nova/releasenotes/notes/resize-api-cast-always-7eb1dbef8f7fe228.yaml
Matt Riedemann a05ef30fb9 Make API always RPC cast to conductor for resize/migrate
This is a follow up to [1] to make the API behave consistently
by always asynchronously casting to conductor during resize
and cold migration regardless of same-cell or cross-cell
migration.

From the end user point of view, not much changes besides
the possibility of some exceptions occurring during scheduling
which would have resulted in a 400 BadRequest error.
The user still gets a 202 response, must poll the server status
until the server goes to VERIFY_RESIZE status or times out, and
can check the instance actions if the resize/migrate fails.

The specific errors that can occur are not really an API contract
and as such end user applications should not be building logic
around, for example, getting a NoValidHost error. It should be
noted, however, that by default non-admin users cannot see
the instance action event traceback that would contain the
error, e.g. NoValidHost.

The only exception types removed from handling in the API are
(1) AllocationMoveFailed which can be raised when the conductor
MigrationTask runs replace_allocation_with_migration and
(2) NoValidHost when the scheduler is called to select destinations.

Because of this, quite a few functional negative tests have to be
adjusted since the API no longer returns a 400 for NoValidHost and
other errors that can happen during scheduling.

Finally, the do_cast kwarg is left on the conductor API method since
the compute service calls it during same-cell reschedule as a
synchronous RPC call and has error handling if rescheduling in
conductor fails.

[1] I098f91d8c498e5a85266e193ad37c08aca4792b2

Change-Id: I711e56bcb4b72605253fa63be230a68e03e45b84
2019-11-13 10:19:53 -05:00

13 lines
656 B
YAML

---
other:
- |
The ``resize`` and ``migrate`` server action APIs used to synchronously
block until a destination host is selected by the scheduler. Those APIs
now asynchronously return a response to the user before scheduling.
The response code remains 202 and users can monitor the operation via the
``status`` and ``OS-EXT-STS:task_state`` fields on the server resource and
also by using the ``os-instance-actions`` API. The most notable
change is ``NoValidHost`` will not be returned in a 400 error response
from the API if scheduling fails but that information is available via the
instance actions API interface.