1) From the 'help' pages:
"...
Try the following:
On UNIX, verify that the Media Manager device daemon (ltid) and the NetBackup Volume Manager (vmd) are running. Start them if necessary. On Windows, verify that both the NetBackup Device Manager service and the NetBackup Volume Manager service are running. Start them if necessary.
Note: ltid or the NetBackup Device Manager service is used only if devices are attached to the system.
..."
2) Possibly happening now because you are running it manually. If this is a frequency based schedule, check what the frequency is set to.
e.g. imagine you have a schedule with a backup window between 1800 to 2200hrs and a frequency set to 12 hours. Now say the job fails & you realise this & re-run it at 0945hrs the following morning. The job will not now start until 12 hours later, ie 2145hrs, & if this falls at a point when your NB environment is busy it will have to queue - but it will only be able to queue for 15minutes before its backup window closes at which point it will error out with a 196. If this IS the case you could reduce this frequency or try not running the job manually (goes against the grain I know!).
I'm sure others have alternatives & will be able to explain them much better than that!
3) Really depends on what has changed - some changes may not require a restart, others may just require device manager restarting. To do the whole hog we use "netbackup stop" & "netbackup start" (does not restart pbx tho') on UNIX & I believe it's bpdown/bpup.exe on Windoze?