> I am using a SCSI tape drive that uses standard data cartidges (DC6525).
> I created an archive on the tape using tar with the -W option to verify the
> archive. It streams along, gets to the end of the tape, then streams back
> the other way. When it gets to the last file, it "shoe shines" for a few
> minutes. It then begins to verify the files and streams along. When it gets
> to the last file it says "tar: /dev/nst0: Cannot read: Input/Output error"
> and repeats this several times. Then it says "tar: too many errors, quitting"
> and "tar: Error is not recoverable: exiting now".
> The cartridge is new. I retensioned it a couple of times before I used it
> and also made some small test archives on the tape and had no problems.
> After having problems with the large archive, I erased the tape, retensioned
> it a couple of more times, and wrote the archive to the tape over again
> and I still had the same problem as before.
> Does the "shoe shining" indicate a problem writing to the tape or is this
> normal behaviour? Is it possible that after retensioning a tape several times
> that it still could be too loose? Any ideas on possible defects in the tape
> or the drive?
> TIA
contents
of a directory. I added one file to the directory, which became the new last
file.
I recreated the archive. This time it got past the orignal last file, but
reported
I/O errors when it got to the new last file which was truncated. The previous
last
file extracted okay, but the current last file did not.
This second attempt should have caused the tape to run past the spot where it
had
stopped last time. This appears to indicate that it wasn't a bad spot on the
tape.
I made a third attempt to make the archive, but this time I changed the blocking
factor. On previous attempts, I had used the default which, according the the
info
pages is 20. On this last attempt I used a blocking factor of 128 and tar
reported
no errors when creating the archive, but it still "shoe shines" for a few
minutes
when it gets to the last file before verifying.