5114b61d52
When a migration is reverted the source host allocation held by the migration is swapped back to the instance. If the instance is bandwidth aware then also the allocation key of the port's binding:profile needs to be updated to point back to the resource providers of the source host provider tree. To be able to do that we have to re-calculate the request group - resource provider mapping for the source host based on the resource requests in the neutron ports of the instance and the resource allocation of the instance on the source host. Alternatively we could store such mapping in the MigrationContext during the move operation. blueprint: support-move-ops-with-qos-ports Change-Id: Ib50b6b02208f5bd2972de8a6f8f685c19745514c |
||
---|---|---|
.. | ||
client | ||
filters | ||
weights | ||
__init__.py | ||
driver.py | ||
filter_scheduler.py | ||
host_manager.py | ||
manager.py | ||
request_filter.py | ||
rpcapi.py | ||
utils.py |