Summary

The image ghcr.io/openclaw/openclaw:latest fails to start
Error: Gateway service check failed: Error: systemctl --user unavailable: spawn systemctl ENOENT
The startup probe fails: Startup probe failed: dial tcp 10.42.0.X:18789: connect: connection refused
Container enters CrashLoopBackOff state
Environment:
Deployment platform: Zeabur (Kubernetes-based)
Image: ghcr.io/openclaw/openclaw:latest
Port: 18789
Expected behavior:
Container should start successfully without requiring systemctl
Application should listen on port 18789
Actual behavior:
Container crashes immediately
Application never starts

Steps to reproduce

1. Deploy the OpenClaw Docker image to a containerized environment (Kubernetes, Docker, etc.)
   • Image: ghcr.io/openclaw/openclaw:latest
   • Port: 18789
2. Set environment variables:
   • OPENCLAW_GATEWAY_PORT=18789
   • OPENCLAW_GATEWAY_BIND=0.0.0.0
   • NODE_ENV=production
3. Start the container
4. Observe the logs

Expected behavior

Expected Result:

Container starts successfully
Application listens on port 18789
Health checks pass

Actual behavior

Actual Result:
Container crashes with error: systemctl --user unavailable: spawn systemctl ENOENT
Container enters CrashLoopBackOff state
Application never starts

OpenClaw version

"Version: latest (deployed on 2026-02-24)"

Operating system

OS: Linux Environment: Containerized (Kubernetes)

Install method

Installation Method: Docker container (via Zeabur)

Logs, screenshots, and evidence

## Container Startup Logs  
[Zeabur] Pod startup sequence:
- Container pulled successfully: ghcr.io/openclaw/openclaw:latest
- Container created and started
- Application begins initialization  
## Error Messages  
Gateway service check failed:  
Error: systemctl --user unavailable: spawn systemctl ENOENT  
Startup probe failed:  
dial tcp 10.42.0.41:18789: connect: connection refused  
## Container Status  
[Zeabur] Pod/service-6995fe4a46ce828f10f1c49a - BackOff:   
Back-off restarting failed container openclaw  
Status: CRASHED  
State: CrashLoopBackOff  
## Application Logs (before crash)  
2026-02-24T18:26:24.805Z [gateway] listening on ws://0.0.0.0:18789 (PID 7)  
2026-02-24T18:26:24.805Z [gateway] agent model: google/gemini-3-flash-preview  
2026-02-24T18:26:24.803Z [health-monitor] started (interval: 300s, grace: 60s)  
2026-02-24T18:26:24.801Z [heartbeat] started  
2026-02-24T18:26:24.779Z [canvas] host mounted at http://0.0.0.0:18789/__openclaw__/canvas/  
## Issue  
Application starts and listens on port 18789, but then immediately crashes.  
The container never passes the startup health check.  
Kubernetes keeps restarting the container in a loop (CrashLoopBackOff).

Impact and severity

Affected Users/Systems

• Users attempting to deploy OpenClaw on Kubernetes-based platforms
• Cloud deployment services (Zeabur, AWS EKS, Google GKE, Azure AKS, etc.)
• Anyone using containerized environments
• DevOps teams and developers using modern cloud infrastructure

Severity

CRITICAL / HIGH
The application is completely non-functional in containerized environments.
Users cannot deploy or use OpenClaw at all on these platforms.

Frequency

ALWAYS - This occurs 100% of the time when deploying the Docker image to any containerized environment.

Practical Consequences

1. Deployment Failure: Users cannot deploy OpenClaw on modern cloud platforms
2. Wasted Resources: Container keeps restarting in CrashLoopBackOff, wasting CPU and memory
3. Blocked Workflow: Users cannot use OpenClaw for AI automation tasks
4. Extra Costs: Continuous failed restart attempts may incur unnecessary cloud charges
5. Poor User Experience: No clear error message explaining the incompatibility
6. Time Wasted: Users spend time troubleshooting a Docker image bug instead of using the product

Who This Affects

• New users trying to deploy OpenClaw on Zeabur or similar platforms
• Teams migrating to containerized infrastructure
• Anyone following the standard Docker deployment method

Additional information

No response