v1.6.6 - Peer History Retention Fix
v1.6.6
This release fixes a long-standing peer history retention bug that caused the Peers tool to lose older data prematurely on high-volume meshes.
Changes
- Moved
/peersstatistics off rawroute_history_segments - Added dedicated rolling peer-history buckets for directional peer counts
- Made peer counts independent of
ROUTE_HISTORY_MAX_SEGMENTS - Added pruning logic so peer-history buckets remain bounded by time
- Added state persistence so peer-history survives restarts
- Added rebuild logic to reconstruct peer buckets from saved route history when needed
Why this release
On larger meshes, the Peers tool could report a 24-hour window while actually only retaining a few hours of data due to early truncation by ROUTE_HISTORY_MAX_SEGMENTS.
This release separates peer statistics from route segment limits, ensuring accurate long-term visibility.
Testing
- Added regression tests to ensure peer counts remain accurate when route history reaches segment limits
- Added regression tests for peer-history persistence across restarts
- Full test suite passing:
57 passed