NFS v4.1 infinite loop with READs or WRITEs returning NFS4ERR_BADSESSION (10052) result in system hangs

Issue

• We have an RHEL 6.4 client talking NFS v4.1 to a Network Appliance fileserver. We've been testing some failover scenarios where one of the fileserver nodes is failed out and another takes over. Sometimes, this seems to cause the Linux clients to get into a state they can never recover from. We're running lmdd to generate traffic to the fileserver, and the lmdd processes become unkillable, and the "df" command hangs on the fileserver mount.
• NFS v4.1 hang
• NFS v4.1 infinite loop with READs or WRITEs returning NFS4ERR_BADSESSION (10052)