20:00:12 <johnsom> #startmeeting Octavia
20:00:13 <openstack> Meeting started Wed Jun 21 20:00:12 2017 UTC and is due to finish in 60 minutes.  The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:15 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:17 <openstack> The meeting name has been set to 'octavia'
20:00:23 <johnsom> Hi folks!
20:00:42 <JudeC> o/
20:01:20 <jniesz> hello
20:01:22 <xgerman_> o/
20:01:49 <johnsom> #topic Announcements
20:02:13 <johnsom> Just my regular reminder, feature freeze is coming:
20:02:14 <johnsom> We are heading towards feature freeze Pike-3 July 24th
20:02:40 <xgerman_> we should probably start an etherpad for the PTG as well to gauge interest
20:02:43 <johnsom> So please keep that in mind for things you want to land in Pike
20:03:02 <johnsom> We can, that is September, so it might be a bit early....
20:03:13 <johnsom> We will have a room dedicated for three days
20:03:25 <johnsom> I have reserved that already
20:03:38 <xgerman_> NICE
20:03:54 <johnsom> Next PTG is September 11-15 in Denver
20:04:00 <johnsom> #link https://www.openstack.org/ptg#tab_schedule
20:05:05 <johnsom> #link https://etherpad.openstack.org/p/PTG-Queens-Octavia
20:05:29 <johnsom> Ok, there we go
20:06:06 <xgerman_> +!
20:06:21 <johnsom> Ok, other announcements?
20:06:48 <johnsom> There is talk on the mailing list about the "big tent" and what we call "official" projects
20:07:15 <johnsom> Also there is an interesting thread on Trove which is proposing a complete re architecture (maybe even new name)
20:07:44 <johnsom> #topic Brief progress reports / bugs needing review
20:07:58 <johnsom> We still have two specs up for review:
20:08:05 <johnsom> #link https://review.openstack.org/453005
20:08:13 <johnsom> l3-active-active spec
20:08:14 <johnsom> and
20:08:19 <johnsom> #link https://review.openstack.org/392485
20:08:26 <johnsom> flavors spec
20:08:32 <johnsom> Please review and comment on those
20:08:52 <jniesz> for the l3 active/active I can update to use amphora lifecycle for distributor functions
20:08:59 <jniesz> if that is the direction we want to go
20:09:42 <johnsom> As for my progress report, I wrote up RBAC policy enforcement for the Octavia v2 API.  It has started to merge, but there are still parts up for review.  I plan to also update the open API patches from JudeC.
20:09:45 <xgerman_> I think we would like to do that. I also ended up promising to refactor the servcie vm part of the IBM patch
20:10:19 <xgerman_> I also fixed gate issues in ioenstack-ansible - now need to get them to merge it
20:10:31 <johnsom> Yeah, I think that makes sense for the VM based distributor, but I also think we need to make sure what is there doesn't *require* the VM if it's not necessary for the driver
20:11:00 <jniesz> yes that makes sense as an implementation might need 0 VMs
20:11:17 <rm_work> +1
20:11:18 <johnsom> Right
20:11:35 <johnsom> So, please, if you have time, dig into the Act/Act patches
20:12:25 <johnsom> The distributor VMs there are basically amphora, so no point in not leveraging the lifecycle stuff we have in place for them and just extending the amp-agent api to support a new "type" of amphora
20:13:38 <nmagnezi> o/
20:13:42 <nmagnezi> (sorry to be late)
20:14:10 <xgerman_> +1 I throw some patch over the wall: https://review.openstack.org/#/c/313006/
20:14:33 <xgerman_> in particular https://review.openstack.org/#/c/313006/81/doc/source/api/haproxy-amphora-api.rst
20:14:39 <johnsom> reedip and others, interested to hear the state of the qos patch.  I think it needs some clean up but is close to being ready for review?
20:15:21 <xgerman_> redip is in India so likely to late for him
20:15:21 <jniesz> +1 xgerman_
20:15:46 <johnsom> It would be nice to get that one in for Pike.  Yeah, but sometimes he is around, so thought I would ping
20:16:03 <johnsom> Ok, any other progress reports from folks?
20:16:11 <johnsom> Some of us have been working on gate issues too.
20:16:12 <xgerman_> He is usually actibve after 10 pm my time
20:17:09 <johnsom> Ubuntu had bad cloud images for a bit, diskimage-builder had a bug with partitioning (fixed in master, not yet released so we may still see it in lbaas gates), and we have been working on the 404 issue seen in the gates.
20:17:40 <johnsom> We have seen one case where the VIP interface never shows up under Linux (or at least after minutes of waiting.)
20:17:49 <johnsom> So, smashing some bugs.
20:18:17 <johnsom> Ok, moving on
20:18:37 <johnsom> #topic Discuss moving the meeting time
20:18:44 <johnsom> #link http://lists.openstack.org/pipermail/openstack-dev/2017-June/118363.html
20:19:00 <johnsom> This got lost in my inbox (gmail and outlook are fighting)
20:19:17 <rm_work> huh, yeah same
20:19:18 <xgerman_> #vote…
20:19:29 <xgerman_> didn’t have that for a while ;-)
20:19:30 <rm_work> that's probably fine by me, actually might be easier for me to attend
20:19:37 <rm_work> and gets me out of my standup once a week :P
20:19:57 <rm_work> i guess we should officially vote on the ML, but... I'm +1
20:20:11 <rm_work> one less thing to split my afternoon into bits
20:20:13 <johnsom> I was going to do one of those online time vote things, but a vote here should work ok.  I'm pretty sure it's a better time for the folks that can't regularly make this meeting
20:20:18 <nmagnezi> sec.. I'll need to convert this to my local timezone
20:20:31 <xgerman_> k
20:20:36 <johnsom> Ok, let me send out an e-mail with the online vote thing.
20:20:37 <rm_work> 3 hours earlier than now
20:20:40 <rm_work> nmagnezi:
20:20:43 <johnsom> Just to give time to look at calendars, etc.
20:20:47 <rm_work> well, 3 hours 20 min
20:21:54 <nmagnezi> for me it's actually much worse than our current time :<
20:22:29 <xgerman_> nmagnezi reply to the e-mail with a new proposal and we can hold the vote next week
20:22:42 <nmagnezi> xgerman_, aye
20:23:17 <johnsom> #link https://doodle.com/poll/kxvii2tn9rydp6ed
20:23:18 <rm_work> nmagnezi: ah when is that for you
20:23:20 <johnsom> Ok, there we go
20:23:35 <rm_work> i would think earlier is better as it's now your night? or does it put the meeting during your dinner time? >_>
20:23:38 <nmagnezi> rm_work, 20:00
20:24:35 <rm_work> johnsom: can we make that poll possible to add more times? :/
20:24:43 <rm_work> so people could propose times
20:24:46 <rm_work> i dunno if that's a thing
20:24:51 <johnsom> Yeah, If you don't see those options, send them to me and I will add
20:24:57 <xgerman_> let them propose on the ML
20:25:17 <rm_work> k
20:25:20 <rm_work> i mean *I* don't have suggestions, but others might
20:25:34 <johnsom> Yeah, please make sure that you find an open IRC meeting room for the timeslots proposed
20:26:08 <johnsom> Ok, so more discussion on the mailing list.
20:26:23 <johnsom> #topic Open Discussion
20:26:28 <johnsom> Other topics for today?
20:27:05 <xgerman_> should we start some priority list of patches again?
20:27:36 <xgerman_> after all P-3 is near and we should make it easy to focus
20:27:39 <johnsom> Yeah, probably.  Maybe for next week.
20:28:12 <cpuga> o/
20:28:32 <cpuga> is it a good time to talk about flavor or do we want to have that conversation on gerrit
20:28:33 <rm_work> yep
20:28:43 <rm_work> we should definitely have a priority list
20:29:02 <xgerman_> cpuga it’s a good time
20:29:17 <rm_work> i find myself mentioning to multiple people which patches need to be looked at, which usually means a central place for this would be good
20:29:19 <rm_work> can we put a priority list etherpad in our channel topic?
20:29:41 <johnsom> Yeah, I can do that
20:29:52 <xgerman_> +10
20:30:57 <cpuga> sorry got distracted, umm was there anything in particular that we don't feel comfortable with the design
20:31:17 <cpuga> it was metioned that it perhaps could be simplified
20:31:38 <xgerman_> yes,  I didn’t like all the key-value tables
20:32:03 <johnsom> Hmm, well, I will have to work on that.  I should be anointed, but it's giving me issues.
20:32:14 <cpuga> which part in particular the capability list?
20:32:30 <xgerman_> yep, and the values we add to it
20:32:53 <xgerman_> but I am no authority I just don’t like to do 100s of curl posts
20:33:59 <cpuga> the flavor profile can be created with one post
20:34:30 <xgerman_> but then I need to set up those key-values for the “flags”
20:35:20 <xgerman_> rm_work sinc eyou are actually have “real” users would that ne helpful for them to query the capabilities?
20:35:47 <cpuga> so the idea currently is that upon the installation of the provider driver the operator would add the providers supported key/values pairs
20:36:26 <jniesz> a provider might not express capabilities with key/values.  They might express with key and json data
20:36:33 <jniesz> or just json data
20:36:45 <cpuga> xgerman_ would you be okay if I would drop those two table and just keep the key/values pair on the flavor_metadata table?
20:37:50 <xgerman_> sure if I can set them all with one json post ;-)
20:39:07 <johnsom> Yeah, I like the json blob idea better myself.  I don't think we want operators in the database.
20:39:21 <johnsom> It's likely to lead to type issues, etc.
20:39:47 <cpuga> the operator wouldn't directly be in the database
20:39:55 <cpuga> it would be via rest calls
20:43:18 <cpuga> I'm flexible with the design, I'd really would just like to get a feel for the teams preference so that it can move forward.
20:44:43 <johnsom> I need to take another pass over it and comment on the patch.
20:45:12 <cpuga> johnsom: would it be fine to request a webex, after you guys take a look at the spec?
20:45:38 <johnsom> Yeah, as long as it is open to the whole community
20:45:47 <xgerman_> +!
20:45:49 <johnsom> I.e. the octavia team
20:45:57 <cpuga> yes definitely
20:46:31 <jniesz> I think it would make sense to add version field to provider and flavor profile
20:46:40 <johnsom> Yeah, I recommend sending something out to the dev mailing list with [octavia] tag in the subject
20:47:30 <jniesz> +1 on that and maybe we can include active/active L3 as well
20:48:14 <jniesz> or I can send separate for that
20:48:32 <johnsom> Oye, we might want two...  Either one of those could be big topics
20:48:42 <xgerman_> +1
20:49:41 <johnsom> Ok, sounds like a plan
20:50:00 <cpuga> k, thx
20:50:29 <johnsom> Other topics for today?
20:51:07 <nmagnezi> yes
20:51:16 <nmagnezi> please have a look at https://bugs.launchpad.net/octavia/+bug/1698654
20:51:17 <openstack> Launchpad bug 1698654 in octavia "API2.0: Loadbalancer create won't accept provider" [Undecided,New]
20:51:25 <nmagnezi> saw this while I tested a patch by rm_work
20:51:36 <nmagnezi> that's all :)
20:51:47 <rm_work> yeah
20:51:52 <rm_work> there's ... not really a provider
20:51:53 <johnsom> Ah, yeah, we haven't implemented that yet....
20:52:04 <JudeC> The provider stuff is just stubbed out afaik.
20:52:09 <johnsom> Correct
20:52:14 <nmagnezi> so.. maybe we should upfate the api-ref for now
20:52:26 <nmagnezi> update*
20:52:44 <johnsom> When I wrote that I expected it to be there, but we lost a bunch of people, so things have changed
20:52:58 <nmagnezi> understandable
20:53:04 <johnsom> flavor is on that list too, which is what we were just talking about.
20:53:10 <nmagnezi> i saw it in the loadbalancer create example
20:53:31 <nmagnezi> dunno if it shows in additional examples.. I imagine it is
20:53:51 <johnsom> Technically it is part of the API, it is just a bug that it is not implemented.
20:54:11 <nmagnezi> yeah but I think flavors are not causing any errors for now so we're good on that
20:54:15 <xgerman_> yeah, we probably should get to it when we actually have providers IHMO
20:54:17 <johnsom> #link https://bugs.launchpad.net/octavia/+bug/1655768
20:54:18 <openstack> Launchpad bug 1655768 in octavia "Need to enable "provider" support to the octavia v2 api" [High,New]
20:54:29 <nmagnezi> so maybe a temporary comment in the api-ref
20:55:21 <johnsom> We can probably make it respond better via the API
20:55:33 <johnsom> I will look at that
20:56:53 <johnsom> nmagnezi I'm going to mark your bug a duplicate of that one we already had open for it.
20:57:00 <nmagnezi> np
20:57:23 <johnsom> Ugh, note to self, I need to go clean out bugs too
20:57:41 <johnsom> We have closed some of these but didn't tag them
20:58:02 <johnsom> Ok, other topics?
20:58:08 <nmagnezi> nothing on my end
20:59:03 <johnsom> Ok, with two minutes left, thanks folks!
20:59:16 <johnsom> #endmeeting