http://bugzilla.spamassassin.org/show_bug.cgi?id=4260
------- Additional Comments From [EMAIL PROTECTED] 2005-05-09 00:17 ------- Subject: Re: rewrite DNS code to use a single socket, event-based model > I modified connect_sock to use a random port between 1024 and 65535 each time it > creates a socket. The currently checked in code creates a new socket with each > message. The result of the one test run I did was two bogus rr messages between > 40% and 50%, and no others for the rest of the run. > > I'm going to try to think of how there could have been a collision that showed > up so soon and then no others. The answer may be to give up on trying to avoid Sidney, don't forget that "random" in no way insures that you won't use the same port number twice in a row, or even 100 times in a row. (Although the later is bloody unlikely.) Or use the same port number 3 requests apart. If this is indeed the necesary solution, then you really should add some history code to insure that you won't use the same socket number used in the last 5 minutes or so. Or include an id with the messages sent by the socket, effectively increasing the number with a sequential key. (And no, using random there would not necessarily be 100% safe either.) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
