Marking bad blocks for non-use

From: Macdonell, Dennis <DennisMacdonell_at_auslig.gov.au>
Date: Fri, 21 Feb 1997 17:26:00 +1100

Hi All,

We have a system running DU3.2g with Polycentre HSM software v1.2 to
control a jukebox. The system maintains a cache of files on a 2.1gb rz28
magnetic disk (the magnetic disk is efs - extended file system, a form
of the advanced file system), the magnetic disk is made to look like the
complete file structure of the information that is stored on the optical
disks.

This magnetic disk has had a number of bad blocks reported in the
syslog, the badblocks have numbers like _3595017, and are reported as
unrecoverable medium errors. I had the impression that if I unmounted
the disk and used the utility scu I could reassign blocks with the
commands

scu -f <raw disk device>
scu> reassign lba 3595017

When I try this it reports that it is unable to read the block in
question and gives up.

I had previously tried the scu utility on a ufs partition with similar
results. In that situation I resorted to good old fsck and lo it did its
trick on the block in question, and you might say that block ain't no
more.

I have used the utility ssgfsck, which is supposed to handle some of the
fsck operations in the jukebox situation but I have never been able to
get that utility to fix things either. The only other utility I know
about is "checkm" but that is used to correct optical with respect to
data held on the magnetic disk.

I imagine that some one out there has had the need to set aside a bad
block on an advanced file system, and if so I'm assuming that if they
were successful then I should be able to follow the same procedure. Of
course if it involves blowing away all the data and recovering from
backup then I'm not sure I will attempt that strategy.

Dennis
mcdonell_at_auslig.gov.au
Received on Fri Feb 21 1997 - 07:38:43 NZDT

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:47 NZDT