[Info-vax] Not getting a dump file when VMS crashes

[Bob Koehler]

In article <4f7cb8f2$0$5610$c3e8da3$eb767761 at news.astraweb.com>, JF Mezei <jfmezei.spamnot at vaxination.ca> writes:
> 
> My all mighty Microvax II had the time to spot memory parity errors and
> write them to the error log (on a slow RD54 nonetheless!) before the
> system failed.

   My second 11/780 was installed in a facility with a diesel powered
   backup generator.  For certain critical tests, switching power was
   tested.

   Folks in the facility were concerned about their then ancient
   collection of IBM 360 mainframes that were running the critical code
   and would prepare them for the switch.

   Nobody told us.

   The 11/780 would crash upon entering power fail recovery code in the
   kernel.  To actually get power fail recovery to work, you had to
   have battery backup power for the RAM, and we did not.  The 11/780 
   saw such a quick power switch that it did not enter a cold start, but 
   it was too long for RAM and something would get corrupted enough to 
   cause the power fail recovery code to fail, and fire off the bugcheck.
   By that time power was stable and everything would proceed to
   complete the dump and automatically reboot.

   SDA would point me to the actual bugcheck code and the power fail
   recovery routines would always be on the stack.

   We started with 1MB RAM and upgraded to 2.  Only full memory dumps
   existed, and they didn't take very long.