This seems to be misguided and would only increase the chances of a DoS-type attack on TCP session resources.

A second 3-second delay upon malloc failure is also removed.