When the z/OS NFS server starts up, sockhang is set to off. Set sockhang to on if you want the z/OS NFS server to create a dump if it detects a potential socket hang condition. This optimizes the ability to detect the situation and capture the necessary diagnostic trace data before the trace data is lost. Set sockhang to trap if you want z/OS NFS server to create a dump if it detects a potential ambiguity in the RPC request.

When sockhang is set to on or trap, tracing levels Network, MVS (DFP_request and DFP_return), and Task_Flow (Dispatch, Resume, Schedule, and Suspend) are set on to ensure that the necessary diagnostic data is collected when a socket hang is first detected. When sockhang is subsequently set to off, tracing levels Network, MVS, and Task_Flow are reset to off; any previous settings of Network, MVS, and Task_Flow before sockhang was turned on must be re-set as needed. Detail level tracing provides the ideal setting to further isolate the source of a socket hang condition or an RPC request ambiguity, but with a greater effect on system performance. Enter the operand as follows:

                                .-off--.   
>>-+-MODIFY-+--mvsnsf,sockhang=-+-on---+-----------------------><
   '-F------'                   '-trap-'   

where

sockhang=off

does not create a dump if a socket hang or an RPC request ambiguity is detected. Both sockhang=on and sockhang=trap are turned off.

sockhang=on

creates a dump if a socket hang is detected. sockhang=trap can be set in parallel.

sockhang=trap

creates a dump if an RPC request ambiguity is detected. sockhang=on can be set in parallel.