Differences in tcpAttemptFails between 2.0.1-Rel and 2.1-Rel
I've been using 2.01-Rel for some time to load balance a busy web service. We have a large number of very short http sessions from all over the world. This setup has been working fine but we are out-growing what a single instance of pfSense can do. On our web hosts we've typically seen the snmp tcpAttemptFails (see rfc1213) counter increase by about 2 per second. I've always assumed that we really have a couple of conn/sec that are really bailing out.
rfc1213 says that tcpAttemptFails are:
"The number of times TCP connections have made a
direct transition to the CLOSED state from either
the SYN-SENT state or the SYN-RCVD state, plus the
number of times TCP connections have made a direct
transition to the LISTEN state from the SYN-RCVD
Fair enough, makes sense that I might see some of this in my application. I recently added an additional instance of 2.1-Rel, I set it up in as an identical manner as possible given some differences in configuration options. This instance essentially works as expected with the exception that we see ten times or more the number of snmp tcpAttemptFails.
I cannot explain why I'd see more issues with 2.1. Any ideas are appreciated.
After more testing this has nothing to do with pfSense. It has something to do with having 2 IPs in DNS.