20:00:14 <johnsom> #startmeeting Octavia
20:00:15 <openstack> Meeting started Wed Dec 14 20:00:14 2016 UTC and is due to finish in 60 minutes.  The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:19 <openstack> The meeting name has been set to 'octavia'
20:00:22 <xgerman> o/
20:00:38 <johnsom> Hello everyone.
20:01:03 <johnsom> I know some folks can't make it today, so it might be a little light on attendance
20:01:18 <johnsom> #topic Announcements
20:01:26 <johnsom> Ocata-2 milestone this week
20:01:27 <xgerman> hope it’s not just us two ;-)
20:01:34 <johnsom> #link https://releases.openstack.org/ocata/schedule.html
20:01:37 <perelman> Hi
20:01:43 <johnsom> Hi perelman
20:01:55 <perelman> It's just three of us :)
20:02:04 <perelman> Hi all
20:02:09 <johnsom> Just a reminder, Ocata is now in the last stretch towards Ocata-3 milestone.
20:02:24 <perelman> I have a question - whe
20:02:37 <johnsom> We still have our merge priority review list:
20:02:40 <perelman> Are there dates for the next mid cycly
20:02:44 <johnsom> #link https://etherpad.openstack.org/p/octavia-ocata-merge-review-priority
20:02:45 <perelman> Ecycle
20:03:02 <blogan> hello
20:03:10 <johnsom> perelman  Currently the next meet up is going to be the PTG
20:03:21 <johnsom> #link http://www.openstack.org/ptg
20:03:25 <diltram> o/
20:03:26 * xgerman hoping he figures out how to attend
20:03:26 <johnsom> Feb 20-24
20:03:31 <kobis> Hi
20:03:32 <perelman> Sorry for mistakes - a bit strange keyboard
20:04:03 <johnsom> NP
20:04:05 <perelman> So we won't have a separate mid cycle.
20:04:06 <nmagnezi> hello everyone
20:04:08 <nmagnezi> o/
20:04:16 <johnsom> Hi there.
20:04:21 <diltram> hey people
20:04:41 <johnsom> Any other announcements before we head into progress reports/ patches needing review?
20:05:25 <johnsom> Folks might also be interested in the "community goals" for Pike.  See the mailing list for the discussion.
20:05:29 <johnsom> #topic Brief progress reports / bugs needing review
20:06:09 <johnsom> Ok, Nir had two on the agenda that have been hanging around a while.  Both related to the legacy namespace driver
20:06:16 <johnsom> #link https://review.openstack.org/#/c/344658/
20:06:24 <johnsom> #link https://review.openstack.org/#/c/299998/
20:06:35 <johnsom> I think some folks have started reviews/ testing on those
20:06:57 <rm_work> o/
20:07:49 <xgerman> hi rm_work
20:07:53 <johnsom> A number of us have been working through the diskimage-builder issues.  Ugh, what a bunch of issues opened by that change
20:08:32 <johnsom> We are down to some issues with the amphora-agent under the new python3 environment
20:08:39 <johnsom> At least as far as I can see
20:08:53 <nmagnezi> thanks a lot for your cycles for ^^ , please ping me for any question you might have
20:09:03 <nmagnezi> (the patches johnsom linked)
20:09:05 <johnsom> Any other updates?
20:09:52 <johnsom> #topic Meeting schedule for the next few weeks
20:10:13 <johnsom> So number of folks have holidays over the next few weeks.
20:10:41 <johnsom> I would like to cancel the 28th as most folks in the US will be taking time off (myself included).
20:10:52 <rm_work> +1
20:10:56 <diltram> +1
20:10:57 <johnsom> Do you want to meet next week or should we follow other teams and cancel it as well?
20:10:59 <nmagnezi> happy holidays guys :)
20:11:32 <johnsom> I guess asking another way, are you all going to be around next Wednesday?
20:11:35 <rm_work> I could probably make it, but not sure how much we'll have
20:11:46 <xgerman> +1
20:11:48 <nmagnezi> johnsom, i think that if anything urgent comes up we can just communicate via mail this time.
20:11:56 * johnsom Hopes a celebration that the DIB issue is resolved
20:12:09 <nmagnezi> +1000
20:13:19 <johnsom> Ok, so no strong signal either way, so I think we should cancel the next two week's meetings and pick up again on the  4th of January.
20:13:39 <johnsom> I will send a note out to the mailing list and update the agenda pages
20:14:17 <johnsom> #topic Meeting time discussion
20:14:24 <johnsom> #link https://etherpad.openstack.org/p/octavia-weekly-meeting-time
20:15:01 <johnsom> Please remember to put your timezone information on the etherpad.
20:15:15 <johnsom> It still looks a little light for the folks that participate in the meeting.
20:16:03 <kobis> Yet crowded with people who usually don't - as this hour is tough on us :)
20:16:07 <johnsom> kobis Maybe have some options/recommendations for times when there is a channel available for the 4th meeting?
20:16:56 <kobis> I guess that we should wait till we have some more feedback before we suggest anything?
20:17:11 <diltram> +1 in my opinion
20:17:34 <johnsom> Yeah, that was why I was thinking the 4th meeting, but I'm happy to delay.
20:17:42 <nmagnezi> kobis, are there any specific votes we are waiting to see?
20:18:18 <kobis> Nms
20:18:20 <kobis> A
20:19:08 <johnsom> I would have expected to see some of the asia-pacific area folks to jump in as well.
20:19:09 <kobis> I guess that most of the poeple who are around the US haven't voted/noted
20:19:32 <kobis> And none from India and Eastern Asia
20:19:49 <kobis> At least at the last time which i checked
20:20:06 <johnsom> Yeah, so maybe a bit more time and a reminder in my e-mail about the meeting cancellation.
20:21:02 <johnsom> #topic legacy namespace driver upgrade path after the merge (nmagnezi)
20:21:15 <nmagnezi> hi everyone
20:21:35 <nmagnezi> so, i added this to the agenda: "[nmagnezi]: Due to neutron-lbaas and Octavia merge, we should discuss the Newton -> Ocata upgrade path for haproxy in namespace users (Howto)."
20:21:51 <nmagnezi> So i think the topic name actually says it all :) , that is actually a question / a start of discussion about the situation of upgrades for current user of the old driver.
20:21:52 <johnsom> To start this topic I should probably mention, given where we are at with the merge work and Ocata-2, I doubt we will be moving drivers in Ocata.
20:22:50 <nmagnezi> johnsom, if we won't make it, will ppl you use the old driver need to activate some sort of proxy to keep it running?
20:23:18 <nmagnezi> i mean a pass through proxy for the API
20:23:28 <johnsom> No, for the Ocata cycle we will probably end up staying the way we are.
20:23:54 <johnsom> Just having the API running in Ocata, but not yet used/advertised as an endpoint.
20:24:37 <johnsom> My hope for the legacy driver is to just need a database migration, but given how intertwined it is with neutron, it may require more.
20:25:06 <nmagnezi> aye, but since we don't know for 100% what will be the situation at the release of Ocata (maybe we will make it..), are we sure that will be the case for the old driver users?
20:25:39 <nmagnezi> if not, i'm willing to help with this and maybe write some spec / doc  , but i will need guidance :)
20:25:45 * johnsom thinks "nothing is ever sure in OpenStack"
20:25:57 <nmagnezi> +1
20:25:58 <nmagnezi> :D
20:26:14 <johnsom> Since work to move the legacy namespace driver has not started, I can't give you more of an answer.
20:26:16 <xgerman> not sure why we need the namespace driver - is there a requirement octavia doesn’t meet
20:26:25 <xgerman> ?
20:26:26 <johnsom> If you would like to start that work, please do
20:26:36 <johnsom> xgerman +1
20:26:50 <nmagnezi> xgerman, operators take their time to move forward
20:26:52 <kobis> Unless there's someone who actually use it... and would like to upgrade
20:27:08 <johnsom> It's a good question.  My feeling is a number of folks are actively using it.  It would at least need to go through deprecation cycle
20:27:41 <johnsom> I know heat and magnum are using it for testing
20:28:14 <nmagnezi> johnsom, xgerman, i know it is being used, the future is Octavia of course, but some folks take time to adopt new things.
20:28:22 <johnsom> That said, I think they could change
20:28:43 <diltram> this is why we have deprecation time
20:28:43 <xgerman> well, octavia has been around for years now ;-)
20:28:51 <diltram> and we gonna work on migrations tools
20:29:07 <xgerman> diltram don’t make promises like that
20:29:20 <diltram> ?
20:29:32 <nmagnezi> xgerman, for the debian tribe ;) , when I finish with https://review.openstack.org/#/c/331841/ our customers can also adapt Octavia
20:29:42 <nmagnezi> adopt*
20:29:48 <kobis> xgerman: if anyone has a customer who uses it outside of openstack testing projects then they'll have to migrate
20:30:11 <xgerman> diltram migration is hard and I am not sure if we should suck it out of DB and recreate in octavia
20:30:18 <xgerman> nmagnezi awesome
20:30:19 <kobis> And as we never officially announced that namespace driver is deprecated... why should they make an effort
20:30:41 <xgerman> kobis don’t get me started why that was never announced
20:30:46 * xgerman looks at blogan
20:30:53 <kobis> :D
20:31:18 <johnsom> Ha
20:31:34 <nmagnezi> kobis, as long is it was not deprecated officially, you cannot force ppl not to use it :)
20:31:49 <kobis> My point exactly
20:31:49 <johnsom> Well, there are two steps here.  One, migration to legacy namespace driver running under Octavia API.
20:32:10 * blogan xgerman looks back confused
20:32:15 <johnsom> The other would be migration to using octavia driver instead of legacy namespace.  Two different problems
20:32:18 * blogan cant type
20:32:27 <kobis> Maybe we should do that asap and avoid the challenge of having namespace under octavia
20:33:05 <johnsom> We can try, but I expect folks to come out of the woodwork (so to say) saying they need it.
20:33:19 <xgerman> johnsom, well then they can support it
20:33:20 <kobis> I wasn't serious about this
20:33:22 <johnsom> We could start with a mailing list post asking about it's use
20:33:31 <xgerman> +1
20:33:40 <nmagnezi> kobis, i would vote against it for now, as long we cannot use octavia without neutron-lbaas
20:33:40 <blogan> well it'll cause those people to need more capacity in their nova deploys, so i assume it'll cause some people issues
20:34:03 <johnsom> True
20:34:16 <johnsom> But free up like 100 network nodes, so....   GRIN
20:34:26 <blogan> lol
20:34:30 * xgerman thinks we should have build that wall between neutron and nova
20:34:32 <kobis> :)
20:35:15 <johnsom> Ok, no harm in sending out a survey e-mail.  "Raise your hand if you are using the legacy namespace driver"
20:35:36 <johnsom> I just expect to see a lot of hands...
20:35:56 <blogan> i expect to see very little because the people who are using it won't be paying attention to the ML
20:36:02 <kobis> When do we expect nlbaas to be out of business?
20:36:05 <johnsom> nmagnezi Did you get enough info out of that conversation?
20:36:23 <nmagnezi> johnsom, for now yes, thank you :)
20:36:47 <johnsom> kobis Realistically at our current rate, it would be Pike when nlbaas is just a pass through proxy.
20:37:14 <johnsom> We just don't have enough in place at the end of Ocata-2 to start moving drivers
20:38:03 <blogan> kobis: are you going to be able to do that conduit between the neutron plugin and an octavia driver?
20:38:45 <kobis> I do - idk if ill have that ready this week...
20:38:53 <rm_work> did devstack change something with the version of qemu-kvm that gets installed??? like, NOW?
20:38:55 <rm_work> bleh
20:39:03 <rm_work> all the sudden lots of dep issues on centos7
20:39:06 <blogan> kobis: oh yeah that'd be surprising if you did
20:39:47 <johnsom> rm_work Don't know
20:39:53 <johnsom> Ok, let's move on.
20:40:02 <johnsom> #topic Octavia event notifier blueprint
20:40:13 <johnsom> There is a proposed blueprint and patch up
20:40:25 <johnsom> #link https://blueprints.launchpad.net/octavia/+spec/octavia-event-notifier
20:40:44 <johnsom> It is being driven by the senlin folks.
20:41:10 <xgerman> I reviewed that patch and they have some work to do
20:41:13 <johnsom> I have some concerns about it
20:41:29 <johnsom> I am also very rusty on that olso notifications stuff they are using.
20:41:47 <xgerman> well, it’s just a shim over rabbit topics
20:41:52 <johnsom> I am interested in the rest of the community comments on this proposal
20:42:21 <johnsom> Yeah, it's more publish-subscribe than the queuing we are using now right?
20:42:30 <nmagnezi> johnsom, sorry for asking but what is senlin?
20:42:49 <kobis> Im guessing that his should enable stats and status collection form lbaas?
20:42:51 <johnsom> #link https://wiki.openstack.org/wiki/Senlin
20:43:01 <johnsom> Don't be sorry, I had to look it up myself.
20:43:16 <nmagnezi> oh, yet another open stack project i didn't knew existed
20:43:17 <nmagnezi> nice :D
20:43:32 <xgerman> there are new ones every day
20:43:45 <johnsom> My initial question is do we want to enable this type of messaging or drive folks towards just using the API.
20:43:57 <johnsom> There are trade offs both ways in my oppinion
20:44:10 <xgerman> to kobis using some topic might be the wrong approach for stats
20:44:40 <johnsom> Then I want to make sure we don't ddos the oslo messaging queues because there is nobody subscribed to these
20:44:54 <xgerman> johnsom as long as it is pluggable we can do both
20:44:58 <johnsom> Yeah, that was also a bit different
20:44:59 <kobis> xgerman: sure - maybe we should just advise about how _we_ prefer to implement stats
20:45:43 <xgerman> yeah, and make them all use kafka :-)
20:45:55 <johnsom> zeromq
20:46:55 <johnsom> So, I'm not hearing feedback that it's bad in concept, we just need to work on implementation details.
20:47:21 <johnsom> I get fidgety when we are integrating with other projects outside of the API....
20:48:00 <xgerman> yeah, but we were talking about some pluggable reporter all along
20:48:03 <johnsom> Ok, please comment any concerns on the blueprint and patch
20:48:29 <johnsom> #topic Open Discussion
20:48:33 <johnsom> Other topics for today?
20:48:41 <xgerman> Horizon?!
20:49:10 <xgerman> we need to see if we can get that to work again
20:49:15 <johnsom> Ok, what about it.  I asked if folks could help with it last week.  More on that topic?
20:49:34 <xgerman> do we have a list of what is broken?
20:49:38 <johnsom> Yes, I think ankur-gupta-f was looking at it.
20:49:54 <xgerman> ok, cool
20:50:07 <johnsom> #link https://bugs.launchpad.net/neutron-lbaas-dashboard
20:50:13 <johnsom> That is the bug list for it.
20:50:18 <johnsom> That is a good place to start
20:50:32 <diltram> ankur failed on this
20:50:35 <johnsom> I don't think ankur-gupta-f got very far looking at it however.
20:50:48 <xgerman> yeah, noone of us is horizon
20:51:10 <johnsom> Right, I think all of the folks that were working on it have changed jobs now....
20:51:22 <johnsom> So, help is needed.
20:51:28 <xgerman> yep
20:52:05 <xgerman> who is the PTL over there those days?
20:52:19 <johnsom> I am, it is now an Octavia project
20:52:28 <johnsom> Or are you asking about horizon?
20:52:35 <xgerman> yep, horizon
20:52:54 <johnsom> Richard Jones
20:53:24 <xgerman> ok, wil look him up
20:54:24 <johnsom> #link https://governance.openstack.org/tc/reference/projects/horizon.html
20:54:59 <johnsom> Any other topics for the last few minutes?
20:55:19 <xgerman> he likes to fix bugs: http://lists.openstack.org/pipermail/openstack-dev/2016-September/103515.html
20:55:21 <xgerman> :-)
20:55:46 <johnsom> Grin
20:56:11 <johnsom> Ok, thanks folks.  Enjoy your break from Octavia/LBaaS meetings!
20:56:30 <johnsom> #endmeeting