16:01:38 <Sukhdev> #startmeeting networking_ml2
16:01:39 <openstack> Meeting started Wed Jul 16 16:01:38 2014 UTC and is due to finish in 60 minutes.  The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:43 <openstack> The meeting name has been set to 'networking_ml2'
16:02:00 <Sukhdev> #topic Agenda
16:02:07 <banix> Hi
16:02:22 <Sukhdev> #info: https://wiki.openstack.org/wiki/Meetings/ML2#Meeting_July_16.2C_2014
16:03:00 <Sukhdev> #topic: Announcements
16:03:22 <Sukhdev> Spec Approval deadline for Juno is 7/20
16:03:30 <Sukhdev> I hope everybody knows about it
16:04:02 <Sukhdev> So, if you have any comments to address, please move on it and let us know if you need our help in getting it pushed
16:04:10 <trinaths> Sukhdev: is the deadline for J-2 or for J relase
16:04:11 <shivharis> can we ping cores to do the reviews for specs?
16:04:27 <Sukhdev> trinaths: for Juno
16:04:39 <Sukhdev> shivharis: please do
16:05:04 <rkukura> Sukhdev: Do we know which spec are definitetly deferred to K, and which still need review to make J?
16:05:30 <Sukhdev> rkukura: I did not have a chance to filter those
16:05:34 <rkukura> Sukhdev: I guess we can cover that under spec reviews
16:06:01 <trinaths> Sukhdev: how to we know specs are deffered to K release.
16:06:26 <rkukura> trinaths: I think mestery has -2’ed some that he doesn’t think will make J
16:06:30 <Sukhdev> trinaths: If one is not approved by 7/20 - it will be moved to K release
16:06:56 <shivharis> 7/20 is this Sunday
16:07:07 <rkukura> sorry for bringing this up now, lets cover the spec reviews when we get to them on the agenda
16:07:08 <Sukhdev> shivharis: Oh well
16:07:28 <Sukhdev> Next announcement -
16:07:44 <Sukhdev> Third Party CI systems must be operational by 7/24
16:07:52 <Sukhdev> We are all impacted by this
16:07:56 <banix> i think for those not approved the owner should move under k when that opens up
16:08:13 <Sukhdev> Please be aware of this
16:08:58 <Sukhdev> In case you did not know you can influence the name of K-release
16:09:18 <Sukhdev> Vote here -  https://www.surveymonkey.com/s/openstack-k-naming
16:09:42 <banix> the point is the move of a spec to K won’t happen automatically but i am sure we will hear about it … sorry for late messages here. moving on.
16:10:21 <Sukhdev> #info rkukura, kmestry, banix and I were in MN for mid-cyle sprint
16:10:56 <Sukhdev> Lots of good stuff took place
16:11:18 <Sukhdev> check out this wiki - https://wiki.openstack.org/wiki/Neutron/MigrationFromNovaNetwork/HowTo
16:11:29 <Sukhdev> If you want to test VM migration
16:11:41 <Sukhdev> banix is expert in this -
16:11:56 <Sukhdev> Anybody has any other announcement?
16:12:00 <banix> i feel i migrated my life ;)
16:12:20 <Sukhdev> banix: you migrated mine as well :-)
16:12:28 <banix> Sukhdev: hahaha
16:12:40 <Sukhdev> OK moving on
16:12:49 <Sukhdev> #topic:  Bugs
16:12:59 <Sukhdev> shivharis: floor is yours
16:13:18 <shivharis> I need status from the bugs listed in the agenda
16:13:31 <shivharis> we can go one at a time..
16:13:52 <shivharis> https://bugs.launchpad.net/neutron/+bug/1331857
16:13:56 <uvirtbot> Launchpad bug 1331857 in neutron "ML2 Nexus MD does not appear to configure multiple switches" [High,Confirmed]
16:14:20 <shivharis> nlahiuti: ?
16:14:36 <Sukhdev> shivharis: Looks like Cisco internal issue
16:14:38 <nlahouti> yes
16:15:09 <shivharis> this shows up as high... is that correct?
16:15:24 <nlahouti> shivharis: I though it was low
16:15:34 <rcurran> it should be junked
16:15:55 <shivharis> rcurran: can you please
16:15:57 <nlahouti> rcurran: I was going to talk to you about it
16:16:09 <shivharis> next
16:16:10 <Sukhdev> rcurran nlahouti: can one of you do the honors please
16:16:11 <nlahouti> rcurran: so it is not valid bug?
16:16:24 <shivharis> https://bugs.launchpad.net/neutron/+bug/1246737
16:16:26 <uvirtbot> Launchpad bug 1246737 in neutron "ML2 plugin deletes port even if associated with multiple subnets on subnet deletion" [Medium,Confirmed]
16:16:59 <shivharis> padkrish: ?
16:16:59 <rcurran> i commited that fix in I ... but i think it didn't get in until j-1
16:17:20 <rcurran> i'll update bug
16:17:28 <shivharis> rcurran: thanks
16:17:35 <nlahouti> rcurran: thx
16:17:36 <padkrish> shivhari# It got committed around Apr, i guess...
16:17:55 <shivharis> padkrish: it is about back port
16:17:57 <padkrish> #shivhari# So, should it be back-ported to havana? that's what it said
16:18:20 <shivharis> how far do we want to go? havana is way old
16:19:09 <shivharis> can we update the bug and way that we will not fix and reopen if someone eeally needs it
16:19:24 <shivharis> way->say
16:19:29 <yamamoto> https://bugs.launchpad.net/neutron/+bugs?field.tag=ml2 might be better than https://bugs.launchpad.net/neutron?field.searchtext=ml2
16:19:44 <Sukhdev> rkukura: what do you think?
16:19:46 <padkrish> shivhari# ok
16:19:58 <rkukura> Is the fix in both trunk and icehouse already?
16:20:20 <shivharis> I have two other bugs listed as review needed, can someone we review internally in ML2
16:20:39 <shivharis> banix, sukhdev: ?
16:20:59 <banix> shivharis: will do after the meeting
16:21:11 <Sukhdev> shivharis: ditto
16:21:11 <shivharis> ok thanks. thats all i have for today.
16:21:14 <irenab> there is another bug for ml2 https://bugs.launchpad.net/neutron/+bug/1338202
16:21:16 <uvirtbot> Launchpad bug 1338202 in neutron "ML2 plugin update_port changes binding:profile when not changed" [Undecided,Confirmed]
16:21:50 <shivharis> irene: j2 or j3?
16:22:10 <banix> Also want to point out the DVR work makese many changes to ML2 code. Please keep reviewing them.
16:22:14 <irenab> please add it to the list, I am planning to submit fix in coming days
16:22:21 <rkukura> I had discussed this with irena, and it sounds like a real issue
16:22:34 <shivharis> irenab: will track that
16:22:43 <irenab> shivharis: thanks
16:23:00 <Sukhdev> Anything else on the bugs?
16:23:03 <rkukura> irenab: should we target J2 or J3?
16:23:04 <shivharis> irenab: sorry typo in name before
16:23:20 <irenab> shivharis: not a problem
16:23:35 <irenab> I will try till j2
16:23:49 <rkukura> banix: I will be reviewing the DVR changes to ML2 this week - other should as well
16:24:06 <Sukhdev> rkukura: I will as well
16:24:20 <Sukhdev> OK moving on
16:24:23 <shivharis> same
16:24:27 <banix> rkukura: thanks; Sukhdev: thanks. reasonably significant changes. mostly additions but some changes
16:24:42 <Sukhdev> #topic: Spec Review
16:25:10 <Sukhdev> Like I mentioned at the top of the meeting deadline for Juno is 7/20
16:25:58 <Sukhdev> Please ensure that as an owner of a spec - address the comments as soon as possible and wait until the last moment
16:26:14 <rkukura> we really need to focus on the specs of general interest that have a chance of making Juno, but also need to get vendor-specific specs approved
16:26:22 <Sukhdev> shivharis: I gave a -1 on your spec this morning - please be sure to check it
16:26:38 <shivharis> Sukhdev: will take a look...
16:27:43 <Sukhdev> rkukura: https://wiki.openstack.org/wiki/Tracking_ML2_Subgroup_Reviews#Under_Review has priorities on them
16:28:21 <rkukura> Sukhdev: I’ll update that with the hierarchical port binding spec, which is now in review
16:28:33 <Sukhdev> Lets make sure the high priority ones get attention right away
16:28:45 <Sukhdev> rkukura: thanks
16:29:02 <Sukhdev> Anything on the Specs?
16:29:04 <rkukura> I’m also working with nlahouti to update the extension spec to add a minimal ExtensionDriver API
16:29:16 <irenab> Can we discuss code review requests now?
16:29:20 <nlahouti> rkukura: yes.
16:29:35 <nlahouti> rkukura: hopefully we can make the deadline
16:30:04 <rkukura> nlahouti: I can work with you this afternoon on the update
16:30:15 <Sukhdev> irenab: I forgot to put on the agenda - do you have something specific in mind?
16:30:17 <rcurran> rkukura, what's the status on https://review.openstack.org/#/c/106185/1
16:30:26 <nlahouti> rkukura: thx I'll ping you. What time is good ?
16:30:45 <rkukura> rcurran: It needs review, I’ll add it to the wiki
16:31:19 <Sukhdev> rkukura: please do - I am reviewing specs from the wiki only :-)
16:31:27 <rkukura> nlahouti: 3-5 EDT
16:31:42 <Sukhdev> irenab: ?
16:31:44 <nlahouti> rkukura: will do that.
16:31:58 <shivharis> irenab: any specific code review you have in mind?
16:32:11 <shivharis> irenab: please add to the wiki
16:32:12 <irenab> the following patch for approved spec: https://review.openstack.org/#/c/74464/
16:32:42 <irenab> shivharis: thanks, will add to the wiki
16:32:59 <Sukhdev> irenab: please be sure to add to wiki
16:33:21 <Sukhdev> OK - moving on
16:33:29 <irenab> Sukhdev: thanks, will do
16:33:29 <Sukhdev> #topic: ML2 Sync
16:33:57 <Sukhdev> Folks I added this topic on the agenda as I wanted to spend couple of minutes on this topic
16:34:16 <Sukhdev> #link https://docs.google.com/document/d/17fATwZkJEonH0pIb1-mPD0UB5RKnJzcHYqkBesJhirE/edit
16:34:42 <Sukhdev> We decided to defer it to K - release
16:35:25 <Sukhdev> But, we should allocate some time to discuss this - as this will help, in general, all ML2 drivers in terms of dealing with the error states
16:35:42 <Sukhdev> The proposal on the document is just a draft
16:35:52 <Sukhdev> to kick off the thought process
16:36:01 <shivharis> this ends up deferring more dependent stuff to K :-(
16:36:40 <Sukhdev> shivharis: it is a bigger task, hence, we decided to defer it
16:36:46 <shivharis> k
16:36:53 <shivharis> i mean ok
16:37:07 <banix> i ewill add the modular l2 agent spec to the wiki as well; again sorry for being late!
16:37:12 <Sukhdev> There are several ways to go about implementing it
16:37:48 <Sukhdev> We would need volunteers to work on ML2 Sync
16:37:54 <shivharis> modular l2 already has an implementation?
16:38:15 <Sukhdev> shivharis: not really
16:38:45 <Sukhdev> banix, shivharis: you reviewed the doc - want to add anything?
16:38:47 <rkukura> Sukhdev: Goal should be to have a solid ML2 sync proposal for the K summit, right?
16:39:05 <Sukhdev> rkukura: that would be great
16:39:27 <banix> Sukhdev: nothing beyond what you said
16:40:27 <Sukhdev> Folks, please review the document and add your comments to it
16:40:32 <Sukhdev> Moving on
16:40:43 <Sukhdev> #topic Provide node specific information to ML2 plugin
16:40:55 <Sukhdev> Who added this to the agenda?
16:41:01 <pczesno> Hi
16:41:10 <Sukhdev> pczesno: floor is yours
16:41:14 <pczesno> adding to the agenda was suggested by kmestry
16:41:38 <Sukhdev> pczesno: please explain it to the team
16:41:47 <pczesno> we want to support dpdk enabled ovs with mech_openvswitch and mech_odl
16:42:09 <pczesno> dpdk enabled ovs requires different vif bindig as normal ovs
16:42:36 <pczesno> during binding we would need to pass some additional info to nova
16:42:51 <pczesno> that dpdk enabled ovs is running on the node
16:43:45 <rkukura> pczesno: I recall seeing the binding:vif_type and binding:vif_details attributes mentioned in the email thread. Is setting these when your MD binds sufficient, or is some additional mechanism needed in ML2 and/or nova?
16:43:53 <pczesno> question is if there is a mechanism in place to do that or if not what approach would be the best
16:43:56 <banix> pczesno: does the nova understand the new info?
16:44:29 <pczesno> rkukura: puting info in vif_details would be good
16:44:46 <irenab> pszesno: you can get the is_dpdk_ovs info from L2 agent, and ther send it to nova in vif_details
16:45:08 <pczesno> yes we could get that info from l2 agent
16:45:19 <pczesno> but we want it to work with odl as well
16:45:25 <rkukura> pczesno: I believe you can do that now, and it will be available to nova’s generic VIF driver
16:46:03 <pczesno> rkukura: can you elaboratye
16:46:47 <pczesno> so we would like l2 plugin to gather host specific info
16:46:50 <rkukura> pczesno: The path from MD to VIF driver should be covered already. With no agent, I think the MD would need to get node-specidic info from config or from some DB.
16:46:59 <pczesno> l2 agent or odl could provide info
16:47:30 <sean-k-mooney> rkukura: are you refering to the vhostuser vif that will be added? if so this is not compatable
16:47:48 <sean-k-mooney> sorry pressed enter too early
16:48:04 <rkukura> pczesno:  your MD’s bind_port() ask ODL?
16:48:36 <pczesno> rkukura: is that a sugestion or question? :)
16:48:53 <pczesno> rkukura: yes it could be done that way
16:48:53 <rkukura> sean-k-mooney: I wasn’t referring to a specific VIF driver, just pointing out that the vif_type and vif_details set by the MD that binds the port should be available to nova’s VIF driver
16:49:32 <rkukura> pczesno: Is having bind_port() ask ODL a workable solution?
16:49:49 <pczesno> rkukura: i think it could be
16:50:03 <rkukura> BTW, the “bind ports outside transactions” patch merged yesterday, so making remote calls within bind_port() is now allowed.
16:50:24 <Sukhdev> rkukura: yes I saw that
16:50:34 <pczesno> rkukura: but it would be better if rpc was started on odl side
16:50:59 <rkukura> pczesno: Do you mean an RPC from ODL to neutron?
16:51:10 <pczesno> rkukura: yes
16:51:36 <rkukura> Do you mean specifically the AMQP-based RPC, or just a REST invocation on a neutron API?
16:52:08 <pczesno> rkukura: ampq based rpc,  i think this already happens for port state updates
16:52:58 <rkukura> pczesno: I wasn’t aware of RPC interfaces being exposed outside neutron itself. Even nova uses REST APIs to talk to neutron. Are you sure?
16:53:11 <pczesno> rkukura: not sure
16:53:21 <Sukhdev> Time check - 5 min.
16:53:40 <pczesno> rkukura: ok, will try come up with spec
16:53:43 <rkukura> Should we move this discussion back to the email thread?
16:53:50 <pczesno> rkukura: yes
16:54:04 <rkukura> pczesno: Feel free to ping me on IRC if you want to brainstorm on this
16:54:05 <Sukhdev> pczesno: I think writing up a spec will be a good idea
16:54:26 <Sukhdev> OK - folks, time to move on
16:54:26 <pczesno> rkukura: could this be considered for J?
16:54:37 <Sukhdev> #topic: Open Discussion
16:54:54 <Sukhdev> Any body wants to discuss anything?
16:55:20 <yamamoto> nothing from me
16:55:22 <rkukura> pczesno: If you’ve got an approved BP for an MD that can use the existing ML2 capabilities, I don’t see any reason it can’t make J.
16:55:25 <padkrish> #sukhdev# is the deadline for code review for approved specs July 24?
16:55:36 <padkrish> or is it for submitting the code for review?
16:55:38 <shivharis> all: Please focus on spec reviews, Sunday deadline
16:56:03 <banix> july 24 is J-2 deadline
16:56:05 <Sukhdev> padkrish: no - for that you have plenty of time - I believe 8/20
16:56:21 <pczesno> rkukura: thanks for your time
16:56:24 <banix> that would be for J-3
16:56:41 <banix> have the code in by early august to be safe
16:56:47 <padkrish> #sukhdev# 8/20 is J-3?
16:56:57 <Sukhdev> padkrish: correct
16:56:57 <nlahouti> even for the J-2 approved BP?
16:57:17 <banix> J-3 is that last deadline for new code (unless a bug fix)
16:57:41 <Sukhdev> here is the release schedule - https://wiki.openstack.org/wiki/Juno_Release_Schedule
16:57:55 <nlahouti> banix: so it means for any approved BP code must be committed by 8/20?
16:58:07 <banix> nlahouti: earlier
16:58:09 <mxu> Hi,  earlier <irenab> asked  can we discuss code review requests now?  Can we review this now?
16:58:21 <banix> J-3 is Aug 4th
16:58:28 <Sukhdev> nlahouti: by 8/20 the code should be submitted - you can still continue to work on it
16:58:45 <Sukhdev> nlahouti: drop dead date is 9/4
16:59:12 <Sukhdev> mxu: shoot
16:59:24 <banix> New code won’t get in after J-3; be aware. plan accordingly.
16:59:32 <banix> unless as an exception
16:59:51 <mxu> what is the code request process?
16:59:52 <banix> So generally after J-3 is for bug fixes not new code
17:00:20 <padkrish> Thanks sukhdev, banix
17:00:23 <mxu> sorry, mistyped, what is the  code review process to get approval for  commit?
17:00:24 <Sukhdev> mxu: please submit the patch for review, add it to the wiki and we will review it
17:00:34 <banix> nlahouti: i think 8/20 is an internal deadline for marking specs as approved or not
17:00:35 <Sukhdev> Out of time guys
17:00:36 <Sukhdev> bye
17:00:40 <Sukhdev> #endmeeting