UNDHEP Dump Procedures
							JMB, 10/9/90

 If our cluster hangs again as it did Oct 8, please crash it with a
   dump, as described on the last pages of the UNDHEP Console Manual.
   If all nodes are hung, crash-dump UNDHE5, otherwise only the node 
   that is hung.  DO NOT crash-dump more than one node, since there is
   only one common dump file & the last will overwrite previous dumps.

 If all nodes will not let you log in, you must get HE5 going before there
   is any point in trying to reboot the other nodes.  

 Halt the system with the halt button on the system box.  On HE2 & HE4,
   the button latches in, but does not halt the system until it is
   pressed again to pop it out.  Then enter the sequence of commands 
   to force the dump at the workstation monitor, NOT at the terminals 
   in 402.  The E (examine) lines are optional.  Write down the PC(s)
   displayed when the console acknowledges the halt (HE5 will have 2).

 Save the dump (from HEPoperator, after the system is back up) by:
	$ copy HE5EXE:sysdump.dmp DU51:[SCRATCH]
   Then rename the copy in DU51:[SCRATCH] to indicate the system dumped
   and the date of the dump.  These dumps are 32k blocks long, so the
   copy will take a while.

 If you crash HE5 and it hangs coming up, Halt HE5, disconnect the ethernet
   tee from the front of HE5, then boot it.  There is no need to attempt
   to take a second dump.  Be sure HE5 is all the way up and you have
   logged in and copied the original dump before reconnecting the ethernet,
   because the rest of the nodes will immediate crash and dump, overwriting
   the primary dump.

 DECnet must be successfully started on HE5 before any other node can start.
   If HE5 has been rebooted with the ethernet disconnected, DECnet will
   not start correctly.  You must shut down and restart DECnet before 
   any satellites will boot.  Be sure the dump is copied before doing this.
   Do this (from HEPoperator) by:
	$ set def sys$manager
	$ @shutdecnet
	$ @startdecnet