Red Hat Bugzilla – Bug 1284493
Delete instance should veify instance is deleted before returning
Last modified: 2015-11-25 12:58:08 EST
Description of problem:
Nova operations are async and return before they are done. This is problematic during cleanup scripts and automation (Tempest, Khaleesi, etc...) where VM deletion returns before the VM is deleted and subsequent resource deletions will fail because VM is still alive on the system
As a result each script needs to implement a waiter to handle nova resources
Please add a flag that will force nova operations to become sync like other clients and return only when operation is actually done.
$ nova delete <vm> && nova show <vm>
VM is shown in "deleting" state
404 error - VM doesn't exist.
This aynchrony is intended behavior in the nova API and realistically-speaking will not be changed upstream.
Can't we ADD a sync mode where we wait for actions to complete?