Hi, I am having a problem with the shutdown sequence on a 2 node cluster. The first node of the cluster shuts down fine. However, when I attempt to shutdown node2, I get the following error:

Unmounting NFS filesystems: Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Unable to receive
Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Unable to receive
umount2: Device or resource busy
umount: /mnt/log: device is busy
Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Unable to receive
Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Unable to receive
Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Unable to receive
Cannot MOUNTPROG RPC: RPC: Port mapper failure - RPC: Unable to receive
[FAILED]

System hangs here on shutdown. Both systems are set up the same way and clustering is always stopped first on both nodes. This is on a redhat enterprise 2.1 server using 2.4.20 kernel and nfs-utils-0.3.3-4

I think I have narrowed this down to where I believe that the netfs script is where the system is hanging. I see that the netfs script uses umount -f -a -t on the nfs mounted file systems and also runs fuser -k to free up any filesystems being used, but I can not figure out where this message is coming from or how to get around it and get the system to shutdown cleanly.

Any help appreciated.

you should unmount your nfs shares before doing this

but i am not sure if it will work if there are locked files - my be script something to check those cases and delay the shutdown untill everything is fine - or kill all that uses nfs mounts from the cluster that's going down