From: Carlos Chua (chuacarlos@hotmail.com)
Date: Thu Mar 27 2003 - 10:40:34 EST
Thanks to Dr. Thomas Blinn and Charles Ballowee for the reply.
It seems like I could either get a good copy from the cd-rom or delete away
the file and run some commands to re-create the file.
Below are the replies I received.
Thanks again for the help. I'll try out the solutions once I get the
dowtime.
Reply 1:
It does sound like a hardware problem has corrupted your dsfl.dat file.
(Note that the name contains the letter "l", not the digit "1", there
is not supposed to be a "/etc/dsf1.dat" file on the system.)
If you have a backup of the file, you should restore it from backup.
If you don't have a backup, then you MIGHT get lucky by restoring it
from the file on the installation media, as you suggest.
Good luck. This is why it's important to have backups, and also to
pay attention to the hardware error reporting logs. You really need
to figure out how ingres monitor output got written into the file; if
that's happening with this file, it may be happening to other parts
of your root file system, and that will eventually lead to a total
system failure.
Reply 2:
You can force a rebuild of the hardware db etc. by following the steps in:
http://www.ornl.gov/cts/archives/mailing-lists/tru64-unix-managers/2002/09/m
sg00280.html
I've had luck with the second response listed there, i.e.:
2) Chris' response
I did something like this (installed on an old AlphaServer 1000 for an
ES40). After I did the install and got everything "set", I did the
following on the AS1000 right before shutting down to remove the disk:
#*#*#*#
#!/bin/sh
# Run in the root directory of the filesystem to "clean"
rm ./etc/dec*
rm ./etc/dccd.*
rm ./etc/dcdd.*
rm ./etc/dfsc.*
rm ./etc/dfsl.*
rm ./cluster/members/member/etc/cfginfo
rm ./etc/cfginfo
rm ./cluster/members/member/etc/dfsl.*
rm ./dev/tty0*
rm ./dev/lp*
rm ./dev/kevm*
rm ./dev/scp*
for dir in cport disk rdisk changer tape ntape; do
rm -rf ./devices/$dir
rm -rf ./cluster/members/member/$dir
rm ./dev/$dir
done
rm ./cluster/members/member/.Booted
#*#*#*#
That basically wipes out the hardware database, and when you boot (on
the real system), it will be recreated for the real hardware.
--------------------------------
I don't know that it's necessarily supported, but it does work. Just
be careful if you have devices that will be sorted in a different order
to make note of which dsk device names map to which devices. hwmgr -show
scsi
before you do that and save the output somewhere so you can shuffle things
around before going past single user on restart.
----- Original Message -----
From: "Carlos Chua" <chuacarlos@hotmail.com>
To: <tru64-unix-managers@ornl.gov>
Sent: Tuesday, March 25, 2003 11:12 PM
Subject: How to fix /etc/dsfl.dat corrupted?
> Hi,
>
> I'm trying to add a new disk to my Alphaserver ES40 running Tru64 Unix
> v5.1A. After running "hwmgr -scan scsi", the system cannot create the
device
> file. When I run "dsfmgr -v", the system complains that it can't open the
> /etc/dsf1.dat although the file is present.
>
> When I run "strings /etc/dsf1.dat", the output will show something like a
> device listings, however, in between there would be some ingres monitor
> output. Does this looks like my dsfl.dat is corrupted? if yes, what's the
> proper way to fix the problem? Can copy the file from the installation
> cd-rom and then run dsfmgr -F to recover? or is it more complicated than
> that?
>
> Thanks in advance for any advice.
>
> Regards,
>
> Carlos
>
>
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:14 EDT