17:01:56 <SumitNaiksatam> #startmeeting Networking Advanced Services
17:01:56 <glebo> yo
17:01:56 <openstack> Meeting started Tue Dec  2 17:01:56 2014 UTC and is due to finish in 60 minutes.  The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:57 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:02:00 <openstack> The meeting name has been set to 'networking_advanced_services'
17:02:04 <SumitNaiksatam> pc_m: SridarK glebo: hi
17:02:11 <johnsom> o/
17:02:11 <SumitNaiksatam> #info https://wiki.openstack.org/wiki/Meetings/AdvancedServices#Agenda
17:02:46 <SumitNaiksatam> copy pasting from the neutron IRC for the benefit who missed it -
17:02:47 <SumitNaiksatam> #info SPD is 12-8-2014 and SAD is 12-15-2014
17:02:53 <SumitNaiksatam> #info Kilo-1 is 12-18-2014
17:03:24 <sbalukoff> Could you expand those acronyms please?
17:03:29 <SumitNaiksatam> any other info/announcements people would like to share? (apart from the services’ split discussion, of course)
17:03:40 <SumitNaiksatam> Spec Proposal Deadline
17:03:43 <dougwig> Spec approval deadline, and the first is the submission deadline (not sure on the P)
17:03:46 <SumitNaiksatam> Spec Approval Deadline
17:03:46 <sbalukoff> Thanks!
17:04:17 <blogan> it'll be a SAD day
17:04:28 <xgerman> because you spec doesn't make it?
17:04:33 <sbalukoff> badum-tish!
17:04:40 <banix> not knowing what that is, probably no
17:04:48 <SumitNaiksatam> blogan: exactly - seemed like SAD had become common neutron lingo
17:04:58 <banix> sorry; wrong window
17:05:53 <sunil_> hello everyone!
17:05:58 <SumitNaiksatam> okay next topic - our raison d'être
17:06:22 <SumitNaiksatam> #topic Advanced services’ spin out
17:06:26 <SridarK> SumitNaiksatam: seems Paris has left its impression on u :-)
17:06:43 <SumitNaiksatam> SridarK: oh good catch!
17:06:47 <SumitNaiksatam> Services' split: #link https://review.openstack.org/#/c/136835
17:07:21 <SumitNaiksatam> and there is a somewhat related spec which factors into the extensions’ discussions: #link https://review.openstack.org/#/c/136760
17:07:35 <SumitNaiksatam> to summarize for those who missed the meeting in the morning
17:07:40 <dougwig> please note the imminent changes that are coming, based on today's neutron meeting:
17:07:42 <dougwig> https://www.irccloud.com/pastebin/5qRvFm1X
17:08:12 * glebo yeah, thinks 06:00 is PAINFUL, but was commuting through inches of rain at the time.
17:08:14 <SumitNaiksatam> dougwig: thanks, i was just about to summarize that, seems like you had it prepared ;-)
17:08:27 <SumitNaiksatam> glebo: yeah, commute was pretty painful
17:08:35 <dougwig> that was my gerrit comment when I did -A just a few ago.  :)
17:09:11 <SumitNaiksatam> dougwig: you mean workflow-a
17:09:12 <xgerman> neat
17:09:17 <SumitNaiksatam> *workflow-1
17:09:43 <trinaths1> Hi
17:10:01 <SumitNaiksatam> dougwig: those read like executive decisions ;-P
17:10:23 <dougwig> i realized after hitting submit that i should've put, "from the neutron meeting..."  :)
17:10:33 <mestery> SumitNaiksatam: That's because they were discussed and agreed to in the neutron meeting today
17:10:38 <mestery> dougwig: ++
17:10:52 <SumitNaiksatam> mestery: yes sure, just pulling dougwig’s leg here
17:10:58 <mestery> lol
17:11:14 <dougwig> with both the split and flavors, it's a good thing that i'm secretly a spider.
17:11:31 <sbalukoff> Wait, are we now getting to accuse dougwig of being a merciless tyrant?
17:11:38 <SumitNaiksatam> mestery: glad you could make it to this meeting as well
17:11:50 * mestery is omniscient
17:12:16 <SumitNaiksatam> sbalukoff: that we thought was already established, no? ;-)
17:12:40 <glebo> mestery:  lol
17:12:43 <sbalukoff> As long as I get to carry around a pitchfork and torch, I'm happy.
17:13:02 <SumitNaiksatam> okay, so any thoughts feedback on those points, or anything else in the latest spec that we need discuss today/here?
17:13:03 <sbalukoff> Aaanyway.
17:13:05 <xgerman> sbalukoff likes to be the only merciless tyrant in town
17:13:08 <glebo> sbalukoff:  wait, we already passed halloween, no?
17:13:20 <sbalukoff> xgerman: +1
17:13:48 <sbalukoff> I feel like the current spec is pretty close to what we need to get progress to happen.
17:14:00 <xgerman> +1
17:14:01 <SumitNaiksatam> sbalukoff: okay
17:14:20 <sbalukoff> As such, if you've got objections, please make specific comments to that effect soon!
17:14:37 <xgerman> and if not +1
17:14:41 <dougwig> the work needs to start next week at the latest, so we need to get things solid.
17:14:43 <sbalukoff> xgerman: +1
17:14:59 <xgerman> yep, originally split was aimed for k1
17:15:07 <xgerman> and we can still make it
17:15:10 <dougwig> i don't mind some rework later, but if you've got something major, please don't wait.
17:15:28 <trinaths1> I'm new here then, How to propose a spec here?
17:15:34 <sbalukoff> And make time *now* to review if you haven't done so and feel like you have a horse in this race.
17:15:46 <SumitNaiksatam> trinaths1: hi, same as neutron spec
17:15:50 <dougwig> trinaths1: our specs are neutron specs
17:16:12 <xgerman> there is a wiki somewhere explaining the process
17:16:41 <SumitNaiksatam> is pc_m here?
17:16:46 <pc_m> yes sir
17:16:47 <trinaths1> okay. How will they be pointed to Advanced services?
17:17:14 <SumitNaiksatam> trinaths1: thats not needed, its all one neutron track
17:17:21 <trinaths1> okay.
17:17:37 <trinaths1> SumitNaiksatam: understood.
17:17:39 <SumitNaiksatam> #topic Services’ impact by L3 agent refactoring
17:17:58 <SumitNaiksatam> this would impact fwaas and vpnaas
17:18:10 <SumitNaiksatam> pc_m: you want to summarize where we are at with this?
17:18:15 <pc_m> sure
17:18:35 <pc_m> Carl has a BP out for the effort (I think 131535)
17:19:08 <pc_m> I did a commit for review to start the observer pattern to allow agents to register with L3 agent to get notifications for events.
17:19:24 <pc_m> 136549
17:19:57 <pc_m> Assaf has refined it some, and will do a updated patch that merges his with mine. Will have it tomorrow or Thur I suspect.
17:20:13 <pc_m> Currently, we're trying to work out the notifications.
17:20:14 <SumitNaiksatam> #link https://review.openstack.org/#/c/131535/ (for the spec)
17:20:43 <pc_m> Seems like there may be some variation as to where notifications will occur.
17:21:02 <pc_m> Will need help from FW folks to see about where hooks are needed for that agent.
17:21:06 <SumitNaiksatam> #link https://review.openstack.org/#/c/136549/ (for pc_m’s patch)
17:21:31 <SridarK> pc_m: i volunteer
17:21:43 <pc_m> I think Assaf's is #link https://review.openstack.org/#/c/137668
17:21:51 <pc_m> SridarK: thanks!
17:22:22 <pc_m> We can hook up offline. Looking at router add/delete/update and where agents need to be notified, for agent specific actions.
17:22:27 <glebo> pc_m: i can also have Gary Duan from our team take a look
17:22:29 <SridarK> pc_m: there are few places that notifications are needed esp on router stuff
17:22:37 <SridarK> pc_m: yes exactly
17:22:51 <pc_m> glebo: super! The more the merrier.
17:23:21 <pc_m> I think we want to get Assaf's and my commits merged, and then it is a matter of working out the notification points and refine the notification method.
17:23:24 <SumitNaiksatam> pc_m: so this does not propose to move away from the l3 agent inheriting from the fwaas agent?
17:23:37 <hareeshp> pc_m: Is is the plan to still go for multiple agents (per service) or service specfic classes somehow loaded by a single agent?
17:24:00 <pc_m> SumitNaiksatam: I think it will split them up. We're trying to tackle it piecemeal.
17:24:08 <SumitNaiksatam> pc_m: ah ok
17:24:18 <SumitNaiksatam> hareeshp: yeah, my earlier question
17:24:26 <hareeshp> SumitNaiksatam: yes ;)
17:24:46 <pc_m> I posted on ML (a few weeks back) some steps Carl and I discussed, and I think Carl has it in his BP.
17:24:54 <carl_baldwin> SumitNaiksatam: It is proposed to move away from the inheritence for FW.  Very soon.
17:25:02 <SumitNaiksatam> carl_baldwin: okay
17:25:17 <SumitNaiksatam> pc_m: so your patch is the first one to be reviewed in this chain, right?
17:25:57 <pc_m> SumitNaiksatam: Sort of... Assaf's is dependent on mine. There are a few others as well.
17:26:14 <pc_m> Carl has one to split up the L3 agent into modules.
17:26:24 <SumitNaiksatam> ah there is a topic for this:
17:26:26 <SumitNaiksatam> #link https://review.openstack.org/#/q/status:open+project:openstack/neutron+branch:master+topic:bp/restructure-l3-agent,n,z
17:26:48 <carl_baldwin> As soon as we identify how FW needs to be notified and what callbacks are needed on the l3_agent and router then we should be able to break the inheritence relationship.
17:26:52 <pc_m> Take a look at the BP first of all, so that we can get closure on carl_baldwin's BP.
17:27:07 <dougwig> at the end of this discussion, i just want to make sure that my naive assumption that this work exists in parallel to the split is actually true, and we don't have to block on each other.
17:27:14 <SumitNaiksatam> carl_baldwin: pc_m: sounds good
17:27:38 <SridarK> carl_baldwin: currently it is mostly on the router_interface_add
17:27:48 <pc_m> dougwig: good question, and concern
17:28:05 <SridarK> so we should be able to do this with some minimal restructure of the code
17:28:16 <SridarK> (FW agent code)
17:28:21 <SumitNaiksatam> dougwig: i would tend to think that this refactoring of the agent has to be accomplished prior to the fwaas and vpnaas split
17:28:54 <SumitNaiksatam> so that the respective agents can move with the services when the split happens
17:28:57 <dougwig> what is the timeline for that?  alternately, i'm ok with a temporary circular dependency with the l3 agent
17:29:16 <dougwig> temporary == never shipped.
17:29:31 <SumitNaiksatam> dougwig: yeah i agree, if there is a workaround, we should consider that
17:29:33 <xgerman> also since we are now splitting into 3 repos do we eed to split all at once?
17:29:43 <pc_m> dougwig: We're hot on it now... not sure how long to tease it all apart.
17:29:47 <carl_baldwin> dougwig: This is currently our highest priority.  The bp just merged yesterday and so we should working on it in earnest.
17:30:31 <dougwig> we have this, split, rest refactor, and vendor split, all aimed straight at each other.  anything we can do to sever dependencies will make the impending management nightmare easier.
17:30:44 <xgerman> +1
17:30:57 <SridharRamaswamy> dougwig: +1
17:31:00 <sbalukoff> +1
17:31:19 <s3wong> +1
17:31:24 <johnsom> +1
17:31:28 <SumitNaiksatam> agreed, so it might help to get some of these things in first, so that we dont have too many loose ends after the split
17:31:28 <trinaths1> +1
17:31:38 <pc_m> +1
17:31:49 <dougwig> SumitNaiksatam: that is the opposite of severing dependencies.  :)
17:31:59 <trinaths1> (confused) what does a split mean?
17:32:06 * pc_m +1 to dougwig's comment
17:32:08 <SumitNaiksatam> dougwig: not all dependencies are equal
17:32:17 <dougwig> trinaths1: https://review.openstack.org/#/c/136835/
17:32:43 <pc_m> So REST refactoring starts next week at the mid-cycle?
17:32:56 <dougwig> anytime you say "before" or "after", it is a dependency, and adds risk, IMO.
17:33:04 <SumitNaiksatam> mestery: if you are still around, is there a timeline we are targeting for the services’ repo split?
17:33:34 <trinaths1> dougwig:  thanks
17:34:02 <dougwig> it's next week, unless the rest refactor makes it impossible.
17:34:04 <xgerman> SumitNaiksatam at the summit we said K1
17:34:09 <carl_baldwin> dougwig: Did you have a plan for moving FW out of tree without breaking the inheritence hierarchy?
17:34:13 <xgerman> I would still aim for that
17:34:16 * carl_baldwin has not completely read the bp.  Sorry.
17:34:23 <SumitNaiksatam> that would help to evaluate if its feasible at all to address any percieved dependenices for the split
17:34:25 <dougwig> carl_baldwin: yes, but you'll need a bucket handy.
17:34:43 <pc_m> :)
17:35:30 <dougwig> carl_baldwin: the short-form would be treating the classes/repos as if they're not separate for the very short-term.
17:35:56 <trinaths1> dougwig: okay. thanks for the clarity.
17:36:02 <dougwig> then the fw team can refactor to independence on their own timeline.
17:36:05 * carl_baldwin wonders where he left that bucket.  ;)
17:36:13 <dougwig> i did warn you.
17:36:49 <mestery> SumitNaiksatam: The work will be done next week in SLC
17:37:03 <carl_baldwin> So, initially, the two repos are co-dependent?
17:37:08 <SumitNaiksatam> so the idea would be that neutron would depend on adv services’ repo, and adv services’ repo would depend on neutron?
17:37:21 <SumitNaiksatam> carl_baldwin: yeah, you summarized it better :-)
17:37:23 <dougwig> carl_baldwin: until the l3 refactor is done, yes.  that's the only circular link that i know of.
17:37:44 <carl_baldwin> dougwig: I think that is fine for the (very) short term but I’ll be looking again for the bucket if Kilo ships like that.
17:37:55 <dougwig> i don't think we can allow kilo to ship like that.
17:38:06 <SumitNaiksatam> i guess we would need the co-dependency would be needed regardless of the L3 agent dependency issue
17:38:33 <SumitNaiksatam> * i guess the co-dependency would be needed regardless of the L3 agent dependency issue
17:38:48 <dougwig> only for the upgrade scenario.  the plan is to have a hacking rule to prevent any other circular usages.
17:38:53 <dougwig> with an exception for l3 for now
17:39:22 <SumitNaiksatam> dougwig: okay
17:40:02 <SumitNaiksatam> so if the spin out is planned for the next week, i am guessing that is probably too short a time frame to achieve the l3 agent refactoring
17:40:16 <carl_baldwin> dougwig: okay.  Does it mean more complexity in the build / test environment to support this co-dependent thing?
17:40:42 <pc_m> SumitNaiksatam: yup
17:40:43 <dougwig> no.  the circular link is already there for the upgrade scenarios.  it just means abusing it slightly for fun and profit.
17:40:51 <dougwig> the only extra work is the exception to the new hacking rule
17:42:20 <carl_baldwin> dougwig: okay.  I’m okay going forward with that.  We can rebase our work to react to it.
17:42:21 <SumitNaiksatam> perhaps we can set the services’ agents (fwaas/vpnaas) as a target for K2, that way we are more or less guaranteed to get it done in Kilo
17:43:12 <SumitNaiksatam> if it lands earlier, great, but we can proceed with the split regardless
17:43:13 <dougwig> i'll include this refactor and breaking this link in the split timeline.  for K2 as SumitNaiksatam suggests?
17:43:39 <SumitNaiksatam> dougwig: i think that will be good to call out
17:43:45 <xgerman> +1
17:43:50 <dougwig> will do.
17:44:03 <SumitNaiksatam> dougwig: at least it identifies what we need to follow up on after the split as an immediate requirement
17:44:04 <dougwig> look for a spec with a detailed roadmap by lunchtime.
17:44:09 <carl_baldwin> SumitNaiksatam: I’d like to target an initial break out of the inheritence hierarchy for a couple weeks from now.  I’m not sure how much work will be needed to get to where you would like the service agents to be.
17:44:10 <SumitNaiksatam> dougwig: thanks
17:44:42 <SumitNaiksatam> carl_baldwin: agreed, i think, like SridarK mentioned, on the fwaas side we dont anticipate it to be as much
17:44:44 <carl_baldwin> I’m focused on detangling the L3 agent first.  Service agents second.
17:44:55 <SumitNaiksatam> perhaps pc_m can talk to for the vpn side of things
17:44:56 <carl_baldwin> SumitNaiksatam: dougwig: +1
17:45:11 <SridarK> SumitNaiksatam: carl_baldwin: yes should not be too bad on FW
17:45:34 <SridarK> I have probab jinxed it by saying that. :-)
17:45:37 <SumitNaiksatam> okay we have 15 mins left, and we haven’t ye reached “flavors” :-)
17:45:43 <SridharRamaswamy> SumitNaiksatam: pc_m: i can help out if any vpn work spills over
17:45:43 <pc_m> I don't think there is too much on VPN. only a handful of notifications.
17:45:46 <sbalukoff> Heh!
17:45:58 <SumitNaiksatam> SridharRamaswamy: pc_m: okay thanks
17:46:04 <SumitNaiksatam> #topic Flavors
17:46:11 <SumitNaiksatam> #link https://review.openstack.org/#/c/102723
17:46:23 <SumitNaiksatam> the spec has been repurposed from Juno ^^^
17:46:29 <sbalukoff> It's back from the grave!
17:46:38 <dougwig> and it wants your brains
17:46:39 * sbalukoff prepares his pitchfork and torch again
17:46:46 <dougwig> err, eyes.
17:46:48 <dougwig> i meant eyes.
17:46:53 <sbalukoff> Yes, eyes.
17:47:00 <SumitNaiksatam> dougwig: thanks for resurrecting it in time for the SPD
17:47:07 <johnsom> Mmmm, the flavor...
17:47:32 <dougwig> it's roughly similar to mark's spec from juno.  did anyone have major issues with that one?
17:47:43 <dougwig> and by roughly, i mean almost entirely identical.
17:47:45 <xgerman> I loved it@@
17:47:50 <SumitNaiksatam> not to rat hole into names, but i believe there were some suggestions to drop the use of “flavor” terminology in general in OpenStack
17:48:24 <dougwig> do we have a good alternative?
17:48:27 <SumitNaiksatam> if those still persist, perhpas they will show up early in the review
17:48:47 <SumitNaiksatam> dougwig: not sure, i was just trying to look up the threads that i recall reading a week or so back
17:48:56 <sbalukoff> If we want this approved before the SAD date, those reviews / objections better come quick!
17:49:04 <SumitNaiksatam> sbalukoff: agree
17:49:17 <dougwig> sbalukoff: +100
17:49:20 <SumitNaiksatam> dougwig: will look up and post if i can dig it out
17:49:26 <dougwig> SumitNaiksatam: thanks
17:49:54 <SumitNaiksatam> anything more of “flavors”?
17:50:05 <SumitNaiksatam> not sure if people had a chance to read it since it was resurrected
17:50:22 <sbalukoff> Well, they should take the opportunity now. :)
17:50:24 * pc_m unfortunately not yet
17:50:42 <SumitNaiksatam> ok moving on
17:51:07 <SumitNaiksatam> #topic Open Discussion (and any FWaaS/LBaaS/VPNaaS updates)
17:51:24 <dougwig> for lbaas updates, scroll up one hour to the meeting minutes.  :)
17:51:30 <SumitNaiksatam> dougwig: :-)
17:51:31 <sbalukoff> Haha
17:51:35 <trinaths1> :)
17:51:42 <sunil_> quick admin question: do we have a separate mailing list for advanced services?
17:51:55 <SumitNaiksatam> sunil_: not really
17:52:03 <trinaths1> SumitNaiksatam: When can I propose the FW Spec for K?
17:52:04 <SumitNaiksatam> sunil_: in fact there isnt one for neutron either
17:52:16 <SridarK> SumitNaiksatam: FWaaS putting together spec for FW insertion on router - should have it out by tomorrow
17:52:19 <SumitNaiksatam> trinaths1: you can and should before SPD (dec 8th)
17:52:20 <SridharRamaswamy> my understanding is the BPs on the flight need to be reproposed against advanced services spec repo ? Is this correct ?
17:52:38 <SumitNaiksatam> SridharRamaswamy: there isnt a separate repo
17:52:40 <xgerman> there is only one spec repo
17:52:55 <SumitNaiksatam> okay, just to clarify for everyone’s benefit -
17:53:00 <dougwig> SridharRamaswamy: no, only neutron-specs, and only re-propose if it was from juno and needs to be for kilo
17:53:11 <SumitNaiksatam> if you need propose a spec, please propose a spec in Neutron Specs, nothing has changed
17:53:12 <trinaths1> SumitNaiksatam: Okay, Here is my spec https://review.openstack.org/#/c/126187/
17:53:13 <glebo> re: FW insertion…
17:53:21 <glebo> how are we doing on the use cases?
17:53:30 <glebo> I've lost track since I went on TG break
17:53:40 <trinaths1> SumitNaiksatam: please consider it for K.
17:53:42 <SumitNaiksatam> if you need to send an email regarding advanced services, you send the email to the -dev ML, and prefix with:
17:54:15 <SridharRamaswamy> SumitNaiksatam: okay, I've a vpn spec already in neutron-spec/kilo #link https://review.openstack.org/#/c/136693/
17:54:16 <SumitNaiksatam> [Neutron][advanced-services] or [Neutron][LBaaS] or [Neutron][FWaaS] or [Neutron][VPNaaS]
17:54:26 <SumitNaiksatam> trinaths1: SridharRamaswamy: great
17:54:45 <glebo> SridarK:  I'll keep my eyes open for the insertion on router spec
17:54:47 <trinaths1> SumitNaiksatam: :) reviews bless the specs..
17:54:48 <SumitNaiksatam> on the fwaas side, SridarK will be posting a new spec for the insertion
17:54:54 <SumitNaiksatam> glebo: ^^^
17:54:57 <trinaths1> * reviewers
17:54:57 <SridharRamaswamy> dougwig: this is a new spec for Kilo
17:55:01 <SumitNaiksatam> glebo: we sill need the use cases :-)
17:55:30 <SridarK> glebo: yes have sent a reminder to FW folks on that
17:55:31 <SumitNaiksatam> pc_m: SridharRamaswamy: anything you would like to add for VPNaaS that is relevant to the rest of the adv services’ team here?
17:55:44 <pc_m> nothing from me.
17:56:05 <SridharRamaswamy> nothing from me, waiting for all the refactors to fly in :)
17:56:10 * pc_m still trying to get unburied from vacation and 4 days of no internet
17:56:18 <SumitNaiksatam> altighty, if nothing else, we can call it a wrap and take back  4 mins!
17:56:26 <pc_m> who hoo
17:56:29 <SumitNaiksatam> thanks all for joining!
17:56:30 <SumitNaiksatam> bye
17:56:33 <trinaths1> bye all
17:56:35 <SridarK> bye
17:56:37 <dougwig> bye
17:56:38 <SridharRamaswamy> bye all
17:56:41 <pc_m> bye
17:56:41 <SumitNaiksatam> #endmeeting