Re: Synopsis of a database crash and recovery (or time to bash

Colin.Shaw_at_phs.com wrote:

> Not to start a(nother) NT bashing here, but isn't process clean-up the job
> of a "real" preemptive multi-tasking OS? (i.e. we ain't talking W95 here,
> people.)

*grin* You've never had shared memory segments hang around when a Unix process has quit? Ever had a TCP/IP port stay bound even tho' the application has died? Tried killing a process while it's in the middle of an tape I/O call? Seen a process table get so full of zombies that you can't fork another?

All operating systems suck (except maybe VMS). It even says words to that effect in the Oracle administrator's manual, in the section on resource limits :0)

My most spectacular OS crash happened when I (accidentally) set the scheduling slice so low that the CPU couldn't carry out a context switch before the next interrupt. At least a modern Unix won't let you take all the CPUs offline at the same time...

g

-- 
guy ruth hammond <grh_at_agency.com> | One is punished for being
Technology Analysis & Consulting  | weak, not for being cruel.
07879607148 http://www.agency.com |       -- Baudelaire