# Deploy with Helm

- Canonical URL: https://docs.fairvisor.com/docs/guides/helm/
- Section: docs
- Last updated: n/a
> Production Kubernetes deployment using the official Fairvisor Helm chart.


## Add the Helm repository

```bash
helm repo add fairvisor https://charts.fairvisor.com
helm repo update
```

## Install

```bash
helm install fairvisor fairvisor/fairvisor \
  --namespace fairvisor \
  --create-namespace \
  --set config.upstream.url=https://api.openai.com
```

## values.yaml reference

```yaml
replicaCount: 2

image:
  repository: ghcr.io/fairvisor/edge
  tag: ""          # defaults to chart appVersion
  pullPolicy: IfNotPresent

service:
  type: ClusterIP
  port: 8080

resources:
  requests:
    cpu: 100m
    memory: 64Mi
  limits:
    cpu: 500m
    memory: 256Mi

config:
  upstream:
    url: "https://api.openai.com"
    timeout: 30s
  exporters:
    prometheus:
      enabled: true

autoscaling:
  enabled: false
  minReplicas: 2
  maxReplicas: 10
  targetCPUUtilizationPercentage: 70
```

## Important: no distributed limiter state

Current OSS runtime does not support Redis/distributed shared counters.

- each pod keeps local in-memory counters
- with multiple pods, limits are enforced per-pod (not globally)

For HA patterns with predictable aggregate limits, see [HA Installation](/docs/deployment/ha-installation/).