15:01:30 <frankbrockners> #startmeeting BGS weekly team meeting
15:01:30 <collabot> Meeting started Mon Jun 29 15:01:30 2015 UTC.  The chair is frankbrockners. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:30 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:01:30 <collabot> The meeting name has been set to 'bgs_weekly_team_meeting'
15:01:38 <frankbrockners> #info Frank Brockners
15:02:03 <Jonas1> #info Jonas Bjurel
15:02:04 <arnaud_orange> #info Arnaud Morin
15:02:15 <frankbrockners> #info agenda for today https://wiki.opnfv.org/meetings/bgs#june292015
15:02:16 <David_Orange> #info David Blaisonneau
15:02:19 <fdegir> #info Fatih Degirmenci
15:02:28 <frankbrockners> anything you'd like to see added to the agenda?
15:02:48 <trozet> #info Tim Rozet
15:02:59 <Jonas1> Im fine with the Agenda
15:03:23 <frankbrockners> in a nutshell topics are: (a) see where we are with Arno deployment experiences and path towards SR1 and (b) status on Genesis project creation
15:03:34 <bryan_att> #info Bryan Sullivan
15:03:48 <bryan_att> is there a webex for today?
15:04:12 <frankbrockners> bryan_att - no webex, IRC only as usual
15:04:23 <frankbrockners> let's get going
15:04:27 <bryan_att> FYI its in the calendar invite I still have
15:04:29 <frankbrockners> #topic Arno updates: Deployment experiences and progress towards Arno SR1
15:05:07 <frankbrockners> let's just briefly recap where we are, i.e. immediate plans for the next weeks, open issues, etc
15:05:16 <radez> #info Dan Radez
15:05:21 <frankbrockners> Tim or Jonas - do you want to give  a brief update?
15:05:37 <Jonas1> I can start?
15:05:58 <frankbrockners> bryan_att - sorry for the stale calendar entries
15:06:05 <frankbrockners> Jonas1 - yes please
15:06:33 <Jonas1> #info We have demonstrated the new ODL plugin which is planned for SR1. It was demonstrated last Thursday.
15:07:09 <Jonas1> #info We are also done with the Autodeploy refinements for SR1
15:07:42 <frankbrockners> Jonas1 - excellent news. Do you have a link to the recording, in case others want to take a look?
15:07:58 <Jonas1> #We will now start needed Fuel 6.1 rebasing, and needed refactoring of the build system .
15:08:13 <Jonas1> #That concludes what is planned for SR1.
15:08:22 <Jonas1> All from me.
15:08:35 <frankbrockners> Jonas1 - you forgot the "info"
15:08:55 <chenshuai_> new ODL plugin? lithium?
15:08:57 <Jonas1> #info We will now start needed Fuel 6.1 rebasing, and needed refactoring of the build system .
15:09:12 <Jonas1> #info That concludes what is planned for SR1.
15:09:21 <Jonas1> Now Im done:-)
15:09:30 <trozet> #info for SR1 we are looking at fixes to deploy, much more stable virtual deployment along with putting ODL helium in a more stable places
15:10:34 <frankbrockners> trozet - do you also plan to make deployments which happen to be behind a proxy easier?
15:10:57 <trozet> yes we need to incorporate pbandzi's changes
15:11:06 <frankbrockners> cool
15:11:17 <trozet> #info plan on migrating everything to genesis
15:11:23 <trozet> #info before SR1
15:11:36 <Jonas1> What is the plans for ODL: SR2 or SR3?
15:12:01 <trozet> Jonas1: SR2 or SR3 for opnfv?
15:12:06 <frankbrockners> You mean which Helium release?
15:12:29 <Jonas1> frankbrockners: exactly
15:12:43 <trozet> oh
15:12:52 <trozet> Jonas1: we use SR3
15:13:06 <trozet> Jonas1: lithium is out, so we are moving to that for next release
15:13:12 <Jonas1> trozet: Then we're aligned :-)
15:13:17 <trozet> great
15:13:41 <frankbrockners> #info OPNFV Arno SR1 plans to use ODL Helium SR3
15:13:54 <Jonas1> trozet: But not Lithium for Arno SR1, I believe?
15:14:11 <trozet> Jonas1: thats right
15:14:15 <frankbrockners> last time we agreed that we don't include major upgrades from upstream
15:14:27 <frankbrockners> into OPNFV service releases
15:14:30 <frankbrockners> ok - cool
15:14:58 <frankbrockners> from an overall schedule perspective, are we still good with the idea to have a SR in September?
15:15:12 <frankbrockners> That would mean that we have the code in place sometime in August
15:15:23 <trozet> yeah
15:15:45 <Jonas1> frankbrockners: I believe we are good for September - not telling you which day though :-)
15:15:53 <RandyLevensalor> Is there a target date for the sometime in August for code complete?
15:16:10 <frankbrockners> Jonas1 - Did I ask for a date? ;-)
15:16:12 <uli-k> Hi, sorry for jumping in, but can you explain to me the difference between maintenance and SR?
15:16:22 <frankbrockners> uli-k - same
15:16:42 <uli-k> So only bug fixes ....
15:16:49 <frankbrockners> yup
15:17:08 <uli-k> Then I misunderstood some of the dialogue....
15:17:23 <frankbrockners> RandyLevensalor - let's define a target once we're closer - for now let's have "code complete in August" as objective
15:17:55 <frankbrockners> any other updates on "work in progress for Arno fixes"?
15:18:38 <RandyLevensalor> frankbrockners, ok...The only reason ask is that it may be good to have some time scheduled for people not involved in the development to walk through the install after code complete and before SR1.
15:18:49 <trozet> +1
15:18:56 <Jonas1> What about the discussions on LF lab? I heard that one of the PODs will be removed and used for build, etc.
15:19:23 <frankbrockners> Jonas1 - good point. Does anyone have the latest on those discussions?
15:19:44 <fdegir> frankbrockners: Jonas1: ChrisPriceAB sent an email asking for feedback last week
15:20:09 <ChrisPriceAB> We can put it to the TSC this week if you like fdegir
15:20:11 <fdegir> but no feedback as of yet so we're waiting to get a conclusion
15:20:15 <Jonas1> fdegir: Didnt catch that
15:20:21 <fdegir> ChrisPriceAB: that would be good since there was no objection
15:20:21 <frankbrockners> fdegir - so you have a link to the email?
15:20:31 <fdegir> frankbrockners: looking at it now
15:21:02 <ChrisPriceAB> http://lists.opnfv.org/pipermail/opnfv-tsc/2015-June/001023.html
15:21:19 <fdegir> #link: http://lists.opnfv.org/pipermail/opnfv-tech-discuss/2015-June/003396.html
15:21:24 <frankbrockners> #info Discussion on LF lab usage
15:21:59 <fdegir> Jonas1: was talking about ChrisPriceAB's mail as linked above
15:22:09 <fdegir> so if we go ahead with this
15:22:20 <fdegir> we will need to do some tricks to use hw more cleverly
15:22:26 <frankbrockners> IMHO it does make sense what Chris proposes - and it is inline with earlier plans
15:22:41 <frankbrockners> will also allow us to run build on LF infra
15:22:48 <fdegir> and virtualized deployments
15:23:02 <fdegir> which is wanted by trozet and many others to verify patches from deployment point of view
15:23:09 * ChrisPriceAB will add it to the Agenda for tomorrow
15:23:23 <frankbrockners> thanks ChrisPriceAB
15:23:42 <Jonas1> Why would we not run nested deployments in donated labs?
15:24:10 <fdegir> we have very few functioning community labs
15:24:12 <frankbrockners> Jonas1 - I don't think the email says this, does it?
15:24:34 <fdegir> and they might be tied to "specific" activities
15:25:03 <frankbrockners> ChrisPriceAB - Could you ask the Pharos team for an update on "functioning community labs"?
15:25:22 <frankbrockners> it would be great if we could deployments to community labs going as well
15:25:22 <fdegir> perhaps I can clarify what I mean with "functioning"
15:25:24 <Jonas1> For LF, IMHO I do we need to think bigger in order to utilize our labs better. Some kind of MaaS orchestration is inevitable.
15:25:30 <ChrisPriceAB> I can ping trevor_intel on it.  But he's off this week I think.
15:25:44 <fdegir> the labs that do not go down suddenly without an announcement
15:26:03 <fdegir> and provide SLA kind of info
15:26:31 <frankbrockners> fdegir - do we have those things captured somewhere (i.e. on some pharos page)?
15:26:32 <ChrisPriceAB> Lab availability is an issue to raise to Pharos.  Some form of SLA should be in place
15:26:43 <fdegir> frankbrockners: not that I know of
15:26:59 <fdegir> there was a discussion within pharos some time ago
15:27:08 <fdegir> but it might not have been finalized perhaps
15:27:11 <frankbrockners> fdegir - interested in resurrecting that discussion?
15:27:27 <fdegir> I can poke Trevor, Wenqing and the rest
15:27:38 <frankbrockners> great - thanks fdegir
15:27:41 <fdegir> np
15:28:12 <frankbrockners> #info Everyone: Please review Chris email - will be on the TSC agenda for tomorrow
15:28:39 <frankbrockners> #info fdegir will check with Pharos team to associated SLA-expectations with available community labs
15:28:56 <frankbrockners> any other topic?
15:29:07 <frankbrockners> (associated with Arno / SR)?
15:29:07 <Jonas1> Again, I think we should take a larger grip on this before we decide to change
15:29:47 <frankbrockners> ChrisPriceAB - your thoughts? Should we up-level the discussion and wait another week?
15:30:05 * ChrisPriceAB reads backwards
15:30:25 <frankbrockners> We can also do things step-wise: Decide on LF infra first and then expand to field labs
15:30:45 <frankbrockners> i.e. first we move to sequential deployment on LF infra
15:30:52 <ChrisPriceAB> We have the SR discussion ongoing with a propsal to the TSC, and the lab topic up as well.
15:30:55 <frankbrockners> then we do virtual deployments on LF infra
15:31:05 <frankbrockners> then we move Jenkins built to LF etc. etc
15:31:05 <ChrisPriceAB> I suggest we put an action on Pharos from the TSC on Lab availability
15:31:52 * frankbrockners thinks that it might be better to start with a discussion in the TSC and then decide in a week or two
15:32:30 <Jonas1> frankbrockners: Have no problem with starting in LF. I would however like that Pharos comes up with a strategy/proposal how we easier can allocate resources dynamically.
15:32:47 <frankbrockners> Jonas1: 100% agreed
15:33:12 <frankbrockners> stepwise approach does not remove the need for some for of resource management
15:33:35 <frankbrockners> (even if we start with a spreadsheet-style approach day 1)
15:33:52 <frankbrockners> so let's see what Pharos comes back with
15:33:54 <Jonas1> frankbrockners: On the same page then
15:34:54 <frankbrockners> #info BGS team expects that Pharos team will create some proposal (ideally incl. suggestions for a tool) for managing resources in LF lab and community labs
15:35:05 <frankbrockners> should we move to Genesis discussion?
15:35:57 * frankbrockners takes silence as a "yes"
15:36:07 <frankbrockners> #topic Genesis as evolution of BGS
15:36:35 <frankbrockners> #info There is a draft project page for Genesis: https://wiki.opnfv.org/genesis
15:37:06 <frankbrockners> #info The project pages includes the draft project proposal: https://wiki.opnfv.org/genesis/genesis_project_proposal
15:37:45 <frankbrockners> #info 3 installers have already voiced their plan to join Genesis: Apex, OpenSteak, Fuel@OPNFV
15:38:45 <frankbrockners> #info Expectation is that Genesis plus the different installer projects replace current BGS. I.e. once we have the installer projects off the ground, we can close BGS.
15:39:43 <frankbrockners> #info Plan is to officially create Genesis once the participating installer projects exist
15:39:55 <frankbrockners> so likely in a few weeks
15:40:16 <chigang> Compass will join Genesis later,  has interest, too.
15:40:23 <arnaud_orange> and then genesis will be entry point to select installer?
15:40:23 <frankbrockners> any thoughts/comments?
15:40:27 <arnaud_orange> from user perspective I mean
15:41:00 <uli-k> I think it is important, we can make all installer projects join Genesis.
15:41:01 <frankbrockners> arnaud_orange: Genesis will just create requirements. Users still have the choice to choose the installer they want
15:41:33 <frankbrockners> uli-k - agreed. the broader the support the better
15:41:35 <RandyLevensalor> What about repot management?  With the breadth of installers, it seems like each installer would have a dedicated repo and only put in some ci hooks and shared code in the genesis repo.
15:41:53 <uli-k> Installers out of Genesis will be bad for OPNFV.
15:41:55 <frankbrockners> RandyLevensalor - yes this is the plan
15:42:01 <arnaud_orange> does genesis will let them know which installer to choose based on a selector?
15:42:29 <arnaud_orange> like: If I want puppet modules, let's use foreman or opensteak
15:42:30 <uli-k> I don't think that should be a goal of Genesis
15:42:30 <RandyLevensalor> frankbrockners, great.  that wan't clear from the proposal.  Mind if I update the proposal to clarify that.
15:42:33 <frankbrockners> arnaud_orange: we can formulate this as a common requirement if we see this as needed
15:42:40 <arnaud_orange> ok
15:43:09 <fdegir> RandyLevensalor: the current genesis repo is structured in that way - have <installer_name/ci directories for all installers
15:43:24 <fdegir> RandyLevensalor: and CI executes those scripts without needing to know the specific details
15:43:38 <frankbrockners> RandyLevensalor: Right now the proposal reads as "Maintain common artifacts used by all deployment tools (e.g. build, deploy, or configuration scripts): Evolve and maintain “common” tree of existing genesis repository. Artifacts which are specific to an installer will be moved into the repository of the associated installer project."
15:43:47 <frankbrockners> isn't this clear enough?
15:44:24 <RandyLevensalor> It doesn't say anything specific about the none common parts, which mainly exist under /install name directories.
15:45:01 <frankbrockners> ok - feel free to update if you have better wording in mind
15:45:02 <RandyLevensalor> Maybe the term artifacts is throwing me.  I think of artifacts as not source.
15:45:42 <frankbrockners> if you have a better term.. please :-)
15:45:59 <RandyLevensalor> As in using the artifact repo for external binary dependencies.  Will do..thx for the clarification and +1 to the plan
15:46:13 <Jonas1> One thing on common. Common repos maintained by Genesis is not mandatory to use. They maybe used for a while by an instaler, and after a while they might be absorbed by the installer upstream.
15:47:01 <frankbrockners> Jonas1 - good point - we do want to keep as little as possible "local" to OPNFV
15:47:35 <arnaud_orange> frankbrockners: what do you mean?
15:48:32 <arnaud_orange> do you mean have the different project repo as small as possible?
15:49:02 <frankbrockners> arnaud_orange: "if upstream adopts changes (e.g. in the Fuel case, they integrate ODL natively) then we don't have to maintain a local copy in OPNFV any more"
15:49:16 <arnaud_orange> oh, yes
15:49:19 <Jonas1> arnaud_orange: OPNFV is a mid stream project and should ultimately carry as few patches as possible
15:49:34 <arnaud_orange> what about the github repo?
15:49:38 <arnaud_orange> like the one from trozet
15:49:54 <arnaud_orange> there is some duplicates between tim repo and genesis at the moment
15:50:09 <frankbrockners> trozet already mentioned plans earlier to move all this stuff into OPNFV for Arno SR1
15:50:11 <Jonas1> arnaud_orange: No private repos, but comunity maintained
15:50:13 <frankbrockners> trozet?
15:50:17 <trozet> yeah
15:50:22 <arnaud_orange> should the code be included into a OPNFV repo, or keep it as much as possible externally?
15:50:34 <arnaud_orange> Jonas1: ok
15:50:36 <trozet> going to move bgs_vagrant to Genesis
15:50:49 <arnaud_orange> so most of the "glue " should stay in OPNFV
15:51:02 <frankbrockners> IMHO if it is "OPNFV stuff", it should be in an OPNFV repo
15:51:07 <trozet> anything not upstream tool specific
15:51:18 <trozet> anything created specifically for OPNFV (like bgs_vagrant) should be in opnfv
15:51:31 <trozet> any upstream tool modification should remain in the upstream tool
15:51:33 <frankbrockners> if it is "stuff that is used by many - and OPNFV is just one consumer" then it should live upstream - outside OPNFV
15:51:43 <arnaud_orange> ok got it
15:52:01 <frankbrockners> cool - we all say the same thing now :-)
15:52:08 <fdegir> just to state the obvious - we must have tags/labels on upstream stuff for release
15:52:11 <arnaud_orange> thanks for the clarification
15:52:22 <fdegir> not latest please
15:52:23 <frankbrockners> fdegir - good point
15:52:31 <arnaud_orange> +1
15:53:14 <RandyLevensalor> Would we start pulling more of the upstream dependencies into the artifacts repo? To ensure they are available and we use a consistent version or is the goal to go straight to the source and hope nothing breaks on the external dependencies?
15:53:56 <arnaud_orange> i think it really depends on the dependency
15:54:22 <arnaud_orange> do you have any exemple in mind?
15:54:27 <frankbrockners> RandyLevensalor: Do you want to make LF a proxy for all upstream dependencies we have?
15:55:44 <RandyLevensalor> frankbrockners, I just would like to see a guideline.  But pulling stuff from 6 different repos makes me a little unsettled. Especially pulling code that isn't in a signed package.
15:56:31 <trozet> RandyLevensalor: we tried to version things for Arno release and pull specific upstream versions, for CI env like Octopus we want latest
15:57:33 <RandyLevensalor> But it is difficult to unwind when debugging.  If I knew a file either came from OPNFV source or a package, it would only take a few minutes to trace down the source and see if there are any relevant patches or bugs.
15:58:34 <RandyLevensalor> We can table this discussion.
15:59:38 <frankbrockners> RandyLevensalor: Probably worth a wider discussion. It might also help get some user feedback on their expectations. Would you mind starting an email discussion on the topic (i.e. to have LF host a proxy for upstream repos)?
15:59:58 <RandyLevensalor> happy to
16:00:10 <frankbrockners> thanks RandyLevensalor
16:00:20 <frankbrockners> ... which brings us to the top of the hr.
16:00:35 <frankbrockners> Thanks everyone.... and have a great (in many cases short) week!
16:00:41 <frankbrockners> #endmeeting