How Can I Tell Whether My Tape Problem Is Hardware or Software?

When you have backup or recovery issues, the first step is to identify whether it is a software or hardware problem. For recovery problems, you can use od -c to read the beginning of a tape or archive file and make sure it is still in a readable format. This approach does not work if software compression is used.

Table 9-4 describes the file record layout for tar.

Table 9-4. File Record Layout for tar

Field Size	Field Descriptor	Detail
100	name	Name of file
8	mode	File mode
8	uid	Owner user ID
8	gid	Owner group ID
12	size	Length of file in bytes
12	mtime	Modify time of file
8	chksum	Checksum for header
1	typeflag	Type of file
100	linkname	Name of linked file
6	magic	USTAR indicator
2	version	USTAR version
32	uname	Owner user name
32	gname	Owner group name
8	devmajor	Device major number
8	devminor	Device minor number
155	prefix	Prefix for filename

The following is the tar file header (USTAR format):

# od -c /dev/st0 |more 0000000   w   o   r   k   /  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0 0000020  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0 * 0000140  \0  \0  \0  \0   0   0   0   0   7   5   5  \0   0   0   0   0 0000160   0   0   0  \0   0   0   0   0   0   0   0  \0   0   0   0   0 0000200   0   0   0   0   0   0   0  \0   1   0   3   3   7   6   7   7 0000220   2   6   7  \0   0   1   0   5   6   0  \0       5  \0  \0  \0 0000240  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0 * 0000400  \0   u   s   t   a   r          \0   r   o   o   t  \0  \0  \0 0000420  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0 0000440  \0  \0  \0  \0  \0  \0  \0  \0  \0   r   o   o   t  \0  \0  \0 0000460  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0 * 0001000   w   o   r   k   /   v   m   w   a   r   e   /  \0  \0  \0  \0 0001020  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0  \0 * 0001140  \0  \0  \0  \0   0   0   0   0   7   5   5  \0   0   0   0   0 0001160   0   0   0  \0   0   0   0   0   0   0   0  \0   0   0   0   0 0001200   0   0   0   0   0   0   0  \0   1   0   3   4   0   1   1   6 0001220   0   2   5  \0   0   1   2   0   2   7  \0       5  \0  \0  \0 # dd if=/dev/st0 of=/tmp/foo.out bs=1k count=4; file /tmp/foo.out /tmp/foo.out: POSIX tar archive

As you can see, the first field is the file or directory name, as you would expect. All the other fields including magic look intact.

The same can be done with cpio, as shown in Table 9-5.

Table 9-5. File Record Layout for cpio

Field Size	Field Descriptor	Detail
6	magic	Magic number "070707"
6	dev	Device where file resides
6	ino	I-number of file
6	mode	File mode
6	uid	Owner user ID
6	gid	Owner group ID
6	nlink	Number of links to file
6	rdev	Device major/minor for special file
11	mtime	Modify time of file
6	Namesize	Length of filename
11	filesize	Length of file to follow

The ASCII cpio file header looks like this:

# od -c /dev/st0 0000000   0   7   0   7   0   1   0   0   0   0   4   d   b   b    0   0 0000020   0   0   8   1   a   4   0   0   0   0   0   0   0   0    0   0 0000040   0   0   0   0   0   0   0   0   0   0   0   0   0   1    4   2 0000060   d   d   6   2   4   e   0   0   0   0   0   0   3   0    0   0 0000100   0   0   0   0   f   d   0   0   0   0   0   0   0   0    0   0 0000120   0   0   0   0   0   0   0   0   0   0   0   0   0   0    0   0 0000140   0   0   0   0   1   1   0   0   0   0   0   0   0   0    /   e 0000160   t   c   /   r   e   s   o   l   v   .   c   o   n    f  \0  \0 # dd if=/dev/st0 of=/tmp/foo.out bs=1k count=4; file /tmp/foo.out /tmp/foo.out: ASCII cpio archive (SVR4 with no CRC)

That technique does not work as well with dump because the tape format is more complicated. You can still use dd to read the header and then run file against it, though.

[View full width]
# dd if=/dev/st0 of=/tmp/foo.out bs=1k count=4;  file /tmp/foo4.out /tmp/foo4.out: new-fs dump file (little endian),  This dump Sat Nov 19 14:37:02 2005, Previous dump Wed Dec 31 19:00:00  1969, Volume 1, Level zero, type: tape header, Label none, Filesystem  /work, Device /dev/mapper/vg00-test, Host linux, Flags 3

Another common problem is when you can read a tape on your system but not on others. This problem is usually caused by misalignment of the tape heads on the drive. The tracks are readable and writable on your drive because they align. When the tape is taken to another system, the heads are not aligned with the tracks, so the tape is unreadable. An additional problem results when you have the tape drive fixed. This makes the tapes unreadable to your drive.

Backup failures due to hardware issues are much easier to diagnose than software problems. If a tar to a device fails, you can try a cpio. If both fail, the problem is typically hardware. The one exception to this would be a problem with the device file.

If a problem is diagnosed as hardware, I usually follow these steps:

1.	Try multiple tapes.
2.	Run the cleaning tape through five times.
3.	Try a new tape.

If the backups still fail, it is most likely a bad drive.

If you replace a tape drive, and it fails again after a short period of time, it can be maddening. What are the chances of two bad tape drives in a row? Oddly enough, it is most likely your tapes that have ruined the second drive. Tape drive failures are very commonly caused by damaged or fouled heads. These damaged drives can then damage or foul the tapes. If you replace the tape drive, and you still have dirty or damaged tapes, they can destroy the new tape drive. If you have a second drive failure in a row, then you should replace the drive and replace all the tapes.