Steps to reproduce:
1. Protect the Agent and take base image (at least 30 GB)
2. Remove the base image and cancel deferred delete, then restart the core service
3. Wait till repository mounts and again cancel deferred delete, then force new snapshot
4. Cancel snapshot and check the log files
Actual result:
Error 'unmanagedDvm.storageSubsystem' spams Core's log, also these steps cause repository corruption.
Deferred delete tries to delete the same data twice and deletes needed data. That is why repository corruption appears.
Errors:
ERROR 2017-02-22T12:02:54 [54] - unmanagedDvm.storageSubsystem ()
02/22/2017: record_allocator_nas : invalid dfs_record 834233 (ref_count: 0, mode 5) on device 2 before dfs_record deallocation.
ERROR 2017-02-22T12:02:54 [54] - unmanagedDvm.storageSubsystem ()
02/22/2017: record_allocator_nas : invalid dfs_record 834234 (ref_count: 0, mode 5) on device 2 before dfs_record deallocation.
ERROR 2017-02-22T12:02:54 [54] - unmanagedDvm.storageSubsystem ()
02/22/2017: record_allocator_nas : invalid dfs_record 834235 (ref_count: 0, mode 5) on device 2 before dfs_record deallocation.
Expected result:
Deferred delete works fine, DVM does not spam Core's log, and repository stays healthy