Signaling sympathy sickness can occur when a stalled member on the target system consumes too many I/O buffers and causes the sending system to run out of outbound I/O buffers. The MEMSTALLTIME parameter in the SFM policy allows SFM to take automatic action to alleviate signaling sympathy sickness caused by a stalled XCF group member. This is especially useful in cases where the stall condition is so severe that both manual intervention and system automation are impeded. If you specify MEMSTALLTIME(seconds), XCF will terminate the stalled member that is consuming the greatest number of applicable I/O buffers after sympathy sickness is declared and persists for MEMSTALLTIME seconds. If the sympathy sickness condition continues to persist, XCF will continue to terminate stalled members in this fashion until the problem is resolved.

If a stalled member is to be terminated to resolve a signaling sympathy sickness problem, XCF terminates the associated task or space that the member identified at join time. Such termination of the task or space has the same effect on the member as would the demise of that task or space through any other failure, and therefore suitable cleanup and/or recovery should be expected. There is some risk that the exploiter might not deal well with such termination. However, failure to resolve the signaling sympathy sickness problem in a timely manner may well lead to the demise of one or more systems.