Getting Started

Set up AI-powered alert analysis in 5 minutes.

Prerequisites

✓ Robusta SaaS account (free or paid)
✓ Robusta UI sink enabled
✓ Robusta version 0.22.0 or higher

Use Robusta's hosted AI service with GPT-4o:

  1. Add to your Helm values:

    enableHolmesGPT: true
holmes:
  additionalEnvVars:
  - name: ROBUSTA_AI
    value: "true"

  2. Apply the changes:

    helm upgrade robusta robusta/robusta -f generated_values.yaml

  3. Enable Slack integration (optional):

    • Go to platform.robusta.dev

    • Navigate to Settings → AI Assistant

    • Toggle "Enable Holmes" and connect your Slack workspace

That's it! HolmesGPT will now analyze your alerts automatically.

Test Your Setup

Deploy a crashing pod to see HolmesGPT in action:

kubectl apply -f https://raw.githubusercontent.com/robusta-dev/kubernetes-demos/main/crashpod/broken.yaml

Then check: - Robusta UI: Look for the CrashLoopBackOff alert and click the "Root Cause" tab - Slack: Click "Ask HolmesGPT" on the alert notification

Using Your Own AI Provider

Instead of Robusta AI, you can use your own OpenAI, Azure, or AWS Bedrock account.

kubectl create secret generic holmes-secrets \
  --from-literal=openAiKey='YOUR_API_KEY'

enableHolmesGPT: true
holmes:
  additionalEnvVars:
  - name: MODEL
    value: gpt-4o
  - name: OPENAI_API_KEY
    valueFrom:
      secretKeyRef:
        name: holmes-secrets
        key: openAiKey

kubectl create secret generic holmes-secrets \
  --from-literal=azureOpenAiKey='YOUR_API_KEY'

enableHolmesGPT: true
holmes:
  additionalEnvVars:
  - name: MODEL
    value: azure/YOUR_DEPLOYMENT_NAME
  - name: AZURE_API_VERSION
    value: "2024-06-01"
  - name: AZURE_API_BASE
    value: https://your-org.openai.azure.com/
  - name: AZURE_API_KEY
    valueFrom:
      secretKeyRef:
        name: holmes-secrets
        key: azureOpenAiKey

Important: In Azure Portal, increase your deployment's token limit to at least 450K.

kubectl create secret generic holmes-secrets \
  --from-literal=awsAccessKeyId='YOUR_KEY_ID' \
  --from-literal=awsSecretAccessKey='YOUR_SECRET_KEY'

enableHolmesGPT: true
holmes:
  enablePostProcessing: true
  additionalEnvVars:
  - name: MODEL
    value: bedrock/anthropic.claude-3-5-sonnet-20240620-v1:0
  - name: AWS_REGION_NAME
    value: us-east-1
  - name: AWS_ACCESS_KEY_ID
    valueFrom:
      secretKeyRef:
        name: holmes-secrets
        key: awsAccessKeyId
  - name: AWS_SECRET_ACCESS_KEY
    valueFrom:
      secretKeyRef:
        name: holmes-secrets
        key: awsSecretAccessKey

Using Existing Secrets

If you store the Robusta UI token in a Kubernetes secret (instead of directly in Helm values), you need to pass it to HolmesGPT:

holmes:
  additionalEnvVars:
  - name: ROBUSTA_UI_TOKEN
    valueFrom:
      secretKeyRef:
        name: my-robusta-secrets  # Your existing secret
        key: ui-token             # Your existing key

Common Issues

Not seeing the "Ask HolmesGPT" button?

  • Ensure enableHolmesGPT: true is set

  • Check HolmesGPT pod is running: kubectl get pods -n robusta | grep holmes

  • Verify AI provider credentials are correct

Getting rate limit errors?

  • Azure: Increase token limit in Azure Portal (minimum 450K)

  • OpenAI: Check your API quota and billing

  • Consider using Robusta AI for unlimited investigations

Analysis seems incomplete?

  • Enable additional data sources in HolmesGPT configuration

  • Ensure Prometheus is configured for metrics analysis

  • Check that pod logs are accessible

Next Steps