fix(cleanup): prevent event listener and log stream leaks, tighten smartProxy connection timeouts, and improve graceful shutdown behavior
This commit is contained in:
10
changelog.md
10
changelog.md
@@ -1,5 +1,15 @@
|
||||
# Changelog
|
||||
|
||||
## 2026-02-26 - 9.1.6 - fix(cleanup)
|
||||
prevent event listener and log stream leaks, tighten smartProxy connection timeouts, and improve graceful shutdown behavior
|
||||
|
||||
- Tightened smartProxy connection timeouts and lifetimes (5m socketTimeout, 10m inactivityTimeout, keep-alive multiplier, 1h extendedKeepAliveLifetime, 4h maxConnectionLifetime).
|
||||
- Remove event listeners before stopping services to avoid leaks (smartProxy, emailServer, dnsServer, remote ingress hub).
|
||||
- OpsServer.stop now invokes logsHandler.cleanup to tear down active log streams and avoid duplicate push destinations.
|
||||
- LogsHandler rewritten to use a module-level singleton push destination, track active stream stop callbacks, add cleanup(), guard against hung VirtualStream.sendData with a 10s timeout, and ensure intervals are cleared on stop.
|
||||
- updateSmartProxyConfig removes listeners on the old instance before stopping it.
|
||||
- Dependency bumps: @api.global/typedsocket ^4.1.2, @push.rocks/smartdata ^7.1.0, @push.rocks/smartmta ^5.2.6, @push.rocks/smartproxy ^25.8.3.
|
||||
|
||||
## 2026-02-26 - 9.1.5 - fix(remoteingress)
|
||||
Reconcile tunnel manager edge statuses with authoritative Rust hub periodically; update active tunnel counts and heartbeats, add missed edges, remove stale entries, and clear reconcile interval on stop
|
||||
|
||||
|
||||
Reference in New Issue
Block a user