[OpenAFS] heartbeat and OpenAFS

Brent A Nelson brent@phys.ufl.edu
Thu, 13 Mar 2003 12:09:38 -0500 (EST)

Has anyone already created a failover OpenAFS file server or have any tips 
to share on how to do it?

I have two OpenAFS servers with /vicepa kept in sync between the two
machines with DRBD.  I'd like to have heartbeat failover a single IP
address between the two machines and manage the starting/stopping of fs,
but the volume database seems to be too "smart" for that; when the address
fails over, the volume database rejects the update from the newly started
file server processes on the new machine, as the failover IP address was
already registered with the first machine.

Anyone have any ideas on the best way to "trick it"? I believe it would
work without the common IP address between the two machines, but then the
clients would figure out the change in file servers rather slowly without
executing "fs checkvolumes" everywhere, right?

Thanks in advance,

Brent Nelson
Director of Computing
Dept. of Physics
University of Florida