Article 168526 of comp.os.vms:
Last week, I posted a question about why my dual-host DSSI VMScluster wasn't 
working correctly. There turned out to be two separate problems. I'll describe 
them, and their resolution, very briefly.

One of them was DSSI node number conflicts - there were two devices on each 
DSSI bus with the same address. The 4300 was in its original configuration, 
which as I recall was DSSI node 6 for DSSI bus 0, and node number 7 for bus 1.
The 4700 came in with the same configuration, so we swapped the DSSI plugs on 
the front panel before turning everything on. Our mistake was to assume that 
the 4700 would read the DSSI node number plugs on power-up, as the 4300 does. 
It's a little more complicated. The node numbers are stored internally, and 
read from an internal register on power-up. The plugs are read only when you 
enter the following command:
	>>> set dssi_id 0 F	(for bus 0)
	>>> set dssi_id 1 F	(for bus 1)

If you put in a node number 0-7, that is used. Filling in "F" is the signal 
for the CPU to read the plug. 

The second problem was that some of the disks would not appear when the DSSI 
buses were configured the way we wanted them to be. This turned out to be a 
bad DSSI terminator on the bus that was unable to see all of the connected  
drives. Replacing it fixed the problem.

Thank you, particularly Jonathan Boswell, for your suggestions.

-d
*****************************************************************************
    CCCCCCCCC   NNNNN   NNN     FFFFFFFFF   Dennis Costello, CCP
   CCC         NNNNNN  NNN     FFF         Computer Systems Manager
  CCC         NNN  NN NNN     FFFFFFFFF   Cornell Nanofabrication Facility
 CCC         NNN    NNNN     FFF         Knight Laboratory Cornell University
CCCCCCCCC   NNN      NN     FFF         Ithaca, New York 14853-5403

Voice (607)-255-2329 ext 112   FAX (607)-255-8601
On the Web at http://www.cnf.cornell.edu/    email:  costello@cnf.cornell.edu
*****************************************************************************