Investigate Retries exhausted error

[jvsena42]

Logs

Apparently it was cause by not finding a route with enough liquidity

Video displays 1 710 286 sats of lightning balance and the user was trying to pay an invoice of 3 947 sats

Got retriesExhausted error

Also did another attempt of 13 199 sats

[jvsena42]

17:12:04.592 | refreshCache() - early, before channels synced
17:12:05-06 | channel_reestablish events generating
17:12:06.476 | refreshCache() (1.86s)
17:12:08.366 | canSend: 1123 < 3947 ⚠️
17:12:17.228 | refreshCache() (10.68s - long sync!)
17:12:31.522 | Payment pushed successfully (~1.08M sats available)

The canSend check at 17:12:08 used stale cache data from 17:12:06.

The channels weren't fully synced yet - notice the 10.68 second refreshCache() at 17:12:17. The nextOutboundHtlcLimitMsat was temporarily low (1,123 sats) during channel reconnection, before it was updated to the real value (~1.08M sats).

But the payment still went through because:

1. canSend is only used for UI validation (showing toast, falling back to on-chain)
2. LDK internally had the correct channel state and successfully sent the HTLC
3. The payment failed due to routing issues (intermediate nodes), not the user's channel

The TemporaryChannelFailure(0x1007) came from downstream routing nodes, NOT from lack of outbound capacity. The user had enough sats - the network route just couldn't complete the payment.

Edit: Blocktank probing worked flawlessly

[jvsena42]

Other possibility is that the available routes are more expensive than the default 1% + 50 sats used by LDK
https://docs.rs/lightning/latest/lightning/routing/router/struct.RouteParametersConfig.html

[jvsena42]

Check for anchor reserve

It is not anchor reserve. There is no anchor reserve for thrusted peers

[jvsena42]

Attempt Time Route via Result
1 17:12:31 030ef51... ❌ TemporaryChannelFailure
2 17:12:38 030ef51... ❌ TemporaryChannelFailure
3 17:12:57 02b03a1... ❌ TemporaryChannelFailure
4 17:13:03 038bcc6... ❌ TemporaryChannelFailure