Skip to main content

How does the 'teleporting' function work on a technical level? Based on the documentation on ports (https://docs.cribl.io/stream/ports/), it seems no inbound ports need to be opened on worker or edge nodes for the feature to work. The leader node handles teleport functionality via port 4200. When teleporting to a node, does it simply stream the UI data to the leader, essentially mirroring the node's interface rather than establishing a direct session? I couldn’t find detailed documentation on this, and tools like TCP dump/Wireshark or internal logs haven’t provided further insights.

Thank you :)

https://docs.cribl.io/stream/ports/

No, an inbound port on the Worker is not required nor used. Teleporting uses the existing TCP 4200 connection that the Worker established to the Leader.

Thanks for the answer!

That means, the Worker node maintains the connection used for teleporting at all times, allowing the leader to use it instantly? It feels odd to me that the connection is "officially" initiated via the GUI on the leader node, but, as I understand it, the Worker doesn’t actually receive a direct request from the leader during this process.

The WN maintains the 4200 connection at all times because that's the control channel. Configs, deployments, restart notifications, scheduled collectors…. lot of stuff happens over the control channel. Teleport is one of them (if you have it enabled).

Reply


Terms & ConditionsCookie settings