On 8/28/2018 9:07 AM, Chris Dent wrote:
On Tue, 28 Aug 2018, Bob Ball wrote:
Just looking at Naichuan's output, I wonder if this is because
allocation_ratio is registered as 0 in the inventory.
Yes.
Whatever happened to cause that is the root, that will throw the
math off into zeroness in lots of different places. The default (if
you don't send and allocation_ratio) is 1.0, so maybe there's some
code somewhere that is trying to use the default (by not sending)
but is accidentally sending 0 instead?
If cpu_allocation_ratio isn't in nova.conf, which it's not in this CI
run, then it should default to 16.0 via the ComputeNode object code:
https://github.com/openstack/nova/blob/6bf864df771edb8c0d0af8a868dde21e3d12481e/nova/objects/compute_node.py#L201
Which is used to set the allocation ratio on the VCPUs inventory here:
https://github.com/openstack/nova/blob/6bf864df771edb8c0d0af8a868dde21e3d12481e/nova/compute/resource_tracker.py#L106
Nothing has changed here recently as far as I know.
--
Thanks,
Matt
__________________________________________________________________________
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
