Producing Bad Dumps

Martin McCormick martin at
Fri May 29 14:52:40 UTC 2009

I use the following flags to create a level 0 dump:

dump 0ufaL /home/backups/backup /dev/DISKPARTITION

The dump appears to run just fine. /home/backups/backup is a
pipe to a remote system that fills a regular file from the pipe.

	Everything seems to run well at the time and the dump
file has gigabytes of data in it. I can restore many files from
it and all seems well.

	Today, I practiced restoring a whole system from one of
these dumps and used the following command:

restore -u -fx FILENAME

It prompted for the volume number which is 1 (100% of the dump)
and then I entered none when prompted for the next volume.

	That was about an hour ago and it is still spewing out
the names of thousands of files, many of them OS-related such as
/usr/src/xx which were not being modified or created at the time
so if any files should be there, these should.

	Any idea as to what I did wrong?

	At this point, it is not certain whether the dump is bad
or the restore is bad, but it isn't exactly confidence-en spiring
if the system in question was to melt.

	No file systems filled up and the pipe isn't taken down
until the dump has finished, at least that is what I believe to
be the case.

	Any suggestions are welcome.

	Actually, for this test, I pretended that a directory
on the system called scratch is / so I am just testing the
ability to restore what should be everything under /
before actually trying this on the real / because after that,
you must rebuild the system from CDROM for a proper test.
	Thank you.

More information about the freebsd-questions mailing list