Reconnection doesn't stop after disconnect-all-peers

[TheBlueMatt]

At least based on logs I got from @benthecarman, it appears ldk-node will still reconnect peers after the disconnect-all_peers call in the shutdown sequence.

His logs showed first
2025-08-13 16:52:49 INFO [ldk_node:649] Shutting down LDK Node with node ID
(immediately after which ldk-node will call PeerManager::disconnect_all_peers and expect no more peers to be connected),
then
2025-08-13 16:52:49 TRACE [lightning_background_processor:801] Terminating background processor.
(at which point having peers conneted isn't really safe in LDK cause updates to channel state won't be persisted in the ChannelManager)
followed by
2025-08-13 16:52:49 DEBUG [lightning::ln::peer_handler:1511] Finished noise handshake for connection with
(indicating the reconnection logic is still running)
and finally
2025-08-13 16:52:49 DEBUG [ldk_node:392] Stopping reconnecting known peers.

BDK then took some time (30 minutes?!) to shut down and the LSP forwarded an HTLC, resulting in
2025-08-13 17:29:48 ERROR [lightning::ln::channelmanager:13259] A ChannelManager is stale compared to the current ChannelMonitor!
on reload :(

[tnull]

BDK then took some time (30 minutes?!) to shut down

Mind explaining what exactly you mean by that? What exactly are you seeing where do you get the 30 minutes from?

[benthecarman]

Here's the full log file. There was no 30+ min shutdown. Every shutdown only took a second or two at most, that was just a separate run.

wallet.log

[TheBlueMatt]

Are you sure about that? There's a 2025-08-13 16:52:50 INFO [ldk_node:717] Shutdown complete. but the next starting is 2025-08-13 17:29:48 INFO [ldk_node:234] Starting up LDK Node with node ID

[tnull]

@benthecarman Now opened #612, please let me know if that addresses the observed behavior.