Messages in this thread | | | Date | Fri, 6 Aug 1999 07:25:18 -0400 | From | Dan Hopper <> | Subject | Re: Linux 2.2.x outgoing TCP connects hang after several hours of use |
| |
Meelis Roos <mroos@tartu.cyber.ee> remarked: > DH> I've been noticing a TCP connect problem on a masquerading gateway > DH> box for a few months now, first running 2.2.7, and more recently > DH> 2.2.10 and 2.2.11pre2. They behavior is basically that outgoing TCP > DH> connections originating on the gateway box itself fail, but that > DH> incoming TCP connects succeed. Outgoing TCP connects that are > DH> masqueraded via this box succeed as well. The interfaces are eth0 > DH> for the local private net and ppp0 for everything else. > > If you are using some reverse masquerading techinques (like ipautofw) > then it may happen that the kernel chooses the local port number for > outgoing connections from the range that the masquerade helper uses. > It results in just the same symptoms - no outgoing connections, incoming > and masquerade still work. When a big-enough number of connections fail > then the local port number gets out of the forwarded range and it works > again.
You're absolutely correct. I thought I had check all my firewall stuff thoroughly, but I overlooked the autofw rules I had added a couple months back to handle Jedi Knight. As a hack to get it working correctly, I had removed the control port, so that all TCP ports in the range 2300-2400 were affected without qualification. I commented these back out (as they should have been all the time) and things are back to normal.
Thanks! Dan
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |