Testing my problem has taken nearly 2 weeks, also because my quick tests with tar options but on smaller directories do not produce the erroneous results as the full amdumps do, which take long times. The tests were also rather haphazard.
I might try to identify the problems, but will do so more systematically and using amanda files and no tape, so I can check the contents of the created tar files. I am not too familiar with linux nor amanda and would like to have confirmation of my assumptions: - the gnutar-list files are in fact the straight incremental files created by tar. - The header contains the time stamp (sec-nanosec) and this time (which is the time that the file was made by tar) is valid for all files (i.e. there are no individual time stamps for each file). - this file is also modified each time an incremental tar is done, so I suppose amanda makes a copy of the *_0 file first (for level > 0) and works with the copy, so the *_0 remains intact (until of course a new level 0 is done). - the exclude options in the dumptypes are assembled in temporary files, which are than passed as an '--exclude-from=file' option to tar. All correct? Regards, Charles On Wed, 26 Feb 2014 23:13:32 +0100 Charles Stroom <[email protected]> wrote: > and this evening the first scheduled backup and it failed miserably > with exactly the same symptoms as before: all level 1 DLEs are equal > in size to a level 0, with the exception of only 2 DLEs (and heaven > may know why!) > > I think I give up. > > Regards, Charles > > -- Charles Stroom email: charles at no-spam.stremen.xs4all.nl (remove the "no-spam.")
