On Wed, 15 Mar 2000, Paul Jakma wrote:
> On Tue, 14 Mar 2000, Alex Belits wrote:
>
> > Killing Oracle, or any other server that depends on some process being
> > alive and keeping a valuable, complex, hard to recover data on disk and
> > in memory, is in some cases not any better than just blowing up the box.
> >
>
> definitely. In such a case it'd be much better to freeze things, and
> have a freepage.root watermark so that root can login and clean up.
Which is fine - unless it's a root process which has exploded.
In which case, start praying you have SysRq enabled - you're going to need
it.
> Yes.. i've heard the "do you want to sit by your server" toss. but not
> all machines are webservers where processes are throwaway httpd's. Think
> credit card transaction processing.
Your CC processor will only be killed if it has blown up in a serious
enough way not to recover unaided. In this case, I would FAR rather have a
dead CC processor than one which APPEARS to be working, but isn't. ("Sorry
about that, all the transactions for the last 24 hours went to the
bitbucket - the CC processor had a memory leak, so it stayed there
accepting transactions but not processing them...")
Being killed unexpectedly is something mission critical systems like this
would be designed to handle. That's why you have transactions,
journalling, checkpoints etc. - and failover/load-balancing systems. If my
server process is KILLED, traffic will go elsewhere. If it is just BROKEN,
it may well accept and discard traffic - totally unacceptable.
James.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Mar 23 2000 - 21:00:19 EST