All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- Relative amount parameters worked only from CLI, now they can be specified in the configs too
- Payment after probing used the global rebalancing context instead of attempt context which can result in waiting for the payment to fail for too long (6 hours by default)
- Rapid rebalance now goes all the way down to min amount to squeeze all available liquidity on the route
- Silent crash with zero exit code on FEE_INSUFFICIENT in some cases
- Check route for max htlc during rapid rebalance and limit the max rebalancing amount to not hit that cap
- 2% safety margin for channel reserve and commitment fee to prevent failures due to hitting that limit
- Exclude too young channels by their age in blocks: give liquidity a chance to move to the other side by itself. You can now set the minimum channel age to be considered for rebalance.
- Rapid rebalance now skips some steps if the channel is already constrained by liquidity
- Sudden fee changes are now properly handled: we retry rebalance and also see if the new fee is still within the limit
- Rapid rebalance summary now shows the correct total fee
- Functions doing high-level rebalancing logic are refactored to a separate file
- Rapid rebalance summary now shows the correct total amount
- Rapid rebalance is now accelerated, it tries to double the amount until it fails, then the amount is halved until it becomes lower than the initial one
- Goreleaser and Docker configs
- If any rapid rebalances succeeded, the total amount and fees are displayed at the end
- Stat file (CSV) is now also flock'ed to prevent accidental corruption if multiple instances try to update it
- Fees in route print and success message are now calculated from the target balance (without fees) so for example 50 sat fee and 1 000 000 sat amount would be shown as 50ppm, before it was 49ppm.
- Exit codes for failed rebalances (2 for global timeout and 1 for all other reasons)
- Incorrect lost profit description
- Probing doesn't return actual learned value and as such payment never happens
--exclude-from/--exclude-toparameters that support both channel and node IDs--infoparameter to see the channels selected and other effective settings that will be used during rebalance- .gitignore to prevent commiting sensitive info accidentally
--exclude-channel-in/--exclude-channel-outare deprecated and should be replaced with--exclude-from/--exclude-to(or respective config parameters)
- Channel that receives Ctrl-C signal is now buffered to satisfy the signal.Notify requirement
- Improved help message with grouped settings
- offline nodes used in
--from/--towere silently ignored and channels were chosen based on percentages
- Timeouts can be customized
- Config files updated and fixed to showcase the latest changes
--allow-unbalance-fromand--allow-unbalance-toare deprecated and enabled by default.- Zero amount is no longer supported, specify it explicitly or use
--rel-amount-to/--rel-amount-from --fail-tolerancenow also affects probing the same way it affects failed channels: if the next step is too close to the failed or successful step it's not tried and probing ends with the best known outcome- Use separate contexts for probing attempts and rapid rebalances
- This changelog
- Mission control helper to quickly skip routes that are likely to fail
- Discussion room in Matrix in README
--exclude-channeland--exclude-nodeare deprecated in favor of--excludewhich accepts both channel and node IDs--fromand--tonow accept node ids as well and will use all channels open to the specified nodes
- Channels excluded as targets are also excluded as sources in some situations