Do not continue reconnecting if the last failure was permanent

[laanwj]

I get the following, repeatedly, in the log of my mainnet lightning node:

lightningd(23449): lightning_channeld(<peerid>): STATUS_FAIL_PEER_IO: bad reestablish revocation_number: 316 vs 366
lightningd(23449): peer <peerid>: Peer transient failure in CHANNELD_NORMAL: Owning subdaemon lightning_channeld died (33792)

This happens after the node aborted with a repeated "missing inputs" error (I don't have the exact log of that anymore, sorry.) It looks like that caused my node to be out of whack with another peer, but that doesn't stop it from trying.

[cdecker]

Yeah, we need to stop reconnecting if we terminated with a permanent failure. That particular one indicates that the two endpoints disagree on the version of the state.

[cdecker]

I took the liberty of making this a broader issue that we can use to track progress.

[ZmnSCPxj]

Is this still an issue?

[laanwj]

I don't have any way to test, sorry. From my side it's fine to close this issue (but maybe someone else knows better).

[ZmnSCPxj]

@cdecker ? Is this "broader issue" still existing?

[cdecker]

I think we can close, it got better once we had the backoff mechanism and then I think we fixed it completely by transitioning the channel into a failing state and disabling reconnection completely