[lug] strange kernel happenings...

[Nate Duehr]

Anyone have any ideas on this...?

Seeing some really "Bad Things(TM)" on one of my boxes.  Started after
an upgrade to 2.4.18 from Debian.  Have tried both 2.4.18-bf24 and
2.4.18-k6 flavors...

Basically, what's happening is suddenly and without warning (no logs, no
idea that it even is occurring) the kernel stops processing SIGTERM,
SIGKILL, etc.  Also, programs like top and ps (anything that views a
process list) lock up the terminal session (both after displaying a
partial list).

I'm actually typing this from that machine -- the machine stays running,
daemons act semi-normally, and in general everything works except
anything that processes signals to the kernel.

Yesterday for the first time I noticed that keventd had gone defunct
during this, but usually that doesn't happen.

What are some of the things I could check on such a goofy system to help
find and correct the problem permanently or help someone much smarter than
I know what might be causing this?
As I said, no logs from syslog show anything useful... not even an
interrupt conflict or IDE hiccup... nothing... it just starts acting
strangely.
(One of the interesting side-effects of this bug/situation is that the
system won't process signals, so "shutdown" and "init" don't work...
can't even reboot it without hitting The Big Red Switch.

Oh and even stranger... I waited about eight hours and logged into the box
again and a "reboot" command INSTANTLY dropped it... no warning message,
and earlier today it wasn't responding at all to any kind of signals...
wow.
-- 
Nate Duehr, nate at natetech.com


Random Quote:
--------------
It is error alone which needs the support of government.  Truth can stand by itself. - Thomas Jefferson, Notes on Virginia