Hello,
AFAICR, LM 7.2 comes with a 2.2 kernel, which CAN be rendered unusable by a
user program unless you set some proper ulimits: if a user process just
grabs memory ( and works on it ), the machine begins to swap - and when it
runs out of a swap space, you in fact can do nothing but power cycle the
box. I've written a perl script once, which claimed tons of memory ( by
working on hashes with the db routines ) - and after working a while, the
kernel told me about running out of memory and began to kill processes. I do
not remember anymore what messages appeared in the syslog, but you will know
when you see them. Try to redirect *.* to /dev/tty12 ( or whatever unused
console ) and switch to that console when your machine spins down...
CU Dmitri Barski
--
Unix System Support
Nokia Networks
_______________________________________________________
E-Mail : [EMAIL PROTECTED]
Telephone : 3253 (NTG internal)
: 828 3253 (Nokia internal)
: +49 211 9412-3253 (external)
Address : Duesseldorf, Heltorferstrasse 21
: America, Room 350
_______________________________________________________
> -----Original Message-----
> From: ext D. R. Evans [mailto:[EMAIL PROTECTED]]
> Sent: 05 Juli, 2001 2:41
> To: [EMAIL PROTECTED]
> Subject: [expert] disk activity renders machine unusable,
> forces manual
> reboot
>
>
> Running LM 7.2 on a stock 700 MHz Athlon.
>
> Everything was working fine until sometime in the past day or two
> something got changed such that I now see the following behaviour:
>
> Sometimes when the clock ticks across an hour boundary, the
> disk light
> comes on and stays on. The machine is so busy that the only
> thing I can
> do is to power it down.
>
> This happened yesterday at 2300Z and today at 2400Z.
>
> Is there anything I can do that will let me log what process might be
> causing this, so that after the enforced manual reboot I can
> figure out
> which process is the culprit?
>
> I have made a lot of changes to user programs in the past day or so,
> although I would like to think that Linux would ensure that if the
> problem was a regular user program, I would still be able to
> get enough
> time to log in and run a ps to look to see what's happening.
>
> I also added ntpd (the 4.0 latest release from ntp.org), which seems
> very suspicious to me -- but it's pretty hard to believe that
> anything
> problem this major could be lurking in such well-tested code;
> especially since I'm not doing anything fancy in the config file.
>
> Doc Evans
>
> --------------------------------------------------------------
> Phone: +1 303 494 0394
> Mobile: +1 720 839 8462
> Fax: +1 781 240 0527
> --------------------------------------------------------------
>
