Connect several servers with each other bi-directionally

Hey,
I have created a network via Twingate and added two servers at different locations.

I could then access the server’s resources from my client on the Windows device and IOS. However, I also want the servers to be able to communicate with each other. However, when I ping one server to the other, it does not work, but goes through the outside instead of the tunnel.

Have I misunderstood something? Do I have to install the Twingate Client on every server? How does that work?

Thanks for help :slight_smile:

Hello and welcome to the forum!

Twingate is not bi-directional. Clients communicate with Connectors to connect to servers reachable from the Connector. Connectors aren’t clients and can’t access Resources.

To access Resources in your Twingate network, you need to have the client running. You need to have the client running on any machine that needs to access Resources. Since you’re talking about servers, check out our headless client setup guide.

The “you don’t need to run it everywhere” bit applies to Connectors. You don’t need to have the Connector running on each server for them to be accessible as a Resource. Connectors will connect clients with any machine that they can reach in the network, i.e. you only need a Connector on a Raspberry Pi in your home network to access to your NAS, your cameras, etc.

Hey, thanks for the explanation. I was now able to additionally install Twingate on the servers and connect it with the headless mode. Does this mean that no device will be deducted because it is headless?

In my case, the servers are all in a different global location and that means I have to install the connector everywhere, right?
Should I then create a network and put all the connectors in it from all the servers, or should I create a separate network for each server?

Headless clients are treated as services, not devices, so I don’t believe you’ll run into any limits there.

In my case, the servers are all in a different global location and that means I have to install the connector everywhere, right?
Should I then create a network and put all the connectors in it from all the servers, or should I create a separate network for each server?

It sounds like you need one Connector for each server.

I’d suggest reading our Remote Network documentation to better understand how Remote Networks and Connectors work together. Each Remote Network is a logical container that groups Resources with specific Connectors. Every Resource within a Remote Network must be reachable from all of the Connectors in the Remote Network.

If you have X servers and each individual server can’t talk to any of the other servers, each one will need its own Connector. Sounds like this is your situation.

If, however, you have X servers and they’re all in the same subnet and can talk to each other, you only need one Connector for all of them.

Thank you very much that helps me.
I still have one question, if the servers of Twingate are not accessible, what happens then? Can the servers still communicate, simply no changes can be made or does nothing work?

Depends on what specifically is down, but it will range from degraded service to things not being able to communicate. We have a status page that will let you know what’s up and running.