[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: How to analyze a server crash?

On Thu, Jan 16, 2003 at 09:57:47AM +0100, martin f krafft wrote:
> So my question is: how can I figure out what happened, and why is
> there absolutely no information between my last successful automatic
> fetchmail log entries and the sysklogd restart message? Under what
> circumstances can Linux crash such that it can't even write logs
> anymore and doesn't respond to the console?

Unfortunately, there probably isn't any way to really figure out what
happened short of trying to reproduce it. For example, a hard-lock
caused by a hardware problem or a driver bug might not leave you with
any messages because the machine locks before the kernel can properly
panic. I've had that happen when X crashed, too, and that wasn't really
even a driver bug (it was combination of wine and font servers). Linux
also doesn't have a standard crash dump facility, and even if it did PC
hardware doesn't give you a way to force a dump like other UNIX server
platforms do.

The best suggestion I can think of right now is to read about the "magic
sysreq" kernel debugging option, enable it, and try to rsync another
10GB. But if the system is truly locked it won't be able to respond to
your sysreq.

Michael Heironimus

Reply to: