Bug Description

In a Docker container environment on a Hostinger VPS (Debian 13), openclaw gateway restart does not behave reliably. The observed failure modes are:

1. openclaw gateway restart may orphan the gateway process — the service appears to stop but the process remains running, or fails to restart cleanly. The container itself may become unresponsive.

2. systemctl --user restart openclaw-gateway (as suggested in docs) does not work in this environment — systemctl is not available in the container context.

3. Using kill directly on the gateway process causes the entire container to become unresponsive, requiring a full container restart from outside.

Environment

• OS: Debian GNU/Linux 13 (trixie) in Docker container on Hostinger VPS
• OpenClaw version: 2026.4.15
• Node: v22.22.2
• Container: openclaw-40qa-openclaw-1 (Docker)

Observed Behavior

```

This does not restart cleanly:

openclaw gateway restart

systemctl is not available in the container:

systemctl --user restart openclaw-gateway

\u2192 systemctl: command not found

kill on gateway process hangs the container:

kill <gateway_pid>

\u2192 container becomes unresponsive, requires external restart

```

Expected Behavior

The gateway should be restartable without causing container instability. The suggested systemctl --user restart openclaw-gateway should work, OR there should be a Docker-native restart mechanism that does not require systemctl inside the container.

Workaround

The only reliable restart mechanism found so far is restarting the Docker container itself from the host:

``` docker restart openclaw-40qa-openclaw-1 ```

Suggested Fixes

1. Make openclaw gateway restart work reliably inside Docker without requiring systemctl
2. OR document a Docker-native restart command (e.g., openclaw gateway kill && openclaw gateway start)
3. OR add a flag like --force for restart operations that are known to be disruptive
4. Consider adding a health-check endpoint that can be polled to confirm the gateway is back up before returning