get_local_socket_name() does not always work #35
Comments
|
The problem is the inconsistency between IP address being used and the IP address the far side is told to call back. This may have existed before we switched to get_local_socket_name(), and the problem may be that the server is connecting to the IP it heard from rather than the one requested. Before, since the host was listening on all IPs, this would have worked (but perhaps should not have). |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Isop W. Alexander sent a report showing that a machine with two ports (172.28.0.10 and 192.168.191.130) sent a connection request that came from the 172 IP address but that had the name of the 192 interface in the message as the host to call back. This turned out to be unroutable, so the server failed to call back on the correct IP address. Presumably, this happens when the gateway is configured to be a single outgoing NIC. This was on a Linux box. We need a more reliable way to select the outgoing NIC.
This function was added to keep Windows from complaining about trying to make network connections for the case where we're using loopback. It does work for that case.
The text was updated successfully, but these errors were encountered: