16:02:58 <dougwig> #startmeeting #kosmos
16:02:59 <openstack> Meeting started Tue Jul 26 16:02:58 2016 UTC and is due to finish in 60 minutes.  The chair is dougwig. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:59 <dougwig> yes
16:03:00 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:03:03 <openstack> The meeting name has been set to '_kosmos'
16:03:07 <dougwig> #topic New folks
16:03:10 <mpbnka> hi
16:03:12 <dougwig> I see some new names.  Who's here?
16:03:21 <RamT> o/
16:03:37 <RamT> This is RamT
16:03:49 <mpbnka> I am Kumar
16:04:08 <RamT> Doug I sent you an email sometime back
16:04:58 <qwebirc15517> Hi I am Praveen
16:05:45 <Penchal> I am Penchal
16:06:02 <RamT> Wanted to understand the architecture of kosmos and how it will work with existing GSLB providers
16:06:23 <RamT> Like A10, route 53
16:06:40 <mugsie> well, route53 is not a real GLB
16:07:18 <mpbnka> I went through the existing kosmos, how are we planing ahead
16:07:25 <mugsie> but A10 would be a plugin - the plugin would have a set interface (create endpoint, update endpoint etc)
16:07:44 <bmidgley> last week dougwig explained his company provides a10 as an appliance to customers
16:08:10 <dougwig> #chair mugsie
16:08:11 <openstack> Current chairs: dougwig mugsie
16:08:14 <dougwig> #chair bmidgley
16:08:15 <openstack> Current chairs: bmidgley dougwig mugsie
16:08:19 <dougwig> just in case my network keeps being wonky
16:08:24 <mugsie> :)
16:08:33 <dougwig> first pass will be having the desginate and lbaas REST apis be our backend.
16:08:41 <dougwig> we should be able to get a workable PoC that way
16:08:47 <RamT> Even F5 has to go the same route right?
16:08:50 <mugsie> yeah
16:09:18 <mpbnka> hmm
16:09:34 <RamT> If we use lbaas and the zone goes down then kosmos will have no way to check the stats
16:09:48 <RamT> But services might still be running behind
16:10:18 <dougwig> well, any existing backend of lbaas or desginate can be used with our first pass.  more native integration could come later, if desired.
16:11:23 <dougwig> RamT: yes, unless we configure hm's in multiple regions.
16:12:26 <RamT> Hm's?
16:12:34 <dougwig> health monitors
16:12:49 <RamT> Sorry didn't get that
16:12:52 <RamT> Ok
16:13:27 <RamT> Health monitors exist with the lbaas provider right
16:13:30 <mugsie> RamT: the health monitors should be checking actuall connectivity as well as the data from LBaaS
16:13:33 <dougwig> RamT: yes
16:13:48 <RamT> If lbaas is down then monitors will be down
16:14:12 <RamT> And more over geographical distribution will not be possible with lbaas
16:14:13 <mugsie> no, they are separate
16:14:16 <dougwig> yes, though i expect when we check on them, we'll notice.
16:14:33 <dougwig> RamT: correct on geo. let me stress that was for our first PoC.
16:14:45 <RamT> Ok
16:14:47 <mugsie> they are separate binaries that are ran
16:15:03 <dougwig> i don't think designate has geo yet, so even though some providers support it, we need an open-source implementation.
16:15:21 <RamT> So from an arcitecture perspective we will be having two plugins one for DBA and one for lbaas
16:15:52 <RamT> Idough: I agree with open source impl
16:16:13 <RamT> One for DNS*
16:16:33 <mugsie> Geo is a hard problem
16:16:48 <mugsie> and is usually *very* custom per environment
16:17:27 <RamT> And when someone writes the plugin for something like A10 then will have to write it the same way?
16:17:50 <dougwig> i expect we'll have configurable dns and hm providers.  the first will be designate and lbaas.  but that doesn't mean we can't add one for route 53 on the dns side, or an ADC vendor on both sides, or just an ADC vendor on the hm side.  we have to be careful that the drivers support one set of features, and that each has an open source implementation.
16:17:52 <mugsie> no. designate will have to be generic
16:18:04 <dougwig> drivers are just an abstraction to a given vendor.  the REST api doesn't change.
16:18:19 <RamT> Prrfect
16:18:24 <RamT> Perfect
16:18:41 <RamT> I like this approach
16:19:12 <RamT> Can we get together to review the api blueprint?
16:19:16 <dougwig> is that jiving with what you remember, mugsie?
16:19:40 <mugsie> dougwig: ++
16:19:59 <dougwig> RamT: let's plan on reviewing that at the next meeting.
16:20:09 <mugsie> RamT: yeah, it was merged previously - https://github.com/openstack/kosmos-specs/blob/master/specs/liberty/api.rst
16:20:12 <mugsie> #link https://github.com/openstack/kosmos-specs/blob/master/specs/liberty/api.rst
16:20:16 <RamT> Yeah I will work on that at my end
16:20:45 <dougwig> please review this week, and next week we can discuss any issues you have, and if we need changes, or just to get everyone on the same page.
16:20:53 <RamT> Ok
16:22:32 <RamT> Anything else that we need to discuss?
16:24:22 <dougwig> not by me. welcome to all the new faces.
16:24:24 <dougwig> anyone else?
16:24:54 <RamT> dougwig: thank you...
16:24:57 <RamT> I am good
16:25:01 <mpbnka> nope, thanks
16:25:05 <dougwig> alright, bye folks.
16:25:12 <RamT> Bye
16:25:14 <mugsie> o/
16:25:15 <dougwig> you can also always chat in the openstack-gslb channel
16:25:17 <dougwig> #endmeeting