Good morning,
[reply delayed, I sent from wrong address and post was rejected]
On 13/1/03 at 1:09 AM, nate <[EMAIL PROTECTED]> wrote:
>Charlie Garrison said:
>
>> Any feedback or suggestions would be greatly appreciated. I have been
>> trying to solve this for about 4 months now, and I don't know what else to
>> try.
>
>setup a 2nd system, connect a null modem cable between the 2. setup serial
>console on the system that appears to be crashing. on the 2nd system setup
>something like minicom to log to a file. load minicom in something like
>screen to load it in the background and have it log.
I had thought a 'remote terminal' type solution would help but wasn't sure how
to go about it. And I'm still not sure how to do it in this case as I would have
to buy more rack space to install another system. But I'll ask the data
warehouse if they have a system I can temporarily log to (per your suggestion).
>is the system on a UPS? if so is there software installed to monitor the
>UPS?
Multi-level UPS, including 3 backup diesel generators... I could keep going; the
building contains a *very* professional data warehouse.
>is the location of this server secure? that is someone isn't tripping over
>the power plug, or pulling it out on purpose, or doing something else to
>it?
More than 10 security points between the street and the rack. No one
unauthorized is getting in. And I want to think it's safe to assume that
authorized people won't be tripping the plug.
>is the system under heavy load normally? what is the typical load average
>of the machine(15 minute average). if it's consistantly staying above say
>3.00, then I would consider it pretty heavily loaded.(most of my systems
>are 0.10 or less, some are even 0.04 or less).
It's still a fairly new server and load is normally 0.10 or less. It
occasionally peaks at 5-6 when doing nightly cron jobs. The 'crash' only happens
at apparently normal load levels.
>serial console is the best bet though, if it is crashing such as kernel
>panic or something you'll get the full kernel dump on the serial port,
>logged to a file for later enjoyment.
I do like that idea and will hopefully find a solution with the cooperation of
the data warehouse.
>I also prefer to log to a syslog server but linux isn't very robust when
>it comes to crashing and if it crashes it doesn't tend to log to
>the syslog server(solaris does, at least in the crashes I have experienced).
I will go ahead and do this anyway. But can you suggest what I should log to
another machine. Note, I will be logging to a system on a remote network, so
don't want to over do it.
I would guess that I need the kern facility, any others? And what priority
should I use; warning, err, crit? Eg:
kern.crit @hostname
>Theres tons of docs out there on how to configure a console serial port.
I'll have a look.
Thanks for the information and suggestions.
Charlie
--
Charlie Garrison [EMAIL PROTECTED]
--
redhat-list mailing list
unsubscribe mailto:[EMAIL PROTECTED]?subject=unsubscribe
https://listman.redhat.com/mailman/listinfo/redhat-list
