17:00:16 <ildikov> #startmeeting cinder-nova-api-changes
17:00:16 <openstack> Meeting started Mon Nov  7 17:00:16 2016 UTC and is due to finish in 60 minutes.  The chair is ildikov. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:21 <openstack> The meeting name has been set to 'cinder_nova_api_changes'
17:00:26 <ildikov> scottda ildikov DuncanT ameade cFouts johnthetubaguy jaypipes takashin alaski e0ne jgriffith tbarron andrearosa hemna erlon mriedem gouthamr ebalduf patrickeast smcginnis diablo_rojo gsilvis  xyang1 raj_singh lyarwood
17:00:30 <smcginnis> o/
17:00:54 <ildikov> smcginnis: hi :)
17:01:01 <mriedem> hangouts today?
17:01:12 <ildikov> mriedem: that's the plan
17:01:24 <ildikov> I asked hemna to help out to get it recorded, etc
17:01:42 <ildikov> jgriffith: johnthetubaguy: are you around?
17:02:02 <jgriffith> o/
17:02:15 <johnthetubaguy> o/
17:03:12 <johnthetubaguy> do we have the hangout link yet?
17:03:16 * ildikov is looking at hemna now :)
17:03:38 <mriedem> does it need to be recorded? ideally yes but if we have the people let's just get started.
17:03:50 <mriedem> recap later in the ML if needed
17:04:20 <ildikov> mriedem: no, we just thought this is the best way of not loosing info and get people on track, who missed it
17:04:26 <ildikov> here's a link: https://hangouts.google.com/call/x5nrih3zinbvdbb66cqdvlnn2me
17:04:40 * smcginnis is at a coffee shop and has no headphones so will catch up afterward...
17:05:12 <ildikov> I'll try to get some notes while our experts talk
17:05:58 <smcginnis> ildikov: thanks!
17:07:32 <ildikov> #info we are trying to figure out whether to go with v1 or v2
17:07:44 <jgriffith> link to v2 https://review.openstack.org/#/c/387712/2/cinder/api/v3/attachments.py
17:08:54 <smcginnis> Did we get a better understanding in BCN that v1 could work for Nova?
17:09:11 <ildikov> #info it would be great having a more simpler approach
17:09:19 <ildikov> smcginnis: I don't think so
17:09:38 <scottda> hi. Sorry, forgot about time change
17:09:53 <ildikov> scottda: no worries
17:10:16 <ildikov> scottda: we're having a Hangout today: https://hangouts.google.com/call/x5nrih3zinbvdbb66cqdvlnn2me
17:11:26 <ildikov> v1 approach: #link https://review.openstack.org/#/c/327408/
17:12:19 <ildikov> summit session etherpad: #link https://etherpad.openstack.org/p/ocata-nova-summit-cinder-session
17:13:17 <ildikov> #info reserve call remains as is
17:13:44 <ildikov> #info it helps avoid races with multiple attach calls, etc.
17:15:26 <ildikov> #info should we have a reserve that creates an attachment_id?
17:17:34 <ildikov> #info the old or new version of the reserve call will get the instance_uuid as the Nova side has that the earliest
17:19:03 <ildikov> #info in the new and shiny version of the proposal reserve will create the attachment record in Cinder and return the id
17:19:47 <ildikov> #info regarding v1 and v2 John Garbutt thinks we need all the v2 calls
17:20:10 <mriedem> hemna: we're all talking
17:20:13 <scottda> we can hear you hemna
17:20:24 <scottda> It is you.
17:20:29 <scottda> We are talking\
17:20:36 <smcginnis> It's not me, it's you? :)
17:20:36 <scottda> You are talking at the same time
17:20:37 <hemna> dangit
17:20:41 <hemna> sorry!
17:20:55 <ildikov> hemna: no worries
17:21:04 <ildikov> hemna: blame Monday! :)
17:22:21 <hemna> that it is!
17:24:44 <ildikov> #info Nova can store additional info and handle locks to not have an attach, while we're detaching
17:25:01 <scottda> hemna: We still hear you
17:27:51 <ildikov> #info we need to shorten the window and have locks to eliminate races
17:28:35 <ildikov> #info Cinder returns info about either safe to detach or the info about the driver whether it shares connection or not
17:29:09 <hemna> I cannot connect to the hangout and get audio
17:29:10 <hemna> :(
17:29:40 <ildikov> hemna: :S, I have no clue how to fix that
17:30:09 <hemna> trying a reboot :(
17:37:40 <hemna> safari works...chrome doesn't.
17:47:08 <ildikov> #info locking needs to be done on Nova side, Nova gets all the info from Cinder that's needed
17:49:10 <ildikov> #info if we have create_attachment, update_attachment and remove_attachment, we should not have reserve, let create handle the lock on Cinder side
17:49:52 <ildikov> #info one attach at a time even in case of multiattach just to be safe
17:50:51 <ildikov> #info after create, Nova sends the connection info to Cinder
17:51:29 <ildikov> #info Cinder returns the info about shared or not shared connection, so Nova has the info for disconnect
17:51:48 <ildikov> #info do not remove attachment from Cinder before disconnect
17:53:35 <ildikov> #info Nova removes the attachments and calls disconnect when no further attachments remained
17:54:32 <ildikov> #info live migrate: same instance_uuid but different hosts is enabled in Cinder
17:54:59 <ildikov> #info we need both connections at the same time for making live migrate work
18:07:33 <ildikov> #info John Garbutt works on live migrate flow and update the spec
18:08:17 <ildikov> #info John Griffith updates v2 with the agreements from today's call on create/delete/update attachment
18:09:06 * johnthetubaguy nods
18:09:41 <ildikov> johnthetubaguy: :)
18:10:14 <ildikov> #info Ildiko handles meeting minutes and administration
18:10:38 <ildikov> thank you! see you all this time next week!
18:10:43 <ildikov> #endmeeting