Any thoughts? Issue with local server

[TOSU]

Ok, this might be randomly posted here, but maybe someone else has ran into this. I created a mumble server within my local network last week for use at a LAN party. I then tested the connection locally and externally from the same client PC and everything worked fine. However starting yesterday this same client PC is no longer able to connect externally, but is able to connect locally. I did some troubleshooting on the network side and don't see any issues, the port isn't being blocked, and tcptrace shows the outside connection hit the CentOS 6 server running mumble.

I then checked the mumble logs and this is what I get when I first try to connect, then let it attempt to reconnect, and then disconnect in the middle of it connecting:

2014-03-22 12:39:45.216 1 => <117-1)> New connection: x.x.x.x:51469
2014-03-22 12:41:17.619 1 => <118-1)> New connection: x.x.x.x:51484
2014-03-22 12:42:09.369 1 => <119-1)> New connection: x.x.x.x:51485
2014-03-22 12:42:38.354 1 => <120-1)> New connection: x.x.x.x:51487
2014-03-22 12:43:07.249 1 => <121-1)> New connection: x.x.x.x:51502
2014-03-22 12:43:36.146 1 => <122-1)> New connection: x.x.x.x:51504
2014-03-22 12:43:40.653 1 => <122-1)> Connection closed: The remote host closed the connection [1]

The client shows this:

[12:39:45 PM] Connecting to server myServer.com.
[12:40:06 PM] Server connection failed: The remote host closed the connection.
[12:41:17 PM] Connecting to server myServer.com.
[12:41:38 PM] Server connection failed: The remote host closed the connection.
[12:42:09 PM] Connecting to server myServer.com.
[12:42:28 PM] Server connection failed: The remote host closed the connection.
[12:42:38 PM] Reconnecting.
[12:42:57 PM] Server connection failed: The remote host closed the connection.
[12:43:07 PM] Reconnecting.
[12:43:26 PM] Server connection failed: The remote host closed the connection.
[12:43:36 PM] Reconnecting.
[12:43:38 PM] Disconnected from server.


So I tried connecting using the local connection and now the client gets this:

[12:47:56 PM] Connecting to server x.x.x.x.
[12:47:56 PM] Connected.
[12:47:56 PM] Server connection rejected: Server is full (max 100 users).

Restarting the server cleared all the ghost connections being created from above.

I was also able to confirm by using the mobile app on my phone that all other outside IP addresses are able to connect (so its not a port issue), it is only when using the internal network to connect using the external IP of the server to connect.

Any suggestions on resolving this issue?

 

[Wooty]

On most routers, when you attempt to use your external IP address to connect to a server on your own LAN, the router will do something called WAN NAT Redirection, in which rather than sending the address out the the internet (where it will prompty be sent back again), it simply redirects it to where it's supposed to go internally. Some servers don't expect this to happen, and it can cause timeouts and such.

I'd recommend:
1) Just don't use your WAN ip to connect to a mumble server on your lan
2) Decrease the timeout parameter in your murmur.ini to something like 20 or 30 (this should cause the fake clients to be dropped automatically rather quickly)

 

[TOSU]

Thanks for such a quick reply!

I see now in my router that filtering NAT redirect was not enabled, and that makes sense why the tcptrace was showing as if it was coming from the gateway instead of an outside IP. I really should have paid more attention in Cisco class lol.

I am doing as you suggested, won't be connecting from within the local network using the WAN, and I turned on the filtering so I won't be tempted to try that again. Also turned on the timeout setting to 30. This also resolved my other issues with my spiceworks and ViconNet install

Thanks for your help!