Non-unique node name for the Remote Client Agent service (also known as the VSS Requestor nodename) can cause a failure for VSS backup and/or restore
The Data Protection for Exchange VSS backup may fail with the error:
ACN5073E None of components entered are in a state to be backed up.
A trace may show the passive databases as "dismounted"
ACN5072W Component <DB1> is dismounted -- skipping
A restore may attempt to restore the database to the wrong machine.
Or an alternate restore may attempt to overwrite the production database.
The wrong machine is contacted to perform the VSS processing
Diagnosing the problem
Check that the Node Name in the dsm.opt file for the VSS Requestor (Remote Client Agent Service) is only utilized on one machine.
Resolving the problem
When performing a VSS backup, the Tivoli Storage Manager Server is contacted during the proxy node initialization and will provide the Data Protection client with the TCP/IP connection information for contacting the Client Acceptor and Remote Client Agent services. This connection information is retrieved from the Tivoli Storage Manager database, based on the nodename that is being specified for the proxynode interface (VSS Requestor).
If the nodename that is used for the VSS Requestor is not unique, the information that is contained in the Tivoli Storage Manager database may specify the TCPIP address for a different machine than the one that is currently be used. When the wrong machine is contacted, based on this 'incorrect" information, sporadic results can occur. This 'incorrect' information can occur if a nodename is shared on multiple systems. Each time a node on a different system connects, the connection information is updated on the Tivoli Storage Manager Server, which can change the IP address or other client information.
For example, if a VSS restore is performed to an alternate machine using the same Backup/Client nodename as is utilized on the production, this can redirect the restore to the production machine, based on the TCPIP address, and result in overwriting the production database. It is possible to perform an alternate restore, but a unique nodename would need to be utilized (and configured with the necessary proxynode permissions).
It is mandatory that the node name be unique for the VSS Requestor on each machine across the entire environment.