Quick Start

Get OnCallM running in your Kubernetes cluster in under 5 minutes.

Prerequisites

Before starting, ensure you have:

• ✅ Kubernetes cluster (v1.20+)
• ✅ Helm 3.x installed
• ✅ Prometheus & AlertManager running
• ✅ OpenAI API key

TIP

Need help with prerequisites? Check our detailed prerequisites guide.

Installation Steps

1. Clone the Repository

git clone https://github.com/OncaLLM/oncallm.git
cd oncallm

2. Configure Values

Create your local configuration file:

cp charts/oncallm/values.yaml values.local.yaml

Edit values.local.yaml with your settings:

oncallm:
  env:
    OPENAI_API_KEY: "your-openai-api-key-here"
    APP_HOST: "0.0.0.0"
    APP_PORT: "8001"
  
  service:
    type: ClusterIP
    port: 8001
  
  resources:
    requests:
      memory: "256Mi"
      cpu: "250m"
    limits:
      memory: "512Mi"
      cpu: "500m"

3. Deploy with Helm

Install OnCallM to your cluster:

helm install oncallm ./charts/oncallm -f values.local.yaml

4. Verify Installation

Check that OnCallM is running:

kubectl get pods -l app=oncallm
kubectl logs -l app=oncallm

You should see output similar to:

NAME                      READY   STATUS    RESTARTS   AGE
oncallm-7d4b8c9f5-xk2m8   1/1     Running   0          30s

5. Configure AlertManager

Add OnCallM webhook to your AlertManager configuration:

# alertmanager.yml
global:
  # ... your global config

route:
  # ... your routing config
  routes:
  - match:
      severity: critical
    receiver: 'oncallm-webhook'

receivers:
- name: 'oncallm-webhook'
  webhook_configs:
  - url: 'http://oncallm.default.svc.cluster.local:8001/webhook'
    send_resolved: true
    max_alerts: 0

6. Test the Integration

Trigger a test alert to verify the integration:

# Create a test alert
kubectl run test-alert --image=nginx --restart=Never
kubectl delete pod test-alert

What's Next?

🎉 Congratulations! OnCallM is now running in your cluster.

Check Your First Report

1. Access the OnCallM service:

   kubectl port-forward svc/oncallm 8001:8001

2. Visit http://localhost:8001/reports to see analysis reports

3. When alerts are triggered, you'll see detailed analysis at http://localhost:8001/report/{fingerprint}

Next Steps

• Configure AlertManager properly
• Set up resource limits
• Learn about the API
• Configure monitoring

Troubleshooting

Common Issues

Pod not starting?

kubectl describe pod -l app=oncallm
kubectl logs -l app=oncallm

Can't connect to OpenAI?

• Verify your API key is correct
• Check network policies allow outbound HTTPS
• Ensure the secret is properly mounted

AlertManager not sending webhooks?

• Verify the webhook URL is reachable from AlertManager
• Check AlertManager logs for webhook delivery errors
• Test connectivity: kubectl exec -it alertmanager-pod -- wget -O- http://oncallm.default.svc.cluster.local:8001/health

Getting Help

• 📖 Read the full documentation
• 🐛 Report issues on GitHub
• 📧 Enterprise support

Need Enterprise Support?

For professional deployment assistance, custom configurations, and dedicated support, contact our team.