Repository navigation
Fixed
Critical: Systemd crash loop when proxy already running
When a manually-started proxy was already listening on port 8765, the systemd service would fail with EADDRINUSE, exit(1), and retry every 3 seconds indefinitely — causing thousands of restarts per session (7,580 recorded in one boot cycle), inotify watch exhaustion, and system instability.
Three-layer fix
server.ts— Pre-flight health check before binding. If a healthy proxy is already on the port,exit(0)so systemd treats it as successsetupService(CLI) — AddedStartLimitIntervalSec=30+StartLimitBurst=5to systemd unit. Also skips auto-restart during install if proxy is already runningcmdStart(CLI) — Callssystemctl --user reset-failedbefore starting, clearing stuck failed state
Verified
NRestarts=0(previously 7,580)- Clean
exit(0)on port conflict — no crash loop - systemd correctly manages proxy lifecycle
Full Changelog: v1.4.2...v1.4.3