NetWorker Architecture and Components
What this domain tests
Architecture questions ask you to separate the control path from the data path, identify which daemon owns each database, and explain how the server, storage node, and client cooperate during a backup or recover.
Server, storage node, client
- NetWorker server: runs the main service (
nsrdon Linux; the NetWorker service on Windows) and owns the resource database under/nsr/res/nsrdb. Policies, clients, devices, pools, and schedules are resources stored here. - Media database: under
/nsr/mm; tracks volumes, save sets, and clone copies. It answers “where is this save set?” - Client file index: under
/nsr/index; tracks individual backed-up file names and attributes. It answers “which files are in this save set?” - Jobs database: tracks running and completed jobs, usually under
/nsr/res/jobsdb. - Storage node: owns devices and runs
nsrmmdto move data. The server can be its own storage node; a remote storage node offloads data movement. - Client: runs
nsrexecd, reads the save set, and sends data to the storage node selected by the server.
Daemons to recognize
nsrd starts and coordinates server daemons. nsrmmdbd manages the media database. nsrindexd manages client indexes. nsrjobd tracks jobs. nsrmmgd and nsrlcpd handle media and library operations. nsrmmd is the data mover on a storage node. nsrexecd runs on the client and launches save or recover processes. NMC and NWUI are management interfaces, not data movers.
Worked example
A backup job starts, then sits at “waiting for media” and fails with a storage-node error. The control path is healthy because nsrd accepted the job and nsrjobd recorded it. The data path is broken: no nsrmmd is available on the selected storage node, the device is not ready, or the target pool has no labeled volume. Check the storage node process, device status, and pool volume state before changing client settings.
Traps
Do not confuse the media database with the client file index. The server does not always move backup data; a remote storage node can. A running NMC console does not prove nsrd is up. Also remember that nsrexecd must be reachable on the client, or the job fails before data movement begins.