Hi

We have a number of Dell servers running RHEL 3 and 4 at our site and have been experiencing a problem on the odd occasion.

We are running Oracle Databases using oracle RAC and have seen a problem where, if one of the servers in the RAC kernel panics, the server is leaving the network interface alive. The problem being that, with the interface alive and pingable, Oracle RAC believes there is no problem and continues operating unchanged.
We have seen this problem on a bunch of different servers over the past couple of years and have now resorted to setting the sysctl parameter to reboot the server after a panic.

Has anyone see this behaviour after a kernel panic?
Does anyone know if there is a way to stop it without making the server reboot itself?

Thanks in advance.

Nick

I've seen this on occasion with servers, and I don't there's any better way to do it. However, you might want to look at setting up netdump to send the panic information to a central server so you can debug what's causing the problem. Netdump will reboot the machine after it finishes sending the crash dump to the server.

Yeah we have done that and it looks like the RAID card caused the last one. Firmware update required.
Thanks for the response, I was beginning to feel we are the only ones seeing this behaviuor.

Nick