Why am I getting MOUNT-F-DEVOFFLINE for local disk

[Russ Carraro]

Cluser has 2 AlphaServer 2100's running OpenVMS 7.1. Each system has an RZ28M local disk (DKA). Yesterday I booted the systems and this disk would not mount during the boot process (has for years) saying "%MOUNT-F-DEVOFFLINE, device is not in configuration or not valid". Running the mount disk command procedure from the other system, via SYSMAN/ set env=node, during the boot or from the system after the boot does mount the disk. Any ideas what is causing the problem during the boot?

[Steven Schweda]

Does "this disk" mean the RZ28M on one of the
systems?

May I assume that neither of these RZ28M
disks is a system disk?

If the systems had been powered down, it's
possible that the disk has become old and
tired, and it did not spin up in time to be
ready when its own system tried to mount it.

Now that the second system has mounted it,
can you get the first system to mount it,
too?

If it's getting slow to spin up, it may be
time to start shopping around for a new disk.

[Jim_McKinney]

"Local" to which system? You say "local disk" and that system1 fails to mount the disk but that system2 is successful. That suggests that the disk is local to system2 and that it must be mounted there first in order to be served by system2 to system1? Is this correct? Or is the disk served by some HSZ controller? Can you mount the "devoffline" disk on system1 once system2 has mounted it?

[Russ Carraro]

System1 has local disk1 and system2 has local disk2. Maybe "internal" instead of "local" would be a better word. We use these disks for system pagefiles. Neither is the system disk and neither system has been shut down.

On system1 I run shutdown with auto reboot. I get the DEVOFFLINE message on system1's console. Shortly (within a minute) after the message is displayed and system1 is still booting; on system2 I go into SYSMAN, set env/node=system1 and run the mount command for disk1 that failed during system1's boot and the disk mounts properly.

If I don't use SYSMAN and wait until system1 finishes booting, I can log onto system1 and run the mount command and disk1 mounts properly.

Both disks are now properly mounted on both systems.

Also, all other disks (DUAxxx) are connected via HSD controllers and are mounted before mounting these local/internal disks (DKAxxx).

[Albert Öttl]

Hi Russ,
there is a little example on SYS$EXAMPLES:CLU_MOUNT_DISK.COM
how to mount disks properly via MSCP.
I usually add some additional code to check
whether the disk is already mounted and to add a
maximum retry counter.

(See attachment)

LG
Albert

[Russ Carraro]

Albert, please see the attachment. It's the portion of our mountdisks.com that mounts one of the local disks.

[Albert Öttl]

Hi Russ,
in case of a locally attached SCSI disk I would
guess that the spin-up of the disk takes too long.
This can be caused by a hardware defect or a missing
spin-up jumper.
So if you did not change any jumpers on your disk,
I would guess that the main drive of the disk is
gradually dying.

LG
Albert

[Volker Halle]

Russ,

in your mount procedure, you might want to check the status of the MOUNT command and wait 10 secs and retry the command, if you get a DEVOFFLINE error. Do this in a loop with a maximum no. of attempts set to e.g. 10.

Volker.

[Russ Carraro]

I'll have to wait until I can have the system again (maybe months) to test changes.