That's helpful. For the software watchdog idea, that sounds
interesting, but I wonder (and will have to research :-) if it will work
with the particular problem I'm having. I've got one server that has
the annoying habbit of running out of RAM and dying. The server isn't
crashed, and the hardware isn't locked, and in fact you can continue to
ping it and it will respond nicely. It just doesn't have any more RAM
or Swap and can't even free-up enough ram to run the "shutdown" or
"init" commands. While this is obviously an issue in itself, and I'll
need to figure out why this is happening, it would be nice if the server
could just reboot itself when it becomes unresponsive.
Maybe a good shell script is the answer in this case ;-)
Nonetheless, in my experience running Linux for the past 6 years, I have
indeed found that it (gasp!) crashes and locks-up, maybe a couple times
per year for a well used system (this is just my experience).
Installing watchdog cards in the servers would save us and our customers
some time.
-Fred
"Mikkel L. Ellertson" wrote:
>
> It depends on what you need the watchgod for. If you just need it to
> recover from hardware/software hangs, then the internal card works
> fine. If you are only worried about software hangs, then the software
> watchdog timer that can be compiled into the kernal may do the job. If
> you need a hardware watchdog, you can look at the addresses in the
> watchdog.txt file included with the kernel docs. I did a search on the
> internet and came up with http://www.outsrc-em.com/watchdogs.html. I
> havn't tried them, but from the sire, it doesn't look like it would be too
> hard to make them work with Linux.
>
> Mikkel
--
To unsubscribe: mail [EMAIL PROTECTED] with "unsubscribe"
as the Subject.
