Your description is misleading:
"the index file is there, is about 9MB, but contains no save sets".
There are several issues:
- the file index info is stored in a set of 3 files in
the ..\nsr\index\client\db6 directory
So it is not only one file
- NW must have the information at least in the media
index, but file index info without a reference in the
media index is simply useless as NW would not be able
to find this.
What do you really mean? - I just guess that the save set
size shows 9MB that have been backed up.
This procedure usually shows more details:
- Create a new directory on your client to identify the
backed up files easily
- Create a test group that you just start of manually
- Create a new client definition for your failing host:
- Assign the group to the failing client
(With NW/Windows Administrator it looks like you
assign a client to group but this is just a
different way how it is visualization to a client)
- For the save set, define the new directory
- Start the group from the command line:
savegrp -v -l full -G group_name
- Investigate the result
- Run mminfo to check how much data really has been saved
and which status the save set has (browsable ?)
- If so, run nwrecover (UNIX) and winworkr - recover
(Windows) to check whether you can see the backed up
files
Potential issues:
- You could also have a 'defect' media index
- If the backup runs successfully
you may have forgotten a local directive (.nsr for
UNIX or nsr.dir on Windows that simply skips files or
directories)
- If using a backup command for the client you may have
forgotten a "save -c client_name" that can be used to
actually fill another client's file index.
- Did you rename the client in the meantime ?
- Is the name resolution really unique ?