16:00:50 <Sukhdev> #startmeeting ironic_neutron
16:00:51 <openstack> Meeting started Mon Jul  6 16:00:50 2015 UTC and is due to finish in 60 minutes.  The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:52 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:56 <openstack> The meeting name has been set to 'ironic_neutron'
16:01:20 <Sukhdev> Good morning folks - ready for the meeting?
16:01:27 <Sukhdev> #topic: Agenda
16:01:40 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/Ironic-neutron
16:01:53 <Sukhdev> #topic: Announcements:
16:02:00 <Sukhdev> I have one -
16:02:30 <Sukhdev> I was eavesdropping on Ironic weekly meeting and found out about Ironic mid-cycle
16:02:45 <Sukhdev> It is scheduled for Aug12-14 in Seattle
16:03:07 <Sukhdev> In case anybody is planning to attend - FYI
16:03:36 <amotoki> hi
16:03:40 <Sukhdev> Anybody has any other announcements
16:03:43 <Sukhdev> amotoki: hello
16:04:10 <Sukhdev> well, lets dive into the agenda then
16:04:22 <Sukhdev> #topic: Spec Reviews
16:04:47 <Sukhdev> Lets cover this spec first - https://review.openstack.org/#/c/188528/
16:05:19 <Sukhdev> amotoki, kevinbenton had some comments on this spec, which we clarified
16:05:40 <jroll> so quick question
16:05:45 <jroll> "What Sukhdev, Kevin and I discussed is all about Neutron API. I believe we three have a consensus and all details will be covered by the devref as Sukhdev commented. As ironic-spec I am fine with the current version, so I vote +1."
16:05:59 <jroll> is it ML2 data structure details covered in the devref or?
16:06:35 <jroll> it seems fine, just making sure I understand correctly
16:06:40 <amotoki> I think we can cover other document (devref, wiki or similar one)
16:06:51 <jroll> well, I want to know which details are covered in the devref
16:07:13 <Sukhdev> jroll: one of the concerns was if something changes on ironic side or ML2 side, how do we coordinate what is needed - hence, we needed some kind of agreement
16:07:22 <Sukhdev> as to what is required
16:07:27 <amotoki> jroll: what I would like to cover is Neutron REST API defail and ML2 impl details.
16:07:39 <jroll> is it "fields other than switch_id, port_id, switch_info" covered there?
16:07:41 <jroll> ok
16:08:16 <amotoki> jroll: I think so.
16:08:27 <jroll> sounds good.
16:08:36 <amotoki> In the current spec, we describe what information is required but the detail data structure is not covered. It will be covered later.
16:08:38 <Sukhdev> jroll: I plan on writing a small document - for ML2 integration - i.e. how to use these fields, etc. and how to write/integrate ML2 drivers for Ironic
16:08:41 <amotoki> and the impact will be small.
16:08:45 <jroll> I still need to do a full review on this spec but I think it's probably good to go from what I've seen
16:08:48 <jroll> yep. +1
16:09:29 <lauramoore> hi sukhdev +1
16:09:50 <Sukhdev> jroll: I think from the neutron integration point of view are good - we need some reviews from the ironic side as well to get this blessed
16:10:01 <jroll> Sukhdev: indeed
16:10:03 <amotoki> Sukhdev: +1. I can collaborate writing the detail in Neutorn side.
16:10:15 <viveknarasimhan> Sukhdev: If i understand correctly: how to use the fields is only a recommendation for ML2 driver writers
16:10:23 <jroll> Sukhdev: we should be able to start writing code now, though, no need to wait for spec to land :)
16:10:27 <Sukhdev> amotoki: perfect - thanks - I will work with you to get this done
16:10:37 <viveknarasimhan> Sukhdev: This does not have impact on core ML2 plugin (or) its extensions, correct?
16:10:50 <Sukhdev> viveknarasimhan: correct
16:11:13 <Sukhdev> viveknarasimhan: just to make it a bit easier for the new guys who want to write ML2 drivers for Ironic
16:11:18 <viveknarasimhan> Sukhdev: Thanks for clarification.
16:11:50 <lazy_prince> Sukhdev: these are mech drivers we are talking about.. Right..?
16:11:58 <Sukhdev> lauramoore: so, based upon the clarification, you can push the updated version
16:12:01 <Sukhdev> lazy_prince: yes
16:12:04 <viveknarasimhan> lazy_prince: yes
16:12:27 <lauramoore> sukhdev: yes i will do that in the next couple of days
16:12:52 <Sukhdev> lazy_prince: we are doing all this work so that folks can write or modify their ML2 drivers to use ironic - hence, a small doc to make it easy for them will be good thing to have
16:13:08 <lazy_prince> +1
16:13:18 <viveknarasimhan> Sukhdev: We have a sequence diagram that we use for our ML2 mech drivers
16:13:37 <viveknarasimhan> Sukhdev: i can share that since that may be of some use for the small doc for ML2 driver writers
16:13:39 <Sukhdev> jroll, lazy_prince : if you can give a full review of the spec as well so that it is done in one go (hopefully) :-)
16:14:03 <Sukhdev> viveknarasimhan: fantastic - any help will be good
16:14:28 <jroll> Sukhdev: yep, will try
16:14:35 <Sukhdev> viveknarasimhan perhaps you, amotoki , and I can collaborate on that
16:14:45 <Sukhdev> cool - thanks
16:14:52 <viveknarasimhan> Sukhdev: Sure. Thanks!
16:14:56 <Sukhdev> So, looks like we are good with this spec -
16:15:10 <Sukhdev> anybody has any comments before we move to the next one?
16:15:39 <Sukhdev> The next spec - https://review.openstack.org/#/c/187829/
16:15:59 <Sukhdev> This has lots of +1's looks like this is ready to go as well -
16:16:17 <Sukhdev> jroll: you mentioned you are going to make a small edit on it
16:16:29 <Sukhdev> other than that - this is good to go as well
16:16:52 <Sukhdev> any question or comment on this one?
16:16:55 <lazy_prince> I think TheJulia had some concerns that jroll will address.. and then we will be good to merge it..
16:17:00 <jroll> Sukhdev: yeah, just need to add julia's comments
16:17:39 <Sukhdev> jroll: I had made a comment on it as well - but, seems like ironic tradition is a bit different - :-)
16:17:59 <Sukhdev> regardless it is a minor thing
16:18:29 <Sukhdev> I am hoping this week both of these specs will be approved barring any last minute issues
16:18:52 <Sukhdev> I think we can now proceed with the implementation part
16:19:14 <Sukhdev> #patches under review
16:19:20 <Sukhdev> Opps - sorry
16:19:30 <Sukhdev> #topic: patches under review
16:19:48 <jroll> are there patches? :)
16:19:49 <Sukhdev> I created a new category in Agenda so that in this section we can cover the reviews
16:20:37 <Sukhdev> jroll: not really - just to create a category - I just put few of the outstanding patches so that we have on "go to" place to see everything
16:20:50 <Sukhdev> s/on/one
16:20:50 <jroll> cool.
16:20:57 <lazy_prince> +1
16:20:58 <lauramoore> sukhdev: seems like a good idea
16:21:02 <lauramoore> +1
16:21:42 <Sukhdev> so currently - i copied three which are WIP sitting out there - feel free to look at them and get some idea or form openions
16:21:59 <Sukhdev> starting next week we will start to pay bit more attention to this area -
16:22:28 <Sukhdev> feel free to put a link here when you push something for review - this way we can all see it
16:23:06 <viveknarasimhan> Sukhdev: Sure, we will do that.
16:23:13 <Sukhdev> I did not plan on going through the patches listed - unless anybody has a question
16:23:41 <lauramoore> sure, thanks sukhdev, will upload once we have a patch for review
16:23:56 <Sukhdev> amotoki viveknarasimhan, we will create another category on the agenda - once we write something for the documentation
16:24:27 <Sukhdev> move right along -
16:24:47 <viveknarasimhan> Sukhdev: ok
16:24:52 <Sukhdev> #topic: Bare metal Physical connectivity scenarios
16:25:07 <Sukhdev> #link: https://docs.google.com/document/d/1a-DX4FQZoX1SdTOd9w_Ug6kCKdY1wfrDcR3SKVhWlcQ/view?usp=sharing
16:25:21 <Sukhdev> we covered this last week - and then ran out of time
16:25:42 <Sukhdev> wanted to make sure everybody had time to review these and we are good with this
16:25:43 <viveknarasimhan> Sukhdev: I opened it up for comments
16:25:56 <Sukhdev> viveknarasimhan: cool - I saw that - thanks
16:26:07 <viveknarasimhan> and provided public access to edit (for comments)
16:26:19 <Sukhdev> If any clarification is needed, feel free to post a comment on the document itsef
16:27:14 <Sukhdev> Any body wants to cover anything?
16:27:34 <Sukhdev> #topic: Open Discussion
16:27:52 <Sukhdev> Anybody wants to bring up anything - this is the time
16:27:58 * Sukhdev waiting
16:28:24 <viveknarasimhan> One query:
16:28:35 <Sukhdev> viveknarasimhan: sure
16:28:44 <viveknarasimhan> do we need to support port-additions and port-removals to an existing LAG ?
16:29:08 <Sukhdev> viveknarasimhan: not in the first cut
16:29:27 <lazy_prince> not in later cut too...
16:29:35 <Sukhdev> lazy_prince: :-)
16:29:41 <lazy_prince> as the port LAG will be provided by admins...
16:29:55 <Sukhdev> lazy_prince: makes sense
16:29:56 <lazy_prince> and users will not be able to manipulate it..
16:30:21 <viveknarasimhan> Sukhdev/lazy_prince: Ok . Thanks.
16:30:40 <lazy_prince> if at all admins change port info for ironic node, it will be used by ironic and neutron later on for next provisioning..
16:30:57 <Sukhdev> lazy_prince: so, the idea is if any such config change needs to take place, admis take down the server and rewire and re-deploy
16:31:05 <lazy_prince> thats my understanding unless i am not missing something..
16:31:36 <Sukhdev> lazy_prince: your understanding seems very reasonable
16:31:55 <lazy_prince> we should not make change to a maching when it is being used by tenants...
16:32:20 <Sukhdev> lazy_prince: +1
16:32:21 <viveknarasimhan> lazy_prince:  so we cycle it out.  get it out of tenant to cleaning network
16:32:34 <viveknarasimhan> lazy_prince: and then boot it again with a port-group in LAG
16:32:52 <viveknarasimhan> lazy_prince: onto the tenant network
16:32:54 <Sukhdev> viveknarasimhan: yup
16:33:02 <viveknarasimhan> Sukhdev: ok
16:33:06 <jroll> s/cleaning/provisioning/ since I'm pedantic :)
16:33:08 <lazy_prince> viveknarasimhan: yes.. but this will be a new instannce
16:33:15 <lazy_prince> not a old one..
16:33:35 <amotoki> "nova rebuild" or something on ironic instance can use a new config, but it is just one option.
16:33:47 <viveknarasimhan> lazy_prince: agreed. it is a brand new nova boot with a ironic port passing in a port-group to neutron right :)
16:34:43 <Sukhdev> viveknarasimhan: yes, that will be correct
16:34:48 <amotoki> I generally agree using a new config for a new instance.
16:34:54 <lazy_prince> not sure of nova rebuild will need to check on that..
16:35:20 <lauramoore> the state machine impact section in the spec has some details about changes to port_group memberhip
16:35:28 <viveknarasimhan> There is no hot-plugin / plugout facility of nova extended for Ironic ?
16:35:28 <amotoki> yeah,.. I just wrote a possible future option, but I am not sure we can.
16:36:13 <lazy_prince> may be jroll can answer that
16:36:37 <jroll> answer what?
16:36:56 <lazy_prince> hot-plugin ^^
16:37:22 <jroll> I have no idea what that would mean
16:37:29 <viveknarasimhan> nova allows ability to add / remove NICs belonging to networks while an instance is running
16:37:33 <jroll> like, so we can re-wire nodes with an instance?
16:37:37 <jroll> ok
16:37:39 <viveknarasimhan> this is provided by some nova drivers
16:37:43 <viveknarasimhan> not all of them
16:37:43 <jroll> how do you propose software adds a physical NIC?
16:38:07 <viveknarasimhan> wiring can be in place already
16:38:20 <viveknarasimhan> the decision to include them in the instance can be dynamic
16:38:31 <jroll> mmmm
16:38:35 <jroll> let's say "later"
16:38:46 <jroll> this kind of goes with selecting nic <-> network mapping, right?
16:38:46 <Sukhdev> jroll: +1
16:39:14 <viveknarasimhan> yes,
16:39:14 <Sukhdev> jroll: not really -
16:39:21 <viveknarasimhan> partly
16:39:24 <lazy_prince> well.. NIC may not be added, but networks can be added for sure... but as jroll said, "later"
16:39:44 <jroll> personally, I'd prefer to add networks via LAG+VLAN, just add a VLAN
16:39:51 <viveknarasimhan> this concept can be used to add ports to a LAG
16:39:55 <jroll> but yeah let's punt on this
16:39:56 <lazy_prince> jroll: +1
16:39:58 <viveknarasimhan> not only new networks. that is why i mentioned that
16:39:58 <jroll> um
16:40:06 <jroll> viveknarasimhan: for a LAG nova sees that as a single NIC
16:40:37 <viveknarasimhan> jroll: Oh...   But Ironic would see them as two unique ports right?
16:41:10 <jroll> viveknarasimhan: correct
16:41:14 <viveknarasimhan> jroll: to put out, Ironic would have two ports both enrolled into the same port-group.   nova will boot with one neutron port passing in Ironic port-group
16:41:23 <jroll> correct.
16:41:27 <viveknarasimhan> jroll: OK, got it!
16:41:41 <Sukhdev> viveknarasimhan: correct
16:42:09 <Sukhdev> viveknarasimhan: hey - you wrote it in the scenario document :-)
16:42:39 <viveknarasimhan> Sukhdev: I just revisited here
16:42:57 <Sukhdev> viveknarasimhan: sure - no worries
16:43:06 <Sukhdev> anything else?
16:43:24 <Sukhdev> shall we call it a day?
16:43:47 * Sukhdev going once going twice :-)
16:44:01 <Sukhdev> Looks like we are done -
16:44:19 <Sukhdev> Thanks for attending the meeting - it was a productive discussion...
16:44:25 <Sukhdev> #endmeeting