To attempt to recover disks that have gone into the state Mount_Verify_Timeout

Log into HEPOPERATOR

If the disk(s) involved are NOT ONLY DU01, do the following:
(If multiple disks are involved, the specific DU01 procedure may still be
necessary.)

1. At the $ type SYSMAN

2. Hit Keypad 0 (KP0) to select cluster
	type DO SHO DEV DUnn   (where nn is the disk in question)
	If all nodes show MountVerifyTimeout, then
	type DO DISM/ABORT DUnn
	If only some nodes are timed out, 
		hit KPn where n is the node number, then type
		the DO DISM/ABORT DUnn, 
		repeat for all timed out nodes.
	Check other disks as needed, from the cluster (KP0).

3. If any node gives the message about open files,
	select that node with KPn, and type
	DO SHOW DEV DUnn/FILES
	Note any processes with open files and kill that process
	with DO STOP/ID=pid   
		(do not kill process 00000000, these are system-open files)
	Then repeat the DO DISM/ABORT DUnn
	Repeat on all nodes with open files.

4. exit SYSMAN with ^Z

5. Get the disk(s) back by cycling power, reconnecting cables, or whatever.

6. Remount the disks with $ @ND:MOUNTDISK on one node only.

Steps 1-4 must be done whether you can get the disk back or not.  
You cannot get the disk back on a node with open files on that disk.
You cannot save a job with the open files; it may appear running but it's not.

If the paging disk is timed out, you must reboot that node, but you must
first clear the timeout on all other nodes with the above procedure.

Paging disks:
	UNDHE4:	DU43
	UNDHE5:	DU50	(or possibly DU51, if it is still there)
	UNDHD0:	DU00
	All others are their only disk.

If the only disk is DU01, and it is timed out on nodes other than UNDHD0,
it is probably because UNDHD0 went down and the other nodes did not.  In this
case, what is hanging the disk is D0's VAXNEWS.  To recover from this, do the
following on each node with DU01 in Mount_verification (can be done with
SYSMAN by saying DO command, or by SET HOST to the node involved):

1.  Do @ND:UNDOVAXNEWS
	Error messages will be displayed, but that is OK.

2.  DISMOUNT DU01
	Check that it worked with SHOW DEV DK.

3.  After doing 1 & 2 on all involved nodes, do
	@ND:MOUNTDISK on one node only (any node is OK, including UNDHD0).

4.  On each affected node, do
	@ND:STARTD0NEWS
						JMB, 1/24/95