nova/nova/scheduler
Matt Riedemann a57b990c6b Address old TODO in claim_resources_on_destination
The TODO at the end of the method was based on some old
nova-compute behavior which was removed with change
I39d93dbf8552605e34b9f146e3613e6af62a1774 in Rocky.

Instead of logging a warning, ConsumerAllocationRetrievalFailed
is now raised since without the instance consumer allocations
on the source node during a forced evacuate we cannot proceed
with making those allocations on the destination host.

The method is refactored a bit for clarity while in here and
to drop the big nesting.

Remember that this method is *only* called in the case of a
live migration or evacuate operation with a forced target host
and evacuate is the only case where source_allocations are not
provided.

Change-Id: I988d1bd4d7eb1a01d443e3d93964bd09afcc4929
2019-03-19 18:32:26 -04:00
..
client Cleanup comments around claim_resources method 2019-03-19 09:46:25 -04:00
filters Remove the CachingScheduler 2018-10-18 17:55:36 -04:00
weights Per aggregate scheduling weight 2019-01-21 11:48:44 +08:00
__init__.py Improve hacking rule to avoid author markers 2014-05-05 14:35:20 +02:00
driver.py Remove the CachingScheduler 2018-10-18 17:55:36 -04:00
filter_scheduler.py Fix WeighedHost logging regression 2019-03-05 17:16:23 -05:00
host_manager.py Refactor scatter-gather utility to return exception objects 2018-10-31 15:18:07 -04:00
manager.py Rip the report client out of SchedulerClient 2019-01-16 18:35:24 +00:00
request_filter.py Honor availability_zone hint via placement 2018-05-22 08:56:50 -07:00
rpcapi.py Use long_rpc_timeout in select_destinations RPC call 2018-11-20 09:03:53 -05:00
utils.py Address old TODO in claim_resources_on_destination 2019-03-19 18:32:26 -04:00