-- bind() system call failed: No buffer space available when processing a high load

Cause : occurs when there is no more socket available on the Windows system

1. monitor the netstat results while the load test is being performed
   netstat - an 30 > netstat.out
   it will create a file listing the netstat result every 30 seconds

2. check the file and verify that most of the TIME_WAIT socket are used by b2bipmserver (default port is 8878) and that the limit 65535 is reached :

ex :
...
...
 TCP    127.0.0.1:65527        127.0.0.1:8878         TIME_WAIT
 TCP    127.0.0.1:65528        127.0.0.1:8878         TIME_WAIT
 TCP    127.0.0.1:65530        127.0.0.1:8878         TIME_WAIT
 TCP    127.0.0.1:65531        127.0.0.1:8878         TIME_WAIT
 TCP    127.0.0.1:65532        127.0.0.1:8878         TIME_WAIT
 TCP    127.0.0.1:65533        127.0.0.1:8878         TIME_WAIT
 TCP    127.0.0.1:65534        127.0.0.1:8878         TIME_WAIT
 TCP    127.0.0.1:65535        127.0.0.1:8878         TIME_WAIT

3. now we are sure of the problem : pmserver is taking too many sockets

4. there are 2 possible solutions :

   4.1- Change the Windows registry setting to decrease the TcpTimedWaitDelay time from 256 -> 60 (in second). such parameter defines the time before the port wil be available again. decresing such value (min valu is 30) free the socket quicker. cautious : it affects the whole windows system, be sure that the administrator is aware of it !

   4.2 Change the B2BI_PMAPI_MAX_CONNECTIONS environment variable from 2 -> 9. this environment variable has to be set in environment.dat, and it defines
The value determines how many connections that stays open (is re-used) when a client connects to the PM-server. The default value is 2 on B2Bi 1.4