Cascade Max TCP Connections

[drodriguez]

Hi everyone,
First of all,
I have set up a Layer-2 VPN just like this one:

So far everything is working as expected, but I noticed that the maximum TCP connections for the Cascade Connection is 2. I tried to increase it both from vpncmd using CascadeDetial and from the Windows Server Manager GUI. But after reconnecting the available -and used- TCP connections are still only 2. From the Windows GUI I can access "Advanced Setting of Communication" for the Cascade Connection and verify that Number of TCP Connections is 8, altough the setting is not currently having effect.
Is this the expected behaviour? Is it possible to increase the number of TCP connections in other way?

PS. I'm using Ver 4.29, Build 9680, rtm

Tanks in advance,
David

[drodriguez]

Today I built both server and bridge from source from the non stable repo, and the behaviour is the same, so I created an issue on GitHub.

[centeredki69]

How are you verifying the number of TCP connections?

TCP connections.jpg

[drodriguez]

Hi, thanks for the reply.

tpc connections 2.jpg

tpc connections 3.jpg

I can also make the same check with CascadeStatus from vpncmd.

[centeredki69]

I have been trying to duplicate your issue but can't seem to.
Try Un-checking the life time limit and try toggling the number of TCP connection using different numbers and hit ok after each choice. Verify that NO security policy has been set on the server side for the user restricting the amount of connections.

Security policy.jpg

[drodriguez]

I have been trying to duplicate your issue but can't seem to.
Try Un-checking the life time limit and try toggling the number of TCP connection using different numbers and hit ok after each choice. Verify that NO security policy has been set on the server side for the user restricting the amount of connections.

Hi, thanks a lot for taking the time trying to replicate the issue. I tried multiple values for TCP connections on the cascade advanced settings, such as 8, 16, 32... with no luck. The security policy for the Hub user is set with a maximum of 32 TCP connections.

Since it appears that you could not reproduce the issue I was concerned about my hardware, so I set up a clean install with latest version from the repo and skipped the steps configuring the local bridge. The result is the same, after changing the Cascade to "online", I click the "status" button and tcp connections ar 2/2.

I will try with different machines and I will try to dig deeper and examine the code, I whant to know exactly where does this MaxTCP - in the Cascade Status dialog - comes from, and how to change it.

David

[drodriguez]

So today I found the issue. It was a D-Link DSR-1000AC router in front of the VPN server creating the bottleneck. I bypassed it and TCP Connections quickly reached 32 and my systems began working properly. There are some machines at both sides of the VPN that rely on heavy broadcast, and removing broadcast limitations + increasing TCP connections made everything work fine for now.

Anyway, thanks a lot for the help!
David

[dr.dick]

So David which exactly limitation you have been changed?
I have the same issue with 2 TCP connect limitation for cascade connections and could not find any broadcast or tcp limitation on my routers similar to 2

[drodriguez]

Sadly I did not find the real cause of this bottleneck. The workaround was simply changing that router, I did not reconfigure it or anything so I can't tell you what to do there.
Broadcast and max TCP connection limitations I was refering to were both SoftEther configuration options, that you can tweak in the GUI or CMD tool.

[dr.dick]

Hmm, it really strange. Hub "knows" about limitation and we can see X/2 - in Managed Sessions. "2" is strong hub limitation. But why?
Please see attached image 1

The other issue is half duplex mode setting ignoring. Please see attached image 2
While in the connection settings in set up also as 8 tcp connection number: Please see attached image 3

It looks like client settings ignoring in cascade connections
Service was restarted several times without any changes. softether is 5.01 build 9674


Image 1

Image 2

Image 3

[dr.dick]

it looks like NAT Traversal issue while direct TCP are ok

[far6955@gmail.com]

guys this is a bug from softether, I have faced with the same issue and I have changed my server name (in my client connection) to server_name.vpnazure.net instead of server_name.sofether.net, and boom, the TCP connection has been changed from 2 to whatever I want to set and also QoS is enable, but now I have another issue which I think it is another bug from sofether, and the issue is that my ping time has been increased to almost double and also my download speed has been decreased to almost in half in most of the time, I mean it works fine for a couple of second in most cases and after that it goes down to half and sometimes it is even works with half speed right after I get connected, I DO NOT have this issue if i get connected with server_name.softether.net