What you are experiencing is a bug bug in Solaris, The bug is a race condition in RME swim/bugtool perl scripts on the OS. The result is that two successful TCP conversations take place with connection. I had to run the sniffer to figure this out, sniffer showed that the swim/bugtool perl script begins a third conversation before receiving the final ACK of the previous TCP conversation. When the perl script in the swim bugtool initiates the third $ua->request, athird TCP conversation begins. However, a TCP with the SYN bit is sent by the operating system immediately followed by a TCP RST milliseconds later with no packets received in between.
Putting a"sleep(1);" statement right before the $ua->request separates these three TCP/IP conversations enough that the bugtool does not fail. This suggests a race condition within the LWP of Solaris 's IP stack where a new TCP
conversation overwrites a previous unended conversation. Cisco has filed a bug but has not provided a workaround. Cisco bug Id is CSCea83646
You can follow the following workaround:
Add sleep(1);" right before "my $res = $ua->request($req);" at line 138 in BTCCOCom.pm file under swim/bugtool.
