6. Make sure the tape is of the correct size for the drive.
Assuming none of the above solves the problem after performing
a Master backup, then we will proceed with more tests.
How much data gets written to the tape before the process fails?
If there is only a few files written before the error, then it
is possible nothing gets written to the tape and the files are
in memory.
If, for example, you are able to write 100MB to the tape before
an error, then we must see if the problem is the file or directory
that displays at around 100Mb. Assuming the error is consistent
at occurring at 100Mb.
7. Do a selective backup of JUST the directory where the error
occurs. If this is successful, then the file where the error
occurred is not the problem. If it does fail on this same
directory, take the tape drive out of the scenario.
Meaning write an archive to a data file rather to the tape device.
For example:
This is the file that seems to fail: /usr/meddata/html/index.html
Back it up to a file name using the following command.
# tar cvbf 20 /tmp/junk.tar /usr/meddata/html.
Notice we took BackupEDGE (edge) out of the picture also,
in the event you have to contact a hardware vendor, they will
not be able to pass it back to BackupEDGE. Also we backed up
the entire directory '/usr/meddata/html' and not just the single
file index.html.
If this is successful use BackupEDGE to bit level the archive.
# edge TTvbf 20 /tmp/junk.tar
If this is successful, then the problem points back to the chain
of hardware related to the tape drive.
Remember to remove /tmp/junk.tar. If this fails, then your problem
is not specific to the tape drive, and points more to the controller
or cabling (including termination).
7. If number 6 is successful, and the tape consistently dies at
the same spot( ex:100Mb), then the problem points more towards
the drive heads, or ribbon cable.
8. Replace the ribbon cable first. Least expensive, most overlooked.
9. Replace the tape drive
10. Replace the controller.
With problems such as this, consistency in the most difficult
thing to pin point.
Remember the driver is reporting errors to the kernel. Try to
get tar to fail also to convince yourself of a hardware issue.
Attempt a complete tar archive, log in as root and type:
# tar -cvf /dev/rStp0 .
Allow this to run, if this command fails, it too will
generate an error.
