Probing for more reliable route fee estimation (routing)

Host: JssDWt  -  PR author: hieblmi


  • In this PR we extend the implementation of the graph based EstimateRouteFee` RPC with a payment probe based approach to obtain a more reliable routing fee estimate.


  • Why would sending a payment probe improve the fee estimate for EstimateRouteFee?
  • What’s the difference in behavior between calling with an invoice or a destination pubkey?
  • What is the purpose of the random payment hash in probePaymentRequest?
  • What is the expected payment response if a probe is successful?
  • What information have you learned about the network if a single htlc was used in SendPaymentV2?
  • And what information have you learned if multiple htlcs were involved?
  • How could you theoretically send a probe that is more reliable?
  • What defines an ‘LSP’ in the code?
  • Why is there an exception in the code for nodes connected only to an LSP?
  • What is the main difference in the way the fee is estimated for an LSP attached node?
  • How does the adjustment of route hints for LSPs work?
  • Why is there a test that probes to a destination without channels?