openstackgerrit | Merged stackforge/akanda-appliance: Fix project URL in setup.cfg https://review.openstack.org/196807 | 01:15 |
---|---|---|
*** puranamr has quit IRC | 02:33 | |
openstackgerrit | David Lenwell proposed stackforge/akanda-rug: Driver code (wip) https://review.openstack.org/195814 | 02:36 |
davidlenwell | adam .. just a heads up.. https://review.openstack.org/#/c/195814 is still very much a wip.. I'm going to write up a spec about what Its trying to accomplish so you have better acceptance criteria | 02:37 |
openstackgerrit | David Lenwell proposed stackforge/akanda-rug: Driver code (wip) https://review.openstack.org/195814 | 03:00 |
*** sarob has quit IRC | 03:54 | |
*** puranamr has joined #akanda | 13:00 | |
*** puranamr has quit IRC | 13:14 | |
*** puranamr has joined #akanda | 14:15 | |
*** puranamr has quit IRC | 14:19 | |
*** rods_ has joined #akanda | 15:01 | |
*** drwahl has quit IRC | 15:25 | |
*** puranamr has joined #akanda | 15:34 | |
*** puranamr_ has joined #akanda | 15:36 | |
*** puranamr has quit IRC | 15:36 | |
*** sarob has joined #akanda | 16:06 | |
*** openstackgerrit has quit IRC | 16:50 | |
*** openstackgerrit has joined #akanda | 16:51 | |
davidlenwell | +1 chicken ;) | 17:07 |
adam_g | davidlenwell, so are the actual bug(s) you were talking about that need fixing? | 17:46 |
adam_g | der, *so what | 17:50 |
davidlenwell | I'll compile a list.. its a lot of small things | 17:57 |
adam_g | k | 17:58 |
adam_g | markmcclain, so it sounds like you had a chance to look at whats on gerrit re rug-ha? | 17:59 |
adam_g | theres one issue: cluster events go onto the same internal notifications queue as neutron events. it could happen that a rebalance event gets added to the back of a couple router events. the rebalance should happen before those router events are processed | 18:03 |
adam_g | trying to think of ways of adding priority to events, short of declaring a separate internal queue for cluster events | 18:03 |
adam_g | maybe declaring a couple new internal queues, priority based? low, normal, critical? i know ryanpetrello had an idea about having health checks take a back seat to CRUD events | 18:04 |
*** puranamr_ has quit IRC | 18:18 | |
*** drwahl has joined #akanda | 18:27 | |
*** sarob_ has joined #akanda | 18:45 | |
*** sarob_ has quit IRC | 18:56 | |
*** puranamr has joined #akanda | 19:17 | |
*** puranamr has quit IRC | 19:18 | |
*** puranamr has joined #akanda | 19:18 | |
*** qwebirc32764 has joined #akanda | 20:17 | |
*** qwebirc32764 has quit IRC | 20:19 | |
adam_g | markmcclain, hmm how critical is the router_id caching that happens @ https://git.openstack.org/cgit/stackforge/akanda-rug/tree/akanda/rug/tenant.py#n120 | 20:41 |
markmcclain | I think dhellmann added that.. I'd have to trace through to seeing what it broke if we removed it | 20:47 |
adam_g | it looks like its just an optimization to avoid a neutron call | 20:48 |
adam_g | for events that have no router_id | 20:48 |
adam_g | was trying to map events to hosts using router_id from the worker, but realized thats not going to work. would it make sense to distribute based on tenant_id instead? | 20:50 |
adam_g | (wrt worker-managed hash ring) | 20:51 |
markmcclain | adam_g: possibly | 21:01 |
markmcclain | wondering if hashing on tenant would give us a lumpy distribution | 21:01 |
markmcclain | but then again a tenant would likely have of have massive quota to really cause it | 21:02 |
markmcclain | we could always move to a tenant based one for now and if needed could switch it up later | 21:03 |
*** puranamr has quit IRC | 21:10 | |
adam_g | markmcclain, the other option is move resolving router_id for events that don't have it outside of the TRM, into the worker but that feels a bit weird | 21:18 |
adam_g | its not entirely clear to me which events dont have an associated router and cause us to do the lookup on the akanda side.. port updates? | 21:19 |
markmcclain | port and subnet updates | 21:33 |
markmcclain | those have ripple effects for the the dhcp service | 21:33 |
*** puranamr has joined #akanda | 21:45 | |
*** puranamr has quit IRC | 21:47 | |
*** puranamr has joined #akanda | 21:48 | |
*** puranamr has quit IRC | 21:59 | |
openstackgerrit | Adam Gandelman proposed stackforge/akanda: Add spec for RUG HA https://review.openstack.org/193680 | 22:01 |
*** puranamr has joined #akanda | 22:05 | |
*** rods_ has quit IRC | 22:13 | |
*** sarob_ has joined #akanda | 23:07 | |
openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Add support for hash-based RUG scale out (WIP) https://review.openstack.org/195366 | 23:14 |
*** sarob has quit IRC | 23:14 | |
*** sarob_ is now known as sarob | 23:16 | |
*** sarob_ has joined #akanda | 23:21 | |
sarob_ | Sarob: test | 23:21 |
sarob_ | Sarob: again | 23:25 |
sarob_ | Again | 23:26 |
sarob_ | Sarob: again | 23:26 |
*** sarob_ has quit IRC | 23:27 | |
*** puranamr has quit IRC | 23:36 | |
*** puranamr has joined #akanda | 23:37 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!