COHERENT manpages
This page displays the COHERENT manpage for restor [Restore file system].
List of available manpages
Index
restor -- Command
Restore file system
restor command [dump_device] [filesystem] [file ...]
restor copies to the hard disk one or more files from floppy disks or tapes
written by the command dump.
restor recognizes the following commands:
r Mass restore both full and incremental dump disks/tapes into the
filesystem. The target file system must have enough data blocks and i-
nodes to hold the dump.
The mass restoration is performed in three phases. In phase 1, restor
clears all i-nodes that were either clear at dump time or are going to
be restored. Any allocated blocks are released. Second, it restores
all files on the disk. The i-numbering is preserved; however, data
blocks are allocated in the standard fashion. Third, a pass is made
over the i-nodes and the list of free i-nodes in the superblock is
updated.
Restoration begins immediately, using the currently mounted disk or
tape.
R Like the r command, except that it pauses to ask for numbers of disks or
reels.
t Read the header from the dump. Display the date the dump was written
and the ``dump since'' date that produced the dump.
x Extract each file from the dump and restore it to the hard disk. All
file names are absolute path names starting at the root of the dump (the
first directory dumped, which is always the root directory of the file
system). A numeric file name is taken to be an i-number on the dumped
file system, permitting restore by i-number.
restor looks up each argument file in the directories of the dumped file
system and prints out each name and associated i-number. restor
extracts the files from the currently mounted dump disk or tape, and
writes the extracted files into the current directory. Extracted files
are named after their i-numbers.
X Like command x, except that before it begins, it asks you for the number
of the disk (or the reel number of the dump tape). It continues asking
for dump disks until all files have been extracted or you types
<ctrl-D>.
Each of the above commands can be modified either or both of the following
modifiers:
f Tell restor to take the next argument as the path name of the dump
device (floppy-disk drive or tape drive). If the f modifier is not
specified, restor uses the device /dev/dump.
v Verbose output. Tell restor to print a step-by-step trace of its
actions when restoring an entire file system. This is for discovering
what went wrong when a mass restore runs into trouble.
Restoring from a Damaged Medium
As noted below, dump requires that its output be written to disks or tapes
that are free of defects. Restoring a file system from a damaged medium is
difficult and is not associated with a high probability of success; if,
however, you must try to do so, the following directions will give you a
fighting chance of restoring your data.
1. Use the command fdformat to format a blank disk. Use the command
badscan to examine it for bad sectors; if it does have bad sectors, put
it aside and try another.
2. Use the command dd to copy the bad disk to directory /tmpfoo1 dd should
die at the bad sector in the disk.
3. dd again to directory /tmp/foo2 using that command's skip=n to skip past
the bad sector (or sectors).
4. Repeat step 3 (if it died too) until the end of the disk is reached.
Now you have a set of directories named /tmp/foo[1...n] that contain
parts of the bad disk.
5. Use the command
dd if=/tmp/foo1 of=/dev/fha0
with the new, defect-free disk.
6. Now, use the command
dd if=/tmp/foo2 of=/dev/fha0 seek=whatever
to place foo2 into the right place on the new disk.
7. Repeat 6 for each directory foo3 through fooN.
8. Create a 512-byte file that contain the string
GARBAGE\n
repeated 64 times. Use dd to copy it into new disk where the bad
sectors were.
Now, you should have a disk that is a mirror image of the old, damaged dump
disk. Each bad sectors will have been replaced by 64 iterations of the
string GARBAGE\n. As noted, there is no guarantee that this scheme will
work in every instance (the chances of error are quite high), but it will
give you a fighting chance to save your data.
Files
/dev/dump -- Dump device
/etc/ddate -- Dump date file
See Also
commands,
dump,
dumpdir
Diagnostics
Most of the diagnostics produced by restor are self-explanatory, and are
caused by internal table overflows or I/O errors on the dump medium or file
system.
If the dump spans multiple disks or reels, restor asks you to mount the
next disk at the appropriate time. Type a newline when the disk has been
mounted. restor verifies that this is the correct disk, and gives you
another chance if the disk number in the dump header is incorrect.
Notes
You cannot perform a mass restore onto a live root partition. Instead,
boot a stand-alone version of COHERENT on a floppy-disk drive, or boot from
an alternative COHERENT file system on another hard-disk partition before
you attempt to do a mass restoration.
The handling of tapes with multiple dumps on them (created by dumping to
the no rewind special files) is not very general. Basically, restor
assumes that tapes holding multiple dumps and tapes holding dumps that span
multiple reels are mutually exclusive. You can restore from any file on a
reel by positioning the tape and then restoring with the x or r commands,
which do not reposition the tape. It is (almost) impossible to use the X
or R commands, as the position of the dump tape will be lost when restor
closes it.
dump requires that its output be written to disks that are free of bad
sectors. If you write a dump to a disk with bad sectors, you will not be
able to restore files from that disk. See dump for directions on
processing disks to ensure that they are free of bad sectors.



