On Sat, Nov 8, 2008 at 3:06 AM, Marcy Cortes
<[EMAIL PROTECTED]> wrote:

> I should have been clearer.
> I want CP to abend - dump - and restart - exactly as if it had some sw
> problem IBM created :)
> Goal : be sure that works after a hyperswap has moved us to alternate
> DASD.

A popular approach with several new installations is a PGT004. Just
drain your page packs and add a few more extra Linux servers ;-)  And
I know from own experience that trying to write CP exits is also a
good way.

But indeed, storing an odd address in a new PSW is very effective (in
one of the prefix pages, that is). I also like the approach to store
something silly in your own VMDCFBUF which makes CP take a FRT016
after completing your STORE HOST command (which Dynam/T VM did once
for us).

Jokes aside, I think it is a very good idea to practice such things if
you have the resources. Personally I think I would stop most service
machines on the system in a clean way first. I know you can't pick the
time for a true unplanned outage, but collateral damage from a test is
often appreciated less by management. Open spool files will be lost,
your accounting data will be incomplete, etc. It's just less pleasant
and if nothing serious breaks this time, it does not prove anything.

Rob

Reply via email to