[H-GEN] Kernel Panic! :)

[Robert Brockway]

On Tue, 20 Jul 2004, James Mills wrote:

> *nods* The box is mission-critical. Some kind person has offered me one

I think Nikolai made an excellent point.  If a box is mission-critical it
needs to be possible to take it down at a moments notice because this can
happen whether you want it to or not.  This initially seems
counter-intuitive I suppose.

Given that all your boxes are identical in h/w you are in a very good
position (loaner/donated boxes aside).  If any of the other boxes are less
critical perhaps the functions can be swapped.

For my personal and business boxes I follow a devel/production schedule.
I have boxes I test on before putting stuff in production.  If a
production box dies these devel boxes can operate in production (after
getting restored from the last production backup).

It sounds like your setup is pretty good so it would be worth having one
extra box around to do devel on and to use as a production backup.

I backup my dev boxes for 2 reasons:

1.  In case of h/w failure I do not want to lose stuff I am working in.

2.  This reduces the amount of time required to drop them into production.
    An incremental backup is all that us needed before you can blow the
    data away and replace it with production stuff.

> only crashed twice since Winter. I purchased it sometime last Summer.

At least it is not crashing every night/morning.  Do you have a serial
console on the box?.  Useful info is often dumped there and it is easier
to copy it than from a vga console (assuming you have the serial console
connected to another box via a null-modem).

ARob

-- 
Robert Brockway B.Sc. email: robert at timetraveller.org, rbrockway at uqconnect.net
Linux counter project ID #16440 (http://counter.li.org)
"The earth is but one country and mankind its citizens" -Baha'u'llah