Re: /proc/loadavg disagrees with top and ps
John Summerfield <debian@ComputerDatasafe.com.au> said on Fri, 06 Aug 2004 13:39:52 +0800:
> Some years ago I used to boot off a Quantum LPS 170. It had some more
> stuff on it, probably /tmp.
>
> It died and managed to hang a couple of process.
>
> I manged to reconfigure the system without taking it down, and the hung
> processes contributed to a higher-than-normail loadaverage for weeks.
> Probably until the next power failure.
I had my web stuff on a friend's computer for a while. They noticed a
high load after a few weeks, and eventually worked out that updatedb
was getting stuck on the same directory every night. It had been like
that for weeks unnoticed (given that it was merely a headless box in
the corner), and was now sitting at a load of 60 or so (mostly
updatedb processes, but a few `ls` in the directory of question).
They sent a notice to everyone that the "computer was experiencing a
bit of trouble, don't do the following:", and then left it until they
could get a new drive :)
--
TimC -- http://astronomy.swin.edu.au/staff/tconnors/
Brown's Theorem (Physics III student, Usyd):
"The only thing that behaves like a billiard
ball, is a billiard ball"
Reply to: