| *** tetsuro has quit IRC | 03:58 | |
| *** tetsuro has joined #openstack-placement | 03:59 | |
| *** dansmith has quit IRC | 05:10 | |
| *** dansmith has joined #openstack-placement | 05:14 | |
| *** tetsuro has quit IRC | 05:28 | |
| *** tetsuro has joined #openstack-placement | 05:31 | |
| *** tetsuro has quit IRC | 06:04 | |
| *** tetsuro has joined #openstack-placement | 06:09 | |
| *** gibi_off is now known as gibi | 07:44 | |
| *** belmoreira has joined #openstack-placement | 07:56 | |
| *** martinkennelly has joined #openstack-placement | 09:32 | |
| *** ttsiouts has joined #openstack-placement | 09:47 | |
| *** belmoreira has quit IRC | 10:44 | |
| *** belmoreira has joined #openstack-placement | 10:45 | |
| *** belmoreira has quit IRC | 10:49 | |
| *** ttsiouts has quit IRC | 11:24 | |
| *** ttsiouts has joined #openstack-placement | 11:52 | |
| *** sean-k-mooney is now known as skm | 11:55 | |
| *** skm is now known as sean-k-mooney | 11:55 | |
| *** cdent has joined #openstack-placement | 11:56 | |
| *** cdent has quit IRC | 12:29 | |
| *** cdent has joined #openstack-placement | 12:41 | |
| *** mriedem has joined #openstack-placement | 13:59 | |
| mriedem | cdent: | 15:10 |
|---|---|---|
| mriedem | (9:09:30 AM) mriedem: efried: maybe a troubleshooting item to document, "oh no my placement db is all screwed up, how can i just start over w/ my existing nova" | 15:10 |
| mriedem | (9:09:43 AM) efried: I thought cdent had that somewhere mebbe? | 15:10 |
| mriedem | is eric high on the drugs again? | 15:10 |
| cdent | did he ever stop? I'm pretty it is "still" not "again" | 15:11 |
| cdent | but yes, that's a thing I thought about, but I need to refresh my mental cache | 15:11 |
| * cdent stirs | 15:11 | |
| cdent | So I think the idea there was sort of with regard to violent upgrades: | 15:13 |
| cdent | the placement db is gone, your only placement client is nova it _may_ be safe to: | 15:13 |
| cdent | cycle all the nova-computes, let them report their providers and inventory, and then run heal allocations to get existing instances back into the database | 15:13 |
| sean-k-mooney | this has come up downstream too in a slightly different context downstream where for backup and restore reasons a customer wanted to be able to chagne the hostname on restore. | 15:14 |
| sean-k-mooney | in the downstream context it was a pre palcment deployment | 15:14 |
| sean-k-mooney | but we were wondering what the best way to handel that would be in a placment worlds | 15:15 |
| sean-k-mooney | letting it start up and healing the allocation seamed like the most likely route to success | 15:15 |
| mriedem | cdent: "cycle all the nova-computes, let them report their providers and inventory, and then run heal allocations to get existing instances back into the database" yup that's what i suggested in nova | 15:15 |
| mriedem | along with (1) backing up your placement db before wiping it and starting over and (2) running nova-manage placement sync_aggregates as well | 15:16 |
| * cdent remembers when the nova-computes could heal themselves completely and is sad about it being gone, even though it is gone for good reasons | 15:16 | |
| mriedem | here is an attempt at docs https://review.opendev.org/698517 | 15:42 |
| *** ttsiouts has quit IRC | 15:48 | |
| *** cdent has quit IRC | 15:49 | |
| *** cdent has joined #openstack-placement | 16:02 | |
| *** martinkennelly has quit IRC | 19:07 | |
| *** mriedem has quit IRC | 19:37 | |
| *** cdent has quit IRC | 19:54 | |
| *** martinkennelly has joined #openstack-placement | 20:43 | |
| *** martinkennelly has quit IRC | 21:40 | |
| *** mriedem has joined #openstack-placement | 22:37 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!