v4.1.3

Fixed

• sling-once: if a channel was closed during this command, sling would not exit the job properly and also cause sling-stats to hang

Release binaries info

• Release binaries were built using rust 1.92.0
• Linux release binaries require glibc>=2.35 (check yours with ldd --version)
• macOS universal binary (x86_64 + aarch64), requires macOS >= 11.0 (deployment target)

v4.1.2

Fixed

• Don't exit if xpay layer was not created yet (should only happen in CI)

Release binaries info

• Release binaries were built using rust 1.91.1
• Linux release binaries require glibc>=2.31

v4.1.1

Fixed

• Important fix if you have more than one channel with the same peer (see Issue#11): They can choose on which channel to forward sats to you and might not choose the scid from sling's route. This can be counter-productive and lead to pointless back and forth on the same channel while losing sats to fees. sling now checks if the sats arrive on the correct channel and if not fails the rebalance and bans the peer for one hour at which point it will try again. Reported by @whitslack, thank you!

Release binaries info

• Release binaries were built using rust 1.89.0
• Linux release binaries require glibc>=2.31

v4.1.0

Added

• sling-deletejob: add an optional boolean argument delete_stats which defaults to false. If set to true it will delete the stats of the job(s) aswell.
• sling-autogo: new option that defaults to false. If set to true it will automatically start all rebalance jobs upon sling startup.

Changed

• raised MSRV to 1.85 since both CLN and debian stable now support it

Release binaries info

• Release binaries were built using rust 1.89.0
• Linux release binaries require glibc>=2.31

v4.0.0

Removed

• Dropped support for CLN <= v24.02, upgrade your node!
• ⚠️ Make sure to comment out any removed options in your config
• sling-refresh-peers-interval: Lowering it from the default 1 doesn't help much and if you must set it higher your node is probably too slow anyways
• sling-refresh-gossmap-interval: Gossip reader is so fast now there is no good reason to keep it

Added

• sling-once: New command to rebalance a specific amount once
• sling-stats: Added human readable table view for when you provide a ShortChannelId, json flag can still be set
• Sling now also periodically reads constraint information from the xpay layer in CLN >= v24.11

Changed

• Optimized gossip file reader to be ~20x faster with similar memory usage. On my system it can read a fully synced gossip file in ~110ms and periodical checks for updates are now instant: ~0ms.
• Optimized route search to be ~2x faster, on my system down from ~25ms to ~13ms.
• Make use of the new trace level logging, some debug logs are now trace
• If possible show node alias in failure log message instead of id
• sling-reset-liquidity-interval: Default value increased to 360m again, if liquidity beliefs are forgotten too quickly it may result in an infinite loop of trying the very cheapest, never succeeding paths
• Slightly increased the minimum channel cost in pathfinding
• Use actual default options so they show up in CLN's listconfigs
• Refactor to show method usage in CLN's help
• switch from jemalloc to mimalloc for compatibility on systems using page sizes >4k

Fixed

• You can no longer add your own channels or your own node id to exceptions, for candidate control use candidatelist/outppm
• Jobstates are now sorted properly in the stats view

Release binaries info

• Release binaries were built using rust 1.88.0
• Linux release binaries require glibc>=2.31

v3.0.6

Fixed

• no longer require scid aliases for private channels during route calculation aswell

Release binaries info

• Release binaries were built using rust 1.86.0
• Linux release binaries require glibc>=2.31

v3.0.5

Changed

• no longer require scid aliases for private channels

Release binaries info

• Release binaries were built using rust 1.86.0
• Linux release binaries require glibc>=2.31

v3.0.4

Added

• more debug logging for candidates selection

Changed

• only add private channels to graph if they have an alias, remove private channels without an alias
• lower log level to debug for repetitive graph refresh lines

Release binaries info

• Release binaries were built using rust 1.86.0
• Linux release binaries require glibc>=2.31

v3.0.3

Changed

• check private channels for existing aliases when adding a job

Release binaries info

• Release binaries were built using rust 1.85.0
• Linux release binaries require glibc>=2.31

v3.0.2

Fixed

• support modified cln clients with prefix in version strings

Release binaries info

• Release binaries were built using rust 1.85.0
• Linux release binaries require glibc>=2.31