openclaw - 💡(How to fix) Fix gateway restart uses systemctl stop which overrides Restart=always, causing dead service

Summary

When the OpenClaw gateway tool or openclaw gateway stop is invoked, it calls systemctl --user stop openclaw-gateway.service but does NOT follow up with systemctl --user start. Since systemctl stop is an explicit management command, it overrides the unit's Restart=always directive, leaving the service in inactive (dead) state.

Environment

• OpenClaw: v2026.5.18
• OS: Ubuntu 24.04, systemd 255
• Service: user-scope systemd unit with Restart=always

Reproduction

1. Run openclaw gateway stop or trigger a gateway restart via the gateway tool
2. Check systemctl --user status openclaw-gateway
3. Service is inactive (dead) instead of restarting automatically

Log Evidence

May 20 09:38:08 systemd[825]: Stopping openclaw-gateway.service
May 20 09:38:08 node[79384]: gateway signal SIGTERM received
May 20 09:38:09 systemd[825]: Stopped openclaw-gateway.service

No subsequent Starting openclaw-gateway.service line.

Service Config (correct, not the problem)

Restart=always
RestartSec=5
SuccessExitStatus=0 143

Root Cause

systemctl stop is an explicit admin action that tells systemd "the operator stopped this service." systemd respects this over Restart=always and does not auto-restart.

Fix

• Replace systemctl stop + delayed start with a single systemctl --user restart openclaw-gateway.service
• Or for process-level restart (SIGUSR1): let the process exit cleanly (process.exit(0)) so systemd's Restart=always handles the restart, rather than calling systemctl stop first

Workaround

Always use systemctl --user restart openclaw-gateway instead of openclaw gateway stop.