[openstack-dev] [POC] Introduce an auto-converge policy to speedup migration

Chao Fan fanc.fnst at cn.fujitsu.com
Thu May 11 09:58:11 UTC 2017


Hi all,

We plan to develop a policy about auto-converge, which can set cpu
throttle value automatically according to the workload
(dirty-pages-rate). It uses the API of libvirt to set the
cpu-throttle-initial and cpu-throttle-increment. 
But the spec file of nova shows the dependent API is not accepted
by OpenStack:

The initial decrease and increment size can be adjusted during
the live migration process via the libvirt API. However these API calls
are experimental so nova will not be using them.

So I am wondering if OpenStack is willing to use this API and accept
the policy mentioned above.

My RFC PATCH of QEMU and test result are here:
https://lists.gnu.org/archive/html/qemu-devel/2017-01/msg03479.html

Thanks,
Chao Fan





More information about the OpenStack-dev mailing list