicke> (doesn't *have* to be that way... but that's usually how its done)
18:32:31 <ariel_noy> i register to the e_mail list but not getting emaill
18:32:39 <ariel_noy> not sure if no one sending
18:32:40 <yapeng> adetalhouet, voted for you, way to go!
18:32:52 <edwarnicke> ariel_noy: There is email: https://lists.opendaylight.org/pipermail/armoury-dev/
18:32:59 <edwarnicke> #link https://lists.opendaylight.org/pipermail/armoury-dev/ <- email archives
18:33:00 <adetalhouet> yapeng: thank you
18:33:17 <edwarnicke> adetalhouet: Would you like to lead the Release Planning discussion here?
18:34:36 <adetalhouet> edwarnicke: sure I can give it a try but I don't really know what it implicates
18:34:39 <alagalah|alt> Sorry, lappy froze
18:34:46 <edwarnicke> adetalhouet: Give it a whirl :)
18:35:14 <adetalhouet> So should we talk about M2?
18:35:27 <adetalhouet> Or we should directly get to M3?
18:35:32 <alagalah|alt> Just go with your gut. What does our project proposal say we want to do then whats doable for Be
18:35:38 <edwarnicke> adetalhouet: If you think we have some unfinished business for M2, lets do that :)
18:36:09 <adetalhouet> ok so let's take it like that. We were admitted at M2 - so let's plan M3
18:36:22 <edwarnicke> adetalhouet: Cool :)
18:36:40 <edwarnicke> ariel_noy: Didn't you have a patch you were trying to push for some API stuff we could look at for M3?
18:36:49 <adetalhouet> so for M3, we want to define models for both NF catalog and Workload Managers
18:37:10 <edwarnicke> adetalhouet: Cool :)
18:37:22 <ariel_noy> I push the first patch
18:37:28 <ariel_noy> need to review
18:37:47 <alagalah|alt> Do we want to consider documenting a sequence diagram for what we want to do?
18:37:47 <adetalhouet> So for NF models, we already have something from mmarsale1
18:37:56 <ariel_noy> also we can talk about the interaction model
18:37:59 <edwarnicke> alagalah: Not a bad idea at all :)
18:38:06 <mmarsale1> adetalhouet: #link https://git.opendaylight.org/gerrit/#/c/26641/3/catalog-api/src/main/yang/catalog.yang initial catalog model
18:38:07 <edwarnicke> ariel_noy: Also a good idea :)
18:38:32 <adetalhouet> For the workload manager, yeah, let's discuss the patch
18:39:04 <edwarnicke> mmarsale1: It looks like ariel_noy came in with a comment after the patch was merged asking about making network_function a grouping?  ariel_noy  Do you want to say more about that?
18:39:14 <mmarsale1> adetalhouet: #link https://git.opendaylight.org/gerrit/#/c/27182/ workload manager
18:39:27 <adetalhouet> mmarsale1: thanks, was actually looking for it
18:40:20 <edwarnicke> mmarsale1: You need to start the line with #link for it to show up correctly in the minutes :)
18:40:39 <adetalhouet> edwarnicke: he did
18:40:45 <adetalhouet> edwarnicke: I'm not doing it
18:40:51 <adetalhouet> so I'll do it
18:41:02 <alagalah|alt> Id like to hear more about nf grouping too
18:41:05 <edwarnicke> adetalhouet: I thought he started the line with " adetalhouet " ;)
18:41:16 <edwarnicke> ariel_noy: Could you say more about your thoughts there?
18:41:19 <adetalhouet> edwarnicke: you're right
18:41:28 <mmarsale1> edwarnicke: I think Ive answered the comment ... we need to decide how WM uses the Catalog
18:41:32 <adetalhouet> ok so let's start with workload manager
18:41:38 <adetalhouet> #topic workload manager
18:41:40 <edwarnicke> ariel_noy: It sounds like you'd like to use NF in the workload manager API... is there a reason you want to use it as a grouping rather than using an instance identifier to point to it?
18:41:47 <edwarnicke> adetalhouet: One momnet
18:41:51 <edwarnicke> #chair adetalhouet
18:41:51 <odl_meetbot> Current chairs: adetalhouet edwarnicke
18:41:56 <edwarnicke> Try setting the topic now :)
18:42:00 <adetalhouet> #topic workload manager
18:42:03 <edwarnicke> #chair mmarsale1 alagalah ariel_noy
18:42:03 <odl_meetbot> Current chairs: adetalhouet alagalah ariel_noy edwarnicke mmarsale1
18:42:04 <adetalhouet> edwarnicke: thxs
18:42:14 <adetalhouet> #link https://git.opendaylight.org/gerrit/#/c/27182/ workload manager patch
18:42:14 * edwarnicke is bad about remembering to chair folks
18:42:22 <ariel_noy> I would expect to get network function type and return an instance
18:43:07 <ariel_noy> did not know how to import network function object which is not in the top level
18:43:27 <edwarnicke> ariel_noy: So, I guess the question is... do you want to *refer* to a network function that is in the catalog
18:43:37 <edwarnicke> ariel_noy: Or are you wanting to repeat some of that information in your RPC call?
18:43:39 <alagalah|alt> So the desire is to be able to query by type?
18:43:59 <adetalhouet> Yes that is what I think alagalah|alt
18:44:17 <alagalah|alt> So a container per type ?
18:44:23 <edwarnicke> alagalah|alt: adetalhouet query *what* by type :)
18:44:27 <alagalah|alt> So you can wildcard query?
18:44:29 <ariel_noy> i would expect to be request to inistantiate a vnf from a type
18:44:29 * edwarnicke lost a beat
18:44:32 <alagalah|alt> Nf
18:44:35 <adetalhouet> edwarnicke: query the NF
18:44:53 <edwarnicke> adetalhouet: So query for what NFs we have in the catalog?
18:44:53 <ariel_noy> and to return an instance or to add the instance to the current running vm list
18:45:09 <ariel_noy> the catalog is definition and work load manager is the running
18:45:13 <edwarnicke> ariel_noy: Ah... I think I'm starting to get it... its more of a NF vs NF-instance thing
18:45:19 <edwarnicke> ariel_noy: Yep :)
18:45:43 <alagalah> ariel_noy, Could this be expressed in terms of CONF and OPER ?
18:45:57 <alagalah> alagalah, Or is it more a hierarchy thing ?
18:46:14 <edwarnicke> ariel_noy: So would it make sense for the catalogVNFInstance to have a nf-type which is pointing to the NF as defined in the catalog?
18:46:14 <mmarsale1> ariel_noy: so what if users configure the instances with an instance-identifier pointing to a specific NF in catalog ? and workload manager then adds the operation/status data
18:46:15 <ariel_noy> yes i agree when run get a request he should get a pointer to conf
18:46:30 <adetalhouet> I agree with mmarsale1
18:46:35 <adetalhouet> I think this is the way to go
18:46:36 <alagalah> ie ODL/config/armoury/nf/nf-type-foo/ returns all foos, ODL/config/armoury/nf/nf-type-foo/nf-type-foo-instance1 ?
18:46:51 <alagalah> mmarsale1, ack
18:46:57 * alagalah I think :D
18:46:58 <adetalhouet> alagalah: yes
18:47:04 * alagalah I always reserve the right to be wrong :)
18:47:34 <adetalhouet> so do I - really new to all this
18:47:51 <ariel_noy> THE workload manager start the VM but need to know which type to start
18:48:13 <adetalhouet> Yes, so by passing an Instance Identifier, it will know which NF to use
18:48:32 <edwarnicke> ariel_noy: Or collection of workloads (VMs/containers, etc...)
18:48:52 <mmarsale1> ariel_noy: user writes NF instance to a list (config data), Workload manager listens and from the input reads the catalog entry (using provided instance identifier) and after the NF is started, adds some operational data about its state
18:48:57 <edwarnicke> mmarsale1: Have you given thought of how to store info about the running instances of an NF in your catalog model?
18:48:58 <ariel_noy> Where this Instance come from if the workload did not create it?
18:49:03 <alagalah> ariel_noy, adetalhouet edwarnicke mmarsale1  This is why I suggested we start a websequence diagram
18:49:15 <edwarnicke> alagalah: You want to fire something up in google docs and point us to it?
18:49:29 <mmarsale1> edwarnicke: I think running instances should be separated from catalog
18:49:35 <alagalah> https://www.websequencediagrams.com/
18:49:35 <mmarsale1> edwarnicke: maybe part of workload manager
18:50:05 <edwarnicke> alagalah: How do we get to the specific one we are collaborating on?
18:50:07 <ariel_noy> The running should be in the workload manager and the definition of the config and drivers in the catalog
18:50:09 <alagalah> edwarnicke, ariel_noy mmarsale1  could we use https://www.websequencediagrams.com/ and save the text (you will know what I mean when you play with it) and email it around ?
18:50:27 <alagalah> Either email it around or put the text in a doc
18:50:42 <edwarnicke> ariel_noy: I'm fine with the workload manager being responsible for the instance info
18:51:00 <mmarsale1> edwarnicke: I think catalog should stay just a static list of images/configurations .... maybe plus a file server
18:51:11 <edwarnicke> ariel_noy: My personal off the cuff feeling though is that it should have an instance identifier that points back to the NF in the catalog for things that are already defined there
18:51:16 <ariel_noy> I am ok with web sequence diagram
18:51:27 <adetalhouet> edwarnicke: mmarsale1:  yes the catalog should just be a static list
18:51:54 <edwarnicke> adetalhouet: Well... maybe not *static*... but probably not instance by instance of a particular NF :)
18:51:54 <alagalah> title Armoury Be sequence
18:51:55 <alagalah> thing1->thing2: Who are you?
18:51:55 <alagalah> thing2->thing1: Does anyone really know who they are?
18:51:57 <ariel_noy> I see we all agree on the catalog
18:52:14 <edwarnicke> alagalah: How do we get an 'instance' of a websequence diagram we can point everyone to to collaborate on?
18:52:29 <alagalah> As I just showed... you keep the text... or else we all $$$ for subs
18:52:41 <alagalah> So we can collaborate in a google doc on the text...
18:53:09 <edwarnicke> alagalah: Maybe a google doc?
18:53:12 <alagalah> you just need to copy pasta it into WSD, make changes, paste back text changes.... take a screen shot of the pic has worked for me too for preso
18:53:18 <ariel_noy> We can put the text in the git?
18:53:31 <alagalah> ariel_noy, We could do that too, but do we want that in the project? Or just github ?
18:53:39 <yapeng> i like the git idea
18:53:41 <alagalah> We could make a public gist etc... lots of options...
18:53:47 <edwarnicke> alagalah: Would make sense to put it in the project I think...
18:53:55 <adetalhouet> alagalah: ariel_noy : what about a wiki page
18:53:56 <edwarnicke> alagalah: if we want to revision control it
18:53:57 <ariel_noy> Will be very nice to have doumentation as part of the project
18:53:58 <alagalah> alagalah, edwarnicke yapeng I think git > googledoc
18:54:20 <yapeng> in gerrit, everyone could make comments
18:54:23 <alagalah> yapeng, yes
18:54:42 <mmarsale1> ariel_noy: alagalah: agree with git ... we could also put this into asciidoc, there is very similar way of writing diagrams + asciidoc nicely renders on github, I think
18:54:43 <alagalah> ariel_noy, Great suggestion...
18:56:26 <ariel_noy> ok
18:56:59 <adetalhouet> ok so someone to initiate the diagram so we can work together on it?
18:57:15 <edwarnicke> #link http://asciidoctor.org/docs/asciidoctor-diagram/ <- I think this may help :)
18:58:37 <ariel_noy> #link https://www.websequencediagrams.com/  <- looks simpler
18:58:50 <mmarsale1> adetalhouet: I volunteer forthe diagram work
18:59:09 <adetalhouet> mmarsale1: thanks you very much mmarsale1 :)
18:59:17 <mmarsale1> adetalhouet: ariel_noy: will use the web sq diagrams to keep it simple for now
18:59:45 <adetalhouet> #action mmarsale1 to kick off the sequence diagram for NF catalog - workload manager interaction
19:00:24 <ariel_noy> Ok do you open a new bug on that to put it in the GIT?
19:00:59 <alagalah|alt> No need for a bug
19:01:10 <mmarsale1> ariel_noy: will put in trello
19:01:20 <alagalah|alt> Just pull repo, make change, push to gerrit
19:01:32 <ariel_noy> ok
19:01:39 <adetalhouet> #link https://trello.com/b/zsERKgeV/odl-armoury trello board
19:01:48 <alagalah|alt> If you need help ping me on irc
19:02:04 <adetalhouet> I will also be available for that :)
19:02:13 <alagalah|alt> But searching wiki for "pulling pushing" has a good tutorial
19:02:48 <alagalah|alt> Just dont skip the settings. Xml bit :)
19:03:20 <mmarsale1> I will start with something simple, push to git and notify on mailing list to start the discussion
19:03:31 <adetalhouet> ok great
19:04:00 <ariel_noy> great tx
19:04:22 <adetalhouet> we're 3 minutes past the hour so I guess we can call off the meeting for today
19:04:54 <adetalhouet> I'll come up with an agenda for next meeting to make it more efficient
19:05:01 <ariel_noy> Ok thanks and have a good day :)
19:05:20 <yapeng> thanks, bye now.
19:05:40 <adetalhouet> edwarnicke: anything to add before ending the meeting?
19:05:48 <edwarnicke> adetalhouet: Nope :)
19:05:53 <adetalhouet> #topic cookies
19:05:58 <adetalhouet> #endmeeting