17:31:37 <SumitNaiksatam> #startmeeting Networking Advanced Services
17:31:37 <openstack> Meeting started Wed Jul 30 17:31:37 2014 UTC and is due to finish in 60 minutes.  The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:31:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:31:41 <openstack> The meeting name has been set to 'networking_advanced_services'
17:31:46 <SumitNaiksatam> #info agenda: https://wiki.openstack.org/wiki/Meetings/AdvancedServices
17:32:01 <dougwig> o/
17:32:18 <SumitNaiksatam> enikanorov__: there?
17:32:59 <SumitNaiksatam> #info Juno 3, i believe is 2014-09-04 (#link https://launchpad.net/openstack/+milestones)
17:33:07 <enikanorov__> SumitNaiksatam: yes
17:33:17 <SumitNaiksatam> which means we can expect code freeze a couple of weeks prior
17:33:42 <SumitNaiksatam> so we should be aiming to be ready with our patches submitted say by aug 21st or thereabouts
17:34:05 <SumitNaiksatam> i think mestery will later annoce what the feature freeze deadline is
17:34:11 <SumitNaiksatam> *announce
17:34:22 <SumitNaiksatam> or he might have already, i missed it
17:34:30 <SumitNaiksatam> #topic Flavors
17:34:32 <banix> i think jun-3 is the feature freeze deadline
17:34:37 <banix> juno-3
17:35:01 <enikanorov__> so, we're still waiting for spec update, while i've pushed first step implementation and CLI for that
17:35:19 <enikanorov__> spec isn't approved still L:(
17:35:19 <SumitNaiksatam> banix: true, but usually we dont allow new features to land a little before that actualy release milestone date
17:35:27 <SumitNaiksatam> banix: so that poeple can review
17:35:36 <SumitNaiksatam> enikanorov__: did notice that
17:35:48 <SumitNaiksatam> enikanorov__: but you seem to have removed your patch out of WIP
17:35:57 <banix> SumitNaiksatam: yes by that day they should have landed meaning much earlier reviews as you mentioned
17:36:07 <enikanorov__> SumitNaiksatam: basically the flavors patch is ready for review
17:36:19 <enikanorov__> technically there are few things to discuss
17:36:28 <SumitNaiksatam> mestery: there?
17:36:36 <enikanorov__> most noticable is the way driver configuration is pushed
17:36:54 <enikanorov__> and it's mostly a question of what kind of API attribute to use for that
17:37:20 <mestery> SumitNaiksatam: Here, reading backscroll around flavors.
17:37:37 <SumitNaiksatam> mestery: hi, was just pinging you :-)
17:38:01 <SumitNaiksatam> mestery: so the my question for you was is it okay for the team here to review enikanorov__’s flavor’s implementation patch?
17:38:11 <SumitNaiksatam> mestery: i believe he is making good progress
17:38:33 <mestery> We really need that spec approved, that spec and the rootwrap ones are the only ones I'm currently going to allow exceptions for at this point.
17:38:36 <SumitNaiksatam> mestery: however the spec has not been updated since july 2nd to the best of my recollection
17:38:51 <mestery> But we need them to close fast, this week, as we're 5 weeks out from the end.
17:38:59 <SumitNaiksatam> mestery: agree
17:39:04 <enikanorov__> correct. that still waiting markmcclain's participation
17:39:33 <mestery> Lets see if we can close this by Friday
17:39:58 <SumitNaiksatam> mestery: ok, meanwhile let people review enikanorov__’s implementation patch as well?
17:40:19 <enikanorov__> SumitNaiksatam: i think some folks already started reviewing
17:40:59 <mestery> SumitNaiksatam: There is no harm in that, sure.
17:41:02 <SumitNaiksatam> enikanorov__: yes sure, just wanted to make sure we are all on the same page in terms of the process
17:41:06 <SumitNaiksatam> mestery: ok great
17:41:37 <SumitNaiksatam> i did not want to get into a situation where people spend a lot of time reviewing the implementation, and then we have to reset it because something else changes in the spec
17:41:54 <SumitNaiksatam> enikanorov__: definitely commend your persistence on this and for making progress
17:42:08 <SumitNaiksatam> mestery: thanks for that direction, and for jumping in :-)
17:42:11 <SumitNaiksatam> mestery: one more questions
17:42:15 <SumitNaiksatam> *question
17:42:29 <SumitNaiksatam> mestery: is Aug 21st the feature freeze deadline?
17:42:47 <mestery> Yes, that's correct.
17:43:17 <SumitNaiksatam> mestery: okay, so aug 21st is the last day that we can post a review for a feature related patch?
17:43:27 <mestery> Yes sir
17:43:42 <banix> wow so earlier than i thought
17:43:50 <SumitNaiksatam> mestery: ok great, thanks, wanted to clarify for the rest of the folks here, so there is no confusion
17:43:57 <SumitNaiksatam> mestery: bug fixes can go in after that, right?
17:44:03 <mestery> Yes
17:44:23 <SumitNaiksatam> mestery: ok, thanks
17:44:33 <mestery> SumitNaiksatam: your welcome :)
17:44:39 <SumitNaiksatam> mestery: :-)
17:44:45 <SumitNaiksatam> enikanorov__: sorry to distract from the focus on the flavors discussion
17:44:55 <banix> so nw features could get merged upto juno-3 or that is not correct?
17:44:56 <SumitNaiksatam> enikanorov__: are there any questions for enikanorov__?
17:45:06 <banix> new
17:45:13 <SumitNaiksatam> banix: yes they will, and until 09/04
17:45:33 <banix> so what is the aug 21st then?
17:45:38 <SumitNaiksatam> banix: however a patch can be only submitted against a blueprint by aug 21st
17:45:50 <SumitNaiksatam> banix: subsequently it will be in review until 09/04
17:45:57 <banix> not revised afterward… i see
17:46:03 <banix> ok thx
17:46:07 <SumitNaiksatam> banix: or it can be approved until 09/04
17:46:15 <SumitNaiksatam> banix: it can be definitely revised
17:46:43 <banix> ok got it thx
17:46:44 <SumitNaiksatam> banix: just that you can post a new patch that references a blueprint (so its an implementtion of a new feature) post aug 21st
17:46:55 <SumitNaiksatam> can -> cant
17:46:59 <banix> yup
17:47:25 <SumitNaiksatam> enikanorov__: hopefully you and the rest of the team can be out of the limbo by this weekend
17:47:33 <SumitNaiksatam> wishful thinking! :-)
17:47:39 <enikanorov__> that would be great
17:47:58 <banix> i think nova mid-cycle is ongoing so perhaps wont hear from Mrk right away
17:47:59 <SumitNaiksatam> enikanorov__: thanks for the update
17:48:34 <SumitNaiksatam> is s3wong here?
17:48:40 <s3wong> SumitNaiksatam: hello
17:48:43 <SumitNaiksatam> also SridarK?
17:48:50 <SumitNaiksatam> #topic Service base and insertion implementation update
17:49:09 <SumitNaiksatam> s3wong: hi, any update for the team on this topic?
17:49:17 <marios> so i have been disctracted this week by something else i'm involved in
17:49:18 <SumitNaiksatam> marios: as well
17:49:19 <SridarK> yes
17:49:24 <s3wong> Not much update for me, unfortunately. This past week and this week I am focusing on GBP - so nothing, sorry
17:49:28 <SumitNaiksatam> marios: ok, np
17:49:31 <marios> i haven't progressed my side as i wanted to (still wip patch)
17:49:41 <SumitNaiksatam> s3wong: okay
17:49:44 <marios> i kinda anticipated this hapening which is why i started the very early wip a few weeks back
17:49:49 <marios> hopefully i get some update for next week
17:49:52 <SridarK> SumitNaiksatam: just got back from PTO yesterday
17:49:53 <SumitNaiksatam> marios: good
17:50:00 <marios> and also looking forward to seeing the base class definitions
17:50:03 <marios> from whoever does that
17:50:09 <marios> so i can rebase onto
17:50:10 <SumitNaiksatam> SridarK: okay, so all of you conspired to not work! :-)
17:50:16 <SridarK> :-)
17:50:25 <LouisF> is kanzhe on?
17:50:25 <SumitNaiksatam> marios: yes, i agree that there is a dependency for you
17:50:39 <s3wong> SumitNaiksatam: have we made a decision on how the serviceInterfacePlugin will get the service instance UUID yet?
17:50:42 <SumitNaiksatam> LouisF: kanzhe is on vacation
17:50:50 <marios> SumitNaiksatam: (dependency yeah but definitely not a blocker by any measure at this point)
17:50:56 <SumitNaiksatam> marios: ok
17:51:14 <SumitNaiksatam> s3wong: i had some ideas on that, which i had run by the team last week
17:51:24 <LouisF> SumitNaiksatam: who else is working on service base and insertion?
17:51:41 <SumitNaiksatam> LouisF: kanzhe is expected to back on this soon
17:51:58 <SumitNaiksatam> LouisF: besides him, s3wong, marios, SridarK, and kevinbenton are working on it
17:51:59 <s3wong> marios: when it becomes available, the base class code will be here #link https://github.com/noironetworks/neutron-group-policy/tree/service-insertion
17:52:02 <SumitNaiksatam> LouisF: myself as well
17:52:11 <marios> s3wong: perfect thx mate
17:52:16 <SumitNaiksatam> s3wong: to your earlier questions
17:52:57 <SumitNaiksatam> s3wong: we can maintain a table in the new service insertion plugin, that will hold a reference to the uuids against their service type
17:53:37 <SumitNaiksatam> s3wong: and we can leverage the notification mechanism (say when a firewall is created, a notification is sent), to learn that a service is created, and then populate the table accordingly
17:54:08 <SumitNaiksatam> s3wong: the notifications already exist today, and are being used ceilometer, etc
17:54:25 <SumitNaiksatam> s3wong: i think kevinbenton had raised some issue with this, but it slips my mind
17:54:36 <marios> SumitNaiksatam: is that really for J though? ^^^
17:54:51 <SumitNaiksatam> marios: yes, we would need to do this
17:54:58 <SumitNaiksatam> marios: do you anticipate any issue?
17:55:19 <marios> i.e. is that the proposed implementation. i see. no i had thought (from the comments) that interfacing with the service plugins would come later.
17:55:23 <SumitNaiksatam> marios: this will have the least amount of code churn among other options
17:55:25 <banix> marios: the notification mechanism is there already
17:55:27 <marios> i guess this way, the plugins can still remain unaware
17:55:36 <SumitNaiksatam> banix: right
17:55:38 <marios> (of service insertion code/framework)
17:55:42 <SumitNaiksatam> marios: exactly
17:55:57 <SumitNaiksatam> marios: the other goal was to be as less intrusive as possible
17:56:03 <marios> ack
17:56:21 <s3wong> marios: that is the goal, we don't want every service to separately implement and maintain service interface info
17:56:24 <SumitNaiksatam> i think we will still need the services to implement the service base, but that might be minor refactoring of existing code
17:56:33 <SumitNaiksatam> s3wong: agree
17:57:22 <SumitNaiksatam> okay, anything else to discuss on this topic?
17:57:34 <SumitNaiksatam> for this week that is
17:57:54 <SumitNaiksatam> i think we need to churn out some code for the rest of the team to understand
17:58:04 <SumitNaiksatam> LouisF: you had some questions earlier?
17:58:15 <SumitNaiksatam> LouisF: not sure if we answered those
17:58:48 <SumitNaiksatam> ok moving on
17:58:53 <LouisF> yes thx
17:59:10 <SumitNaiksatam> hemanthravi: there?
17:59:22 <hemanthravi> SumitNaiksatam: hi
17:59:31 <SumitNaiksatam> #topic Service Chaining implementation update
17:59:48 <SumitNaiksatam> hemanthravi: do you have an update on this?
17:59:54 <SumitNaiksatam> i dont see songole
18:00:26 <hemanthravi> started working on the patch loosely based on GBP plugin/driver model and expect to have the first patch by Mon
18:00:37 <SumitNaiksatam> hemanthravi: ah good
18:00:38 <hemanthravi> songole here with me
18:00:45 <hemanthravi> this will be a WIP patch
18:00:52 <SumitNaiksatam> hemanthravi: okay, so who is working on this patch, you, songole, or both?
18:01:04 <hemanthravi> songole and me
18:01:23 <SumitNaiksatam> hemanthravi: okay, got it, just want to make sure that the rest of the team knows
18:01:52 <SumitNaiksatam> hemanthravi: are there are any blockers for you at this point?
18:02:20 <hemanthravi> not yet, will be pinging you with questions later this week
18:02:27 <SumitNaiksatam> okay
18:02:37 <SumitNaiksatam> any questions for hemanthravi on this topic?
18:03:18 <SumitNaiksatam> ok moving on then
18:03:38 <LouisF> when will the horizon work for chaining be done?
18:03:39 <SumitNaiksatam> i believe this is all we have approved in terms of blueprints for Juno
18:03:59 <SumitNaiksatam> lets shift gears to our features which are currently targeted for Kilo
18:04:11 <SumitNaiksatam> #topic Traffic steering update
18:04:15 <SumitNaiksatam> cgoncalv1s: hi
18:04:31 <cgoncalves> SumitNaiksatam: hi
18:04:45 <SumitNaiksatam> cgoncalv1s: i know this is a little difficult to discuss here since its a kind of a limbo situation
18:04:45 <hemanthravi> LouisF: haven't planned on horizon for chaining, will look at this as we make progress
18:04:52 <cgoncalves> not much to report this week again
18:05:04 <SumitNaiksatam> cgoncalves: no worries
18:05:27 <cgoncalves> SumitNaiksatam: internally we've been testing it, and myself been out of office on business travels
18:05:51 <SumitNaiksatam> cgoncalves: my suggestion is that we work towards a plan on what we want to do prior to the specs opening up for Kilo, so that we dont lose momentum here
18:06:00 <cgoncalves> next two weeks I'll be on vacation so don't expect much activity from my end
18:06:07 <SumitNaiksatam> cgoncalves: sure well deserved break! :-)
18:06:15 <cgoncalves> SumitNaiksatam: sure
18:06:33 <SumitNaiksatam> any questions or suggestions for cgoncalves before takes off on vacation? ;-P
18:07:04 <cgoncalves> I should keep lurking around, don't worry.
18:07:25 <SumitNaiksatam> cgoncalves: sure, i guess one of the suggestions would have been not to - enjoy your time off! :-)
18:07:28 <SumitNaiksatam> alright moving on
18:07:38 <cgoncalves> SumitNaiksatam: I also need to reach out to you and some other folks offline soon.
18:07:44 <SumitNaiksatam> cgoncalves: sure
18:07:48 <SumitNaiksatam> anil_rao: there?
18:07:54 <anil_rao> yes :)
18:08:02 <SumitNaiksatam> #topic Tap as a Service
18:08:07 <SumitNaiksatam> anil_rao: hi
18:08:20 <SumitNaiksatam> i dont see vinay_yadhav here
18:08:26 <anil_rao> Hi. Vinay can't join us today but I can give an update
18:08:34 <SumitNaiksatam> anil_rao: same questions as cgoncalves for you
18:08:41 <SumitNaiksatam> anil_rao: sure go ahead
18:09:07 <anil_rao> We are planing on how to proceed, most likely another small revision to the spec
18:09:15 <SumitNaiksatam> anil_rao: sure
18:09:16 <marios> https://review.openstack.org/#/c/96149/
18:09:23 <anil_rao> However, we need some help on how to address the objection to the *-aaS issue.
18:09:37 <marios> must.not.laugh.
18:09:39 <SumitNaiksatam> marios: thanks, sorry i should have been posting the links during the discussion
18:10:16 <SumitNaiksatam> marios: true, except that its not funny to hold up the spec for this reason! :-(
18:10:35 <marios> SumitNaiksatam: np (i just had it open so pasted). yes this is true.
18:10:56 <SumitNaiksatam> anil_rao: please proceed
18:11:16 <anil_rao> We are also looking into starting the implementation and uploading some initial code for review as WIP
18:11:20 <SumitNaiksatam> anil_rao: i do understand the motivation and need for choosing that name
18:11:31 <SumitNaiksatam> i belive the rest of the team is also aware of that
18:12:07 <marios> plus the *-aas issue is much wider reaching than tap-aas (as in the comments, lb, vpn, fw etc)
18:12:25 <SumitNaiksatam> marios: true, there is already a precedent!
18:12:38 <SumitNaiksatam> i believe the itention was to expose a rich set of features behind this service endpoint
18:12:48 <SumitNaiksatam> port-mirroring is the first capablity
18:13:19 <anil_rao> Perhaps we will make our intentions more explicit in the spec so that there is no confusion going forward
18:13:28 <SumitNaiksatam> anil_rao: sure
18:13:48 <anil_rao> One question if I may.
18:14:00 <SumitNaiksatam> the predicament for the authors here is that if they pack too many features they will be dinged for putting too much in
18:14:07 <SumitNaiksatam> anil_rao: please go ahead
18:14:53 <anil_rao> We are currently referencing an older BP called Port-Mirroring Extension. Perhaps that is what is causing the confusion. Do you all suggest that we drop that BP and create a new one for our Spec.
18:15:22 <SumitNaiksatam> anil_rao: ah, i did not realize that
18:16:00 <SumitNaiksatam> anil_rao: perhaps, however the new BP should have enough in it to suggest that is it sufficiently different or encompasses the existing one
18:16:23 <anil_rao> Ok.
18:16:26 <SumitNaiksatam> anil_rao: you have done the right thing by referencing and trying to reuse an existing body of work
18:16:45 <SumitNaiksatam> anil_rao: however if it is causing issues, perhaps something worth considering
18:17:06 <anil_rao> Ok. I'll discuss this with Vinay.
18:17:16 <SumitNaiksatam> anil_rao: it might also be good to send out an email to the -dev mailer explaining this explicitly
18:17:26 <anil_rao> Sure. That is a good idea.
18:17:32 <cathy_> SumitNaiksatam: Sorry to chime in late. May I ask one question. Regarding our comment about the direction of the chain on service chaining BP, I did not see a reply on the latest comment.
18:18:02 <cathy_> That is, supposing user request FW then IDS, the service chaining API reqires specifying the port for applying the service and the BP gives an example of a Neutron port between a router and network X. But without the direction, it is ambigous whether the traffic flows through Router->FW->IDP->Network X or the reverse sequence.
18:18:36 <SumitNaiksatam> cathy_: sure, i think we have some spare time
18:18:53 <SumitNaiksatam> cathy_: can you hold that question while we wrap up the Tap update?
18:19:01 <cathy_> sure
18:19:07 <SumitNaiksatam> anil_rao: anything else you wanted to discuss?
18:19:15 <SumitNaiksatam> any questions for anil_rao on Tap?
18:19:18 <anil_rao> No at this time. Thanks
18:19:26 <SumitNaiksatam> anil_rao: thanks
18:19:52 <SumitNaiksatam> #topic Open Discussion
18:20:03 <SumitNaiksatam> hemanthravi: still there?
18:20:14 <hemanthravi> yes
18:20:23 <SumitNaiksatam> cathy_: did you post the comment on the approved spec?
18:20:41 <marios> cathy_: i believe the list of nodes is intended to be ordered
18:20:51 <cathy_> yes, I posted. Then Louis posted on a new updated version
18:21:03 <SumitNaiksatam> marios: true, that was my understanding as well (and the original intent)
18:21:09 <SumitNaiksatam> cathy_: ah ok
18:21:12 <SumitNaiksatam> i dont see mandeep here
18:21:21 <marios> "This creates the ordered-list ["FW", "LB"] as the list of services in the	196 chain.
18:21:32 <SumitNaiksatam> hemanthravi: can you take a stab at answering cathy_’s question?
18:22:24 <hemanthravi> currently it's an ordered list and traffic in reverse will the list reversed
18:22:31 <SumitNaiksatam> cathy_: i believe your question is that you want only traffic from one direction to hit the service chain in that order?
18:22:45 <marios> the latest version actually is updated to be more explicit about the direction/ordering https://review.openstack.org/#/c/93524/12..13/specs/juno/service-chaining.rst
18:23:17 <SumitNaiksatam> marios: true
18:23:31 <cathy_> SumitNaiksatam: what is needed is the explicit, If the latest has that, we cna take a look.
18:24:07 <SumitNaiksatam> cathy_: ok, we can circle back to this if your query is not satisified
18:24:19 <cathy_> But from the API, how could the user spacify two different sequences of service chain (one the forward sequence, the other the reverse sequence)?
18:24:37 <LouisF> in the lates there is no defintion of what "ingress" and "egress" traffic is
18:24:48 <SumitNaiksatam> #action hemanthravi to report back on cathy_’s question “from the API, how could the user spacify two different sequences of service chain (one the forward sequence, the other the reverse sequence)?”
18:25:02 <SumitNaiksatam> LouisF: got it
18:25:25 <SumitNaiksatam> hemanthravi: see the above action item ^^^ :-)
18:25:37 <cathy_> SumitNaiksatam: hemanthravi : Thanks!
18:25:46 <hemanthravi> will do, need to work through some of the scenarios
18:25:52 <SumitNaiksatam> cathy_: np, thanks for brining it up
18:26:11 <SumitNaiksatam> we are getting close to our time here
18:26:19 <LouisF> hemanthravi: look at my comments on previous patches
18:26:33 <hemanthravi> LouisF: ok
18:26:41 <SumitNaiksatam> anything else we need to discuss?
18:27:20 <s3wong> 3 minutes early!!!
18:27:28 <marios> goodnight all :)
18:27:28 <SumitNaiksatam> s3wong: yay!
18:27:32 <SumitNaiksatam> thanks all for attending
18:27:36 <SumitNaiksatam> keep up the good work
18:27:38 <SumitNaiksatam> bye
18:27:39 <regXboi> @SumitNaiksatam: a last suggestion
18:27:42 <anil_rao> Thanks!
18:27:49 <SumitNaiksatam> regXboi: just in time
18:27:51 <SumitNaiksatam> regXboi: shoot
18:28:07 <regXboi> send an email out to openstack-dev with the link to the minutes and a note of any #agreed items
18:28:11 <regXboi> just to be friendly
18:28:15 <SumitNaiksatam> regXboi: ah ok
18:28:20 <regXboi> I'm going to send mail suggesting that across the board
18:28:25 <SumitNaiksatam> regXboi: did not have any “agreed” items today
18:28:33 <regXboi> having spent time cross-indexing minutes with ML
18:28:34 <SumitNaiksatam> regXboi: but good suggestion
18:28:43 <regXboi> thanks
18:29:02 <SumitNaiksatam> regXboi: i have tried not to bombard the mailing list with too many emails about routine items
18:29:12 <SumitNaiksatam> regXboi: since i think there is already an overload
18:29:23 <SumitNaiksatam> regXboi: but will definitely keep that in mind in the context that you mentioned
18:29:32 <regXboi> understood - but a note about a "hey we met, here are the minutes" keeps all abreast
18:29:33 <SumitNaiksatam> and right on dot
18:29:40 <SumitNaiksatam> regXboi: sure
18:29:44 <SumitNaiksatam> #endmeeting