UPDATE SUMMARY: ES40 won't boot past Single User Mode

• Previous message: Praveen Patle: "SUMMARY: All the MSA1000 Volume not showing in the SRM Console"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Date: Mon, 13 Dec 2004 09:29:42 +0000
To: "Tru64 (E-mail)" <tru64-unix-managers@ornl.gov>

All,

By request, here is the procedure that I followed to recreate the hardware
database files.

1. View the current disk layout (print this out for later reference).

 # hwmgr -view dev

2. If the system disk is under LSM control then detach one half of the root
mirror

 # volplex det rootvol-02

3. Drop the system to single user mode

 # init s

4. Move the hardware database files and devices to a backup area

 # mkdir /backup
 # cd /backup
 # mkdir etc disk rdisk tape ntape cport member
 # mv /etc/dec* /etc/dccd* /etc/dcdd* /etc/dfsc* /backup/etc
 # cd /cluster/members/member
 # mv .Booted etc/cfginfo etc/dfsl* /backup/member
 # mv /dev/disk/* /backup/disk
 # mv /dev/rdisk/* /backup/rdisk
 # mv /dev/tape/* /backup/tape
 # mv /dev/ntape/* /backup/ntape
 # mv /dev/cport/* /backup/cport

5. Halt the system and boot to single user mode

 # halt

>>> boot -fl s

6. In single user mount to root file system and compare the new device
   database with the old one

 # mountroot
 # hwmgr -view dev

7. Use dsfmgr to rename any devices that have changed

 # dsfmgr -m dskX dskY

8. Halt and reboot to multiuser (this will automatically reattach the
rootvol-02 plex)

 # halt
>>> boot

Regards

Steve

 ---- ---------------------- -------------------------------------------
>
>All,
>
>Thanks to Howard Arnold and Eric Mortensen for their replies (listed
below).
>Unfortunately none of their suggestions worked. In the end I had to
recreate
>the hardware database files, following a procedure from HP.
>
>Howards' reply:
>If you are booting the install CD try and set the bootdef_dev "" and see if
>this helps. I have ran into things like this before and it was tied to the
>bootdef_dev being set to an old install.
>
>Erics' reply:
>I'm not sure if this will solve your problem, but anytime I've ever had
>problems with the scsi device database and related files, the following
>command has helped to fix the problem:
>
>/sbin/dsfmgr -vVF
>
>Run it a few times to see if it will come back clean.
>
>Then if you're using advfs filesystem you could also run:
>
>/sbin/advfs/verify
>
>On each of the filedomains in /etc/fdmns to double-check that there's no
>additional filesystem corruption.
>
>
>Thanks again,
>
>Steve
>
>
>-----Original Message-----
>
>All,
>
>We have an ES40 which we were about to upgrade to version 5.1B from 5.1A
>pk4. We have got as far as installing the new firmware levels (6.7),
>everything seemed to go ok on this part. But since then we have not been
>able to boot the server past the single user mode. We can boot into single
>user mode, but from there on any commands like "mountroot" or "bcheckrc"
>cause the system to crash.
>
>The only message that we can catch in time before it disappears is:-
>
>process_tempid_list: New hardware ID is not unique.
>
>I am thinking that there is a problem with the device driver databases in
>/etc, especially the files dec_unid_db, dec_scsi_db and dec_hw_db.
>
>Possibly related to this is that when we perform the "hwmgr show scsi"
>command all the device names return as being "null". If I do a "dn_setup
>-install_tape" and then show the devices again, all the names are present.
>However this does not have any affect on stopping the problem when we try
>and boot the system any further.

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

• Previous message: Praveen Patle: "SUMMARY: All the MSA1000 Volume not showing in the SRM Console"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]