| *** johnw has joined #openstack-lbaas | 00:46 | |
| dougwig | blogan: sanity check. i recall you saying there was magic in the plugin to hold off telling drivers about listeners until a load balancer was created; correct me if i'm wrong, but there doesn't need to be any magic; since loadbalancer_id is in the listener model, and model link will already auto-magically call listener update, right? | 01:08 |
|---|---|---|
| *** fnaval has quit IRC | 03:40 | |
| *** mlavalle_ has joined #openstack-lbaas | 03:44 | |
| *** mlavalle has quit IRC | 03:46 | |
| *** mlavalle_ is now known as mlavalle | 03:46 | |
| *** mlavalle has quit IRC | 03:49 | |
| enikanorov__ | dougwig: not sure i understand. do you mean it communicates with lbaas plugin directly? | 04:40 |
| dougwig | Yes | 04:41 |
| enikanorov__ | so it's agent in itself | 04:48 |
| *** openstackgerrit has quit IRC | 06:43 | |
| *** fnaval has joined #openstack-lbaas | 15:24 | |
| *** fnaval has quit IRC | 16:09 | |
| *** fnaval has joined #openstack-lbaas | 16:17 | |
| *** woodster__ has joined #openstack-lbaas | 16:57 | |
| *** fnaval has quit IRC | 17:40 | |
| *** fnaval_ has joined #openstack-lbaas | 17:42 | |
| dougwig | fyi, today is the spec approval deadline. | 18:33 |
| dougwig | if you're down to just nits on something, i'd suggest +1'ing, unless they're important enough for a slip to K. | 18:34 |
| dougwig | enikanorov__: i think when we say "agent less", we mean, "does not run a separate process for rpc", but you could interpret it as being the agent in itself. | 18:55 |
| enikanorov__ | dougwig: but that driver runs in it's own process, right? | 18:59 |
| dougwig | no, it runs inside neutron-server | 18:59 |
| dougwig | (like many of the vendor drivers) | 18:59 |
| dougwig | the plan is to make it back into a agent driver, but to do the simpler agent-less driver in order to get the rest of the lbaas code reviewed and merged first. | 18:59 |
| enikanorov__ | not sure i understand then | 19:01 |
| enikanorov__ | does it spawn haproxy on the API controller? | 19:01 |
| dougwig | yep. | 19:01 |
| enikanorov__ | why? | 19:01 |
| dougwig | just as an interface demonstration, because neutron won't accept code with a reference driver, and this was the quickest path to a reference driver. it's not useful in production for anywhere but the tiniest installs. | 19:02 |
| dougwig | we need this merged (and it's dependencies): https://review.openstack.org/#/c/105610 | 19:03 |
| enikanorov__ | right, it looks like it's only suitable for devstack | 19:03 |
| dougwig | and they won't do that without "a ref driver in gerrit, even if it's not useful." | 19:03 |
| dougwig | yep, exactly. the last plan that i heard was to do the agent-ified version in juno-3. | 19:04 |
| dougwig | but we don't want 105610 to wait, because LOTS of stuff is depending on that (all other vendor drivers, horizon, cli, tis, l7). it's just about making the merges and reviews less painful, more than anything. | 19:05 |
| dougwig | (because they didn't want one 5k line review, either.) | 19:05 |
| *** woodster__ has quit IRC | 19:05 | |
| *** fnaval_ has quit IRC | 19:34 | |
| dougwig | blogan: for 105610, can you swap LOADBALANCERV2 for LOADBALANCER, for the noop driver, in etc/neutron.conf? | 19:49 |
| *** sbalukoff has quit IRC | 20:13 | |
| *** sbalukoff has joined #openstack-lbaas | 20:13 | |
| dougwig | blogan: missing step in your setup wiki: ProgrammingError: (ProgrammingError) (1146, "Table 'neutron_ml2.lbaas_listeners' doesn't exist") | 20:21 |
| *** fnaval has joined #openstack-lbaas | 22:06 | |
| dougwig | ctracey and blogan - results from running the code: | 22:20 |
| dougwig | https://www.irccloud.com/pastebin/oD3Rys4q | 22:20 |
| ctracey | dougwig: aware of some of these. Going to be working on it this evening. | 22:22 |
| ctracey | Specifically the session persistence part. | 22:22 |
| *** mlavalle has joined #openstack-lbaas | 22:27 | |
| *** woodster__ has joined #openstack-lbaas | 22:34 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!