[Info-vax] Volume Shadowing & Cluster Satellite

[Bart Zorn]

On Monday, April 23, 2012 4:03:39 PM UTC+2, Paul Sture wrote:
> On Mon, 23 Apr 2012 13:42:58 +0000, Michael Moroney wrote:
> 
> > gartmann at nonsense.immunbio.mpg.de (Christoph Gartmann) writes:
> > 
> >>Hello,
> > 
> >>today I tried to fiddle around with volume shadowing on OpenVMS 7.3-2.
> >>So far everything worked as expteced. The boot node's device changed
> >>from DKA0: to DSA0:. But the satellites were unable to boot from the
> >>"new" cluster system disk. I changed all references in
> >>SYSTARTUP_VMS.COM, SYLOGICALS.COM and in LANCP. Satellites did boot,
> >>entered the cluster but immediately afterwards before the beginning of
> >>STARTUP.COM they reported:
> >>    boot device is already mounted as a shadow set member
> > 
> >>Thus, where else do I have to change the device name?
> > 
> > Did you change the shadowing sysgen parameters (such as SHADOWING,
> > SHADOW_SYS_DISK and SHADOW_SYS_UNIT) ?
> 
> And ensure that ALLOCLASS is set to an appropriate non-zero value.
> 
> -- 
> Paul Sture

There is no need for this. The ALLOCLASS (or port allocation class) of the boot node determines the device names. The satellite nodes just use these names, regardless of their own ALLOCLASS. Only disks local to the satellite get a name based on the satellite's ALLOCLASS. If ALLOCLASS is zero, the satellite's disks get a device name like <nodename>$DKAn, which is guaranteed to be unique throughout the cluster. 

Bart