| *** puranamr has quit IRC | 00:14 | |
| *** sarob has quit IRC | 01:50 | |
| *** puranamr has joined #akanda | 03:37 | |
| *** puranamr has quit IRC | 04:57 | |
| *** puranamr has joined #akanda | 04:58 | |
| *** puranamr has quit IRC | 04:59 | |
| *** puranamr has joined #akanda | 05:03 | |
| *** gsagie has joined #akanda | 05:06 | |
| *** puranamr has quit IRC | 05:44 | |
| *** puranamr has joined #akanda | 06:05 | |
| *** puranamr has quit IRC | 06:10 | |
| *** puranamr has joined #akanda | 13:32 | |
| *** rods1 has joined #akanda | 13:34 | |
| *** puranamr has quit IRC | 13:43 | |
| *** puranamr has joined #akanda | 13:44 | |
| *** puranamr has quit IRC | 13:57 | |
| *** puranamr has joined #akanda | 14:18 | |
| *** puranamr has quit IRC | 14:23 | |
| *** openstackgerrit has quit IRC | 14:41 | |
| *** openstackgerrit has joined #akanda | 14:41 | |
| *** cleverdevil has joined #akanda | 15:51 | |
| *** sarob has joined #akanda | 16:48 | |
| *** puranamr has joined #akanda | 17:11 | |
| adam_g | anyone wanna take a look at this so we can make the devstack job voting? https://review.openstack.org/#/c/189433/ | 17:19 |
|---|---|---|
| ryanpetrello | +2 adam_g | 17:21 |
| adam_g | thanks! | 17:22 |
| sarob | +2a | 17:25 |
| sarob | adam_g: ^ | 17:26 |
| openstackgerrit | Merged stackforge/akanda-rug: Wait for a configured test router to become active https://review.openstack.org/189433 | 17:31 |
| adam_g | cool, thanks | 17:42 |
| *** puranamr has quit IRC | 19:17 | |
| *** puranamr has joined #akanda | 19:30 | |
| openstackgerrit | Sean Roberts proposed stackforge/akanda-rug: remove docs directory https://review.openstack.org/190312 | 19:33 |
| openstackgerrit | Sean Roberts proposed stackforge/akanda: add state machine bits https://review.openstack.org/190314 | 19:35 |
| *** puranamr has quit IRC | 20:03 | |
| adam_g | markmcclain, do you foresee any issues if notifications and neutron rpc's are consumed between two Process's instead of one like it they are now? | 20:40 |
| markmcclain | so with notifications the recv'ing process should be the only one recieving | 20:41 |
| markmcclain | the tenant workers shouldn't need it | 20:41 |
| markmcclain | is the problem that oslo connection is made before the fork? | 20:42 |
| adam_g | markmcclain, so right now we setup a single kombu connection thats listening for both notifications and RPCs, and processing them as one | 20:43 |
| markmcclain | right so if there are two connections for that process we'll be fine | 20:44 |
| adam_g | with oslo we need to split that into two things: one is an oslo notification listener and the other is an RPC server | 20:44 |
| markmcclain | can we run them threaded? | 20:44 |
| markmcclain | the workers should be able to take on notification from multi-producers | 20:45 |
| adam_g | probably | 20:45 |
| adam_g | is the multirocessing.Queue we pass around thread safe? | 20:46 |
| markmcclain | python threads yes | 20:46 |
| markmcclain | eventlet.. who knows | 20:46 |
| markmcclain | multi-processing and eventlet are strange | 20:46 |
| adam_g | oslo.messaging is pluggable there | 20:46 |
| *** puranamr has joined #akanda | 20:47 | |
| markmcclain | yeah... so we should be able to do it | 20:48 |
| adam_g | so, 1 sub-process that spawns threads for each rpc connection? | 20:48 |
| markmcclain | yeah.. that feels about right | 20:49 |
| adam_g | i know neutron sets up its multi-connection consumption via a 'consume_in_threads()' but i imagine that means eventlet | 20:49 |
| markmcclain | I thought Doug had told me before that it works with python threads too | 20:49 |
| *** dhellmann has joined #akanda | 20:50 | |
| adam_g | markmcclain, AFAICS the way in which messages can be processed on the receiving side can be toggled | 20:50 |
| adam_g | oh, hi dhellmann | 20:51 |
| dhellmann | I believe all of oslo.messaging is thread safe, if you use the threaded executor | 20:51 |
| dhellmann | that's certainly our intent, so if you find otherwise we would consider it a bug | 20:51 |
| dhellmann | hi, adam_g :-) | 20:51 |
| adam_g | cool | 20:52 |
| *** puranamr has quit IRC | 20:57 | |
| *** puranamr has joined #akanda | 20:58 | |
| *** ryanpetrello has quit IRC | 21:20 | |
| *** ryanpetrello has joined #akanda | 21:27 | |
| *** ryanpetrello has quit IRC | 21:30 | |
| *** ryanpetrello has joined #akanda | 21:33 | |
| *** puranamr has quit IRC | 21:47 | |
| adam_g | markmcclain, does the rug actually do anything with messages it gets on the dhcp_agent rpc queue? | 22:01 |
| adam_g | https://git.openstack.org/cgit/stackforge/akanda-rug/tree/akanda/rug/notifications.py#n107 | 22:01 |
| markmcclain | right now no.. the dhcp rpc events are identical to the notifications queue | 22:01 |
| adam_g | ok | 22:02 |
| adam_g | so the only rpc we actually sniff out right now is router_deleted | 22:02 |
| adam_g | (on the l3_agent) | 22:02 |
| adam_g | markmcclain, also, whats the difference between messages that arrive via the 'l3_agent' queue and 'l3_agent.' + host_id ? | 22:03 |
| *** puranamr has joined #akanda | 22:48 | |
| *** puranamr has quit IRC | 22:53 | |
| *** puranamr has joined #akanda | 23:01 | |
| *** rods1 has quit IRC | 23:10 | |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Add a log shim to avoid a dependency on openstack.common.log https://review.openstack.org/190400 | 23:38 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Rework messaging layer for oslo.messaging (WIP) https://review.openstack.org/190401 | 23:38 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!