hubert@entropy.UUCP (Steve Hubert) (04/08/86)
I am having a strange problem which I have not run across before.
We are running 4.3BSD on a VAX11/785. The tape drive is a TU80
and the disks are RA81s. The uda device driver is version 6.12
from Berkeley (9/16/85) which seems to be equal to or derived from
a DEC driver from January 84. We have a particular mag tape which was
made on a TOPS 20 system which has 817 blocks of length 11200 bytes.
The data is ASCII with a newline and a return embedded in the
data every 58 bytes. I have tried to read the tape several times
with the following command:
dd if=/dev/rmt8 of=x bs=11200
I have successfully done this with
dd if=/dev/rmt8 of=x bs=11200 count=n
with n = 1, and 10 and the data comes off correctly. On every
occassion on which I've attempted to read a large chunk of the data
I have eventually gotten
write: I/O error
from dd after which the filesystem is corrupted. An attempt to remove
the file will result in
bad block 181818927, bad block 181818927, ino 28
...
messages on the console. The block numbers and inode numbers vary.
The system seems to hang when this occurs and can only be brought back
to life by rebooting and clri'ing the inode. The referenced inode is,
of course, the inode relating to the file x. I have tried this on
different filesystems on different disks all with the same result.
The write error doesn't come at the same place every time. It seems
to be random. There is no tape read error and no console error message
from the uda.
I don't understand how it is possible for dd to produce a write error.
I have done dd's of filesystems from one disk to another with bs=10k and
encountered no problems. The problem seems to be related to this
particular tape, somehow.
Oh, afterwards, fsck says
PARTIALLY TRUNCATED INODE I=28 SALVAGE?
..
181818927 BAD I=28 CLEAR?
Thanks in advance for any help.
Steve Hubert
Dept. of Stat., U. of Wash, Seattle
{decvax,ihnp4,ucbvax!lbl-csam}!uw-beaver!entropy!hubert
hubert%entropy@uw-beaver.arpa