http://bugzilla.spamassassin.org/show_bug.cgi?id=4260
------- Additional Comments From [EMAIL PROTECTED] 2005-04-26 15:49 ------- John Gardiner Myers wrote in sa-dev: > If a new socket is created for each message, you might find the port > wrapping, leading to collisions. I worked out the math for that. Assuming on the order of 50 DNS queries per message (using the test message in t/dnsbl.t as an example) right now what happens is that the 16 bit ID wraps after about 1300 messages. The real problem is that as long as the process is alive it is listening on the same port. All the old replies get read and processed. If there is an ID collision, it has an effect. With each message having a new socket, the port number will wrap 50 times less often than the IDs are now wrapping. If an old reply comes in it is not likely that there is a listener active on the same port, since the lifetime of a listener is one message. With each message having only on the order of 50 DNS queries, the probability that two messages that have the same port number also having their 16 bit IDs in the same range is tiny. A port collision has no bad effect unless the IDs also collide. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
