Hi

For the 3rd time in 4 days my linode has crashed. This is getting rather annoying. And I know this has nothing to do with the /dev/kmem issue because I've had uptime of over two weeks before. (Little has changed on my linode since then)

Can there be a feature to automagically restart the linode in the event of a crash ?

Thanks

Sunny Dubey

Does anything unusual appear in your logs? Look at /var/log/messages and /var/log/syslog

Nothing, I just spent time going over them.

I'd love to provide a crash report to the UML community (unless we can figure it out on our own).

It doesn't sound to me like you can reproduce the crash, but if you can, try having a session open via the console (ssh user@hostX.linode.com) and see your kernel gives an oops/panic.

Otherwise, the "watchdog" feature is something that is in the queue.

-Chris

I don't know if it will help you now, but I add the following to my
/etc/syslog.conf:

kern.* /var/log/kernel
*.alert /var/log/alert


So all kernel messages get nicely logged for me,
and I don't have to stay on console 24/7


-Ashen

I'm looking into logging solutions (other then syslog) as my linode
has now, several times, simply become completely unresponsive
to any attempts to contact it, even though the linode.com page
says it's running. Grrr, I hate mystery problems.

-Ashen

Sounds like your host has a heavy I/O load. This should be fixed in the somewhat near future with simply better kernels.

Sunny Dubey