14:00:22 <mlavalle> #startmeeting neutron_drivers
14:00:23 <openstack> Meeting started Fri Apr 27 14:00:22 2018 UTC and is due to finish in 60 minutes.  The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:27 <openstack> The meeting name has been set to 'neutron_drivers'
14:01:09 <mlavalle> Hi there!
14:01:18 <haleyb> hi
14:01:23 <amotoki> hi
14:02:31 <mlavalle> amotoki: if you have a chance, please look at https://review.openstack.org/#/c/564748/ from haleyb
14:02:31 <patchbot> patch 564748 - releases - Neutron Queens 12.0.2
14:03:20 <amotoki> mlavalle: looking
14:03:39 <mlavalle> I think we have quorum. yamamoto might join us later
14:03:47 <mlavalle> #topic RFEs
14:04:59 <mlavalle> First RFE for today is https://bugs.launchpad.net/neutron/+bug/1748132
14:05:00 <openstack> Launchpad bug 1748132 in neutron "[RFE] Can not create router gateway without external_fixed_ips" [Wishlist,In progress] - Assigned to Guoshuai Li (liguoshuai1990)
14:05:25 <mlavalle> Looking at the notes, I think ubmitter is confused about latest suggestion
14:06:59 <amotoki> yeah, I have a similar feeling.
14:07:12 <amotoki> on the other hand, the submitter commented " I just hope that users can economize it."
14:08:14 <mlavalle> maybe add some furthere clarification to the RFE
14:08:20 <amotoki> if a provider charges users based on the number of consumed public IPs, this would work.
14:08:46 <mlavalle> yeah, I think so
14:09:01 <amotoki> is seems better to clarify what the submitter would like to have more.
14:09:07 <amotoki> it*
14:09:29 <mlavalle> yes
14:10:20 <mlavalle> maybe we don't need to concern ourselves so much what this specific submitter wants / needs
14:10:52 <mlavalle> maybe we just want to open the possibility to users / deployers to save the IP if that is what they want
14:10:59 <amotoki> mlavalle: will you add some comment? I can comment more but it might be better to use another phrase by other person.
14:11:37 <mlavalle> I can do that. But before doing that, waht do you think of my previous comment
14:12:34 <amotoki> I totally agree with your two mentions above
14:12:54 <mlavalle> in that case, why not just go ahead an approve the RFE?
14:13:27 <mlavalle> and let users / deployers decide for themeselves what they want more
14:14:08 <amotoki> it opens a door to save IPs based on users' choice.
14:14:19 <mlavalle> yeah
14:14:35 <mlavalle> if that is what they want
14:14:55 <mlavalle> I don't think we would be breaking anything if we allow it
14:15:44 <amotoki> yes, it is backward-compatible.
14:16:40 <mlavalle> so approve it? I am for it
14:17:29 <amotoki> ah, I got it. the rfe itself requests to support empty fixed IPs for gateway ports.
14:17:48 <amotoki> it help save IP addresses by users.
14:18:06 <amotoki> so what is requested here itself makes sense.
14:18:14 <mlavalle> I think so
14:18:38 <amotoki> It seems I try to clarify the intention too mcuh.
14:18:59 <amotoki> I am okay to approve this
14:19:15 <mlavalle> that's ok, that's what we are here for ;-) I appreciate your questioning
14:19:24 <mlavalle> any concerns haleyb ?
14:19:45 <haleyb> no concerns, you seem to have it under control
14:20:05 <mlavalle> LOL, I hope it doesn't come back to haunt me ;-)
14:20:14 <amotoki> it looks like a good starter RFE :)
14:20:21 <mlavalle> yeap
14:20:30 <mlavalle> I will advertise it in the next meeting
14:20:43 <mlavalle> hang on, let me approve
14:23:03 <mlavalle> done
14:23:29 <mlavalle> ok, next one for today
14:23:35 <mlavalle> https://bugs.launchpad.net/neutron/+bug/1756705
14:23:37 <openstack> Launchpad bug 1756705 in neutron "RFE: New Neutron extension for binding levels" [Wishlist,In progress] - Assigned to Evgeny Fedoruk (evgenyf)
14:25:33 <mlavalle> In the last meeting, yamamoto raised the question as to why the HPB details were not returned in the port response to the API, when HPB was originally implemented
14:25:52 <mlavalle> he raised the possibility that the team had a good reason not to do it
14:26:49 <mlavalle> so I pinged rkukura and had an IRC chat with him. The answer is really simple: they planned to do it but never gout around to actually do it :-)
14:28:25 <mlavalle> The original reasons to return that information in the response had more to do with debugging, rather than appliances set up like in the case of this RFE
14:28:52 <mlavalle> it seems to make sense
14:29:01 <mlavalle> so I am for approving the RFE
14:29:38 <amotoki> it makes sense to me too, though I honestly do not understand how it can be used in loadbalancer impl..
14:30:08 <mlavalle> yeah, Bob was a little mistyfied as well
14:30:47 <amotoki> anyway, some custom controller can do something based on a driver bound to a port.
14:31:09 <mlavalle> maybe they plan to have scripts to scrap that information from the API response
14:31:30 <amotoki> I think so.
14:31:55 <mlavalle> so, approve?
14:32:11 <amotoki> I am okay to approve this. one question is which is a better format for binding:levels  arrays or dict.
14:32:43 <mlavalle> I think it is an array of dicts
14:32:55 <mlavalle> with each dict representing a level
14:33:15 <amotoki> according the description of the RFE, it is a dict and each entry has a level.
14:33:44 <mlavalle> I already discuseed that with him in a docs patch that he submitted in neuton-lib
14:34:01 <amotoki> ah... I wasn't aware of it.
14:34:03 <mlavalle> and he agreed that it had to be array of dicts
14:34:11 <mlavalle> like port's fixed_ips
14:34:13 <amotoki> sounds good to discuss in the review.
14:34:29 <amotoki> +1 for approving it.
14:34:42 <mlavalle> any concerns haleyb?
14:35:13 <haleyb> no, nothing from me on that one
14:35:23 <mlavalle> cool
14:35:45 <mlavalle> hang on. I'll approve it and even include the array of dicts suggestion
14:36:02 <amotoki> we've approved two long standing RFEs :)
14:37:57 <mlavalle> \o/
14:38:29 <mlavalle> Next one is https://bugs.launchpad.net/neutron/+bug/1766380
14:38:30 <openstack> Launchpad bug 1766380 in neutron "[RFE] Create host-routes for routed networks (segments)" [Undecided,Confirmed]
14:40:03 * mlavalle will give the team some time to go through the long description
14:41:32 <mlavalle> good detailed description, btw. no complaints
14:43:20 <amotoki> generally speaking, what the RFE says sounds reasonable (though I haven't looked thru the detail of the model proposed)
14:43:57 <haleyb> so a change to the dhcp agent to detect these multiple segments and install host/network routes automatically
14:44:21 <haleyb> there's a lot to read there
14:44:28 <mlavalle> it makes sense as to the gap he has identified
14:45:32 <mlavalle> and yes it is to install host routed in each subnet in a routed network, so the traffic statys within the network
14:46:10 <mlavalle> would it make sense for you guys if you take some time reading it carefully and retaking it during the next meeting?
14:46:27 <amotoki> that's my understanding so far, but I am refreshing my memory on routed networks and it needs some more time.
14:46:35 <amotoki> mlavalle: +1
14:47:04 <haleyb> mlavalle: so on-link routes where the VM can just send directly?  i think i understand
14:47:05 <mlavalle> of let's come back to it next meeting
14:47:39 <haleyb> great, homework :)
14:47:45 <mlavalle> LOL
14:48:03 <amotoki> haleyb: what does 'on-link routes' mean?
14:49:30 <amotoki> ah, it seems the scope of routes.
14:49:37 <mlavalle> yes
14:50:00 <haleyb> amotoki: where you can just send packets for that route directly on an interface, even if there is no gateway
14:50:07 <haleyb> yes, the scope
14:50:31 <amotoki> thanks. I am sometimes confused with this term :(
14:50:43 <haleyb> in this case all the subnets are on the network
14:51:15 <amotoki> so the scope must be local.
14:51:17 <haleyb> i used to use this trick to skip a router (assuming that's what this RFE is asking for)
14:52:38 <mlavalle> I think it is still a good idea to take some time to digest it
14:52:47 <mlavalle> we can retake it, first thing next meeting
14:53:35 <mlavalle> #topic Open Agenda
14:54:06 <mlavalle> amotoki: I am told there will be a week long holiday next week in Japan. Is that correct?
14:54:21 <amotoki> mlavalle: yes.
14:54:37 <amotoki> I might be travelling somewhere in Japan though I haven't decided yet where I go.
14:54:44 <mlavalle> so yamamoto and you most likely won't make it the meeting, I assume
14:55:17 <amotoki> at now it is 50:50
14:55:44 <amotoki> it is late in Japan, so perhaps I can join.
14:56:00 <mlavalle> so let's cancel next week's meeting and give you and yamamoto the freedom to thoroughly enjoy your holiday
14:56:26 <amotoki> it would be great. thanks
14:56:34 <mlavalle> just out of curiosity and deep jelousy, what is this holiday about?
14:57:15 <amotoki> several holidays are defined...
14:57:47 <amotoki> i found wikipedia page https://en.wikipedia.org/wiki/Golden_Week_(Japan)
14:58:01 <mlavalle> Thanks! Enjoy
14:58:19 * mlavalle GREEN with jelousy
14:58:26 <amotoki> :)
14:58:32 <mlavalle> #endmeeting