> We know: > > * It pretty much does what we intend it to do: allocations are added > and deleted on server create and delete. > * On manipulations like a resize the allocations are not updated > immediately, there is a delay until the heal periodic job does its > thing.
We know one more thing. For some reason we're overrunning the vcpu capacity in a normal tempest run. It doesn't seem to affect any other resources, though. We are configured to not use CoreFilter, which means the scheduler isn't worried about overcommit or honoring the cpu_allocation_ratio value. I put up a devstack change to hack it up to 4.0, thinking that that would give us enough room to stop getting the errors (4 VCPUS x 4.0 = 16), but it doesn't: https://review.openstack.org/#/c/364581/ You can see in anything that runs placement that it's unhappy: > 2016-09-02 00:34:11.768 18251 WARNING nova.scheduler.client.report > [req-cde54d5f-bcef-4670-8864-eaf479cc9bb9 > tempest-ServersAdminTestJSON-1247365597 > tempest-ServersAdminTestJSON-1247365597] Unable to submit allocation for > instance abc7661d-f258-4eed-8c0c-91b17216d32c (409 409 Conflict > > There was a conflict when trying to complete your request. > > Unable to allocate inventory: Unable to create allocation for 'VCPU' on > resource provider '5295c607-fbb8-472d-8e3d-6067b8814ef8'. The requested > amount would exceed the capacity. We should try to get this figured out before newton ships if possible. I don't think I see it locally, but I have a large dev machine, so I'll have to try to poke it harder. --Dan __________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
