[Info-vax] recognizing SCSI disks

[Phillip Helbig undress to reply]

In article <qn5qpl$1ckv$1 at gioia.aioe.org>,
helbig at asclothestro.multivax.de (Phillip Helbig (undress to reply))
writes: 

> > Well, cluster is back with the original hardware.  :-)

Except for the BA356 on one of the nodes.

> I've never had a an expansion box itself fail, as opposed to the disks 
> in it.
> 
> As the model was identical, I though that I could just connect it up and 
> boot, but the console couldn't see the disks.  I power-cycled the 
> machine (a PWS), and then everything is OK.
> 
> Why was it necessary to power-cycle the machine so that it could see the 
> disks?  I remember once when I pulled a SCSI cable by accident; the 
> disks (including the system disk )went into mount verification, then 
> came back when I plugged the cable back in.  That seems much more 
> dangerous than unplugging the cable at the console prompt.  Of course, 
> it is a different box (but same connector at the top of the box, the 
> removable one where the cable plugs in), but surely the SCSI controller 
> doesn't need the ID of the box (assuming that there is one).

Another puzzle: before shutting down the node to swap the disks, from 
another machine I did DISMOUNT/POLICY=MINICOPY on the member of a 
three-member shadow set connected to that node.  Now the node is up and 
all shadow sets are mounted, and I'm getting a full copy.  Why is that?

When I thought that one of the disks was bad and removed them one by one 
to check, I did DISMOUNT/POLICY=MINICOPY and the subsequent mount always 
resulted in a minicopy, not a copy.