fix(rustproxy-metrics): retain inactive per-IP metric buckets briefly to capture final throughput before pruning
This commit is contained in:
@@ -1,5 +1,12 @@
|
||||
# Changelog
|
||||
|
||||
## 2026-04-26 - 27.8.2 - fix(rustproxy-metrics)
|
||||
retain inactive per-IP metric buckets briefly to capture final throughput before pruning
|
||||
|
||||
- adds a bounded retention window for closed IP buckets so short-lived transfers are still included in per-IP throughput sampling
|
||||
- prunes expired inactive IP tracking by TTL and hard cap to prevent unbounded metric map growth
|
||||
- updates Rust and throughput tests to expect zero active connections during the temporary retention period
|
||||
|
||||
## 2026-04-26 - 27.8.1 - fix(rustproxy-metrics)
|
||||
preserve high-throughput IPs in metrics snapshots when active-connection rankings are saturated
|
||||
|
||||
|
||||
Reference in New Issue
Block a user