Hah - our database server is currently handling 150+ queries per second, fairly happily on 1% CPU usage!
I'm trying to do some hard disk maintenance on a machine on the other side of the world, and before I start doing things like reconfiguring its RAID arrays with things still mounted there, I'm giving NFS booting a go. The idea is to configure the system to boot using its usual GRUB boot partition, but then mount its root filesystem off an NFS server. This will let me unmount /boot and do all the hard disk maintenance, and if the worst comes to the worst, I can ask our man in the datacenter to bring the machine back to life with a Debian rescue disk.
I'd installed nfs-user-server on a Debian install on VMware, and configured a second Debian install to mount an NFS export from the server as its root partition. However, it would always fail during the boot with this error:
rpc failed: 2
At first I thought there was a module missing during boot or something, but I eventually figured out that the nfsmount tool used during boot was /usr/lib/klibc/nfsmount, and that it would give the same error if run after a normal boot:
/usr/lib/klibc/nfsmount 192.168.2.64:/nfsroot /mnt
The fix? Install nfs-kernel-server rather than nfs-user-server.
It now gets past that point OK. (It's still failing after 'Setting up networking....' -- I suspect that it's resetting eth0 and losing its IP address, or killing the NFS session.)