3dcc7955b9
The addition of the global Location setting for Resource object broke clustering because it was relying on the Location header to find the Action object associated with the delete action. To fix that, instead of exposing that location on a location property, construct an Action object with the id pulled from the location header in the delete method. This way there is an object with a status property already. fetch will need to be called on the Action returned to fill in status information - but since wait_for_status and wait_for_delete do that already, it should work with those systems as expected. Change-Id: Ifa44aacc4b4719b73e59d27ed0fcd35130358608
14 lines
580 B
YAML
14 lines
580 B
YAML
---
|
|
fixes:
|
|
- |
|
|
Fixed a regression in deleting Node and Cluster resources
|
|
in clustering caused by the addition of the ``location``
|
|
property to all resource objects. Previously the delete
|
|
calls had directly returned the ``location`` field
|
|
returned in the headers from the clustering service pointing
|
|
to an Action resource that could be fetched to get status
|
|
on the delete operation. The delete calls now return an
|
|
Action resource directly that is correctly constructed
|
|
so that ``wait_for_status`` and ``wait_for_deleted``
|
|
work as expected.
|