Skip to main content
Version: Next

Configuration

This guide covers how to configure Bison for your specific environment and requirements.

Helm Chart Configuration​

Bison is configured primarily through Helm values. You can customize the installation by providing a values.yaml file or using --set flags.

Key Configuration Parameters​

ParameterDescriptionDefaultExample
auth.enabledEnable authenticationfalsetrue
auth.admin.usernameAdmin usernameadminadmin
auth.admin.passwordAdmin passwordadminchangeme
apiServer.replicaCountAPI server replicas23
apiServer.image.repositoryAPI server imageghcr.io/supermarioyl/bison/api-server-
apiServer.image.tagAPI server image tag0.0.1latest
webUI.replicaCountWeb UI replicas23
webUI.image.repositoryWeb UI imageghcr.io/supermarioyl/bison/web-ui-
webUI.image.tagWeb UI image tag0.0.1latest
opencost.urlOpenCost API endpointhttp://opencost.opencost-system.svc:9003Custom URL

Example Custom Values​

Create a custom-values.yaml file:

# Authentication
auth:
  enabled: true
  admin:
    username: admin
    password: MySecurePassword123

# API Server
apiServer:
  replicaCount: 3
  image:
    tag: 0.0.1
  resources:
    requests:
      cpu: 200m
      memory: 256Mi
    limits:
      cpu: 1000m
      memory: 512Mi

# Web UI
webUI:
  replicaCount: 3
  image:
    tag: 0.0.1
  resources:
    requests:
      cpu: 100m
      memory: 128Mi
    limits:
      cpu: 500m
      memory: 256Mi

# OpenCost Integration
opencost:
  url: http://opencost.opencost-system.svc:9003

# Node Selection (optional)
nodeSelector:
  node-role.kubernetes.io/control-plane: ""

# Tolerations (optional)
tolerations:
  - key: node-role.kubernetes.io/control-plane
    operator: Exists
    effect: NoSchedule

Install with custom values:

helm install bison bison/bison \
  --namespace bison-system \
  --create-namespace \
  --values custom-values.yaml

Billing Configuration​

Billing settings are configured through the Web UI or API after installation.

Access Billing Configuration​

  1. Via Web UI:

    • Navigate to Settings > Billing Configuration
    • Set pricing for CPU, Memory, GPU, and other resources
    • Configure currency and billing intervals

  2. Via API:

    curl -X POST http://localhost:8080/api/v1/billing/config \
      -H "Content-Type: application/json" \
      -d '{
        "enabled": true,
        "currency": "USD",
        "pricing": {
          "cpu": 0.05,
          "memory": 0.01,
          "nvidia.com/gpu": 2.50
        },
        "billingInterval": "hourly"
      }'

Billing Parameters​

ParameterDescriptionExample
enabledEnable/disable billingtrue
currencyCurrency for billingUSD, CNY, EUR
pricing.cpuCPU price per core-hour0.05
pricing.memoryMemory price per GB-hour0.01
pricing["nvidia.com/gpu"]GPU price per GPU-hour2.50
billingIntervalBilling aggregation periodhourly, daily
lowBalanceThresholdWarning threshold (%)20
suspendThresholdAuto-suspend threshold (%)5

Example Billing Configuration​

{
  "enabled": true,
  "currency": "USD",
  "pricing": {
    "cpu": 0.05,
    "memory": 0.01,
    "nvidia.com/gpu": 2.50,
    "nvidia.com/mig-1g.5gb": 0.50,
    "nvidia.com/mig-2g.10gb": 1.00
  },
  "billingInterval": "hourly",
  "lowBalanceThreshold": 20,
  "suspendThreshold": 5,
  "alertChannels": ["webhook", "dingtalk"]
}

Team Configuration​

Creating Teams​

Teams can be created through the Web UI or API:

Via Web UI:

  1. Navigate to Teams page
  2. Click Create Team
  3. Set team name, quota, and initial balance

Via API:

curl -X POST http://localhost:8080/api/v1/teams \
  -H "Content-Type: application/json" \
  -d '{
    "name": "ml-team",
    "description": "Machine Learning Team",
    "quota": {
      "cpu": "20",
      "memory": "64Gi",
      "nvidia.com/gpu": "4"
    },
    "balance": 1000.00
  }'

Team Quotas​

Team quotas define resource limits:

quota:
  cpu: "20"              # 20 CPU cores
  memory: "64Gi"         # 64 GB RAM
  nvidia.com/gpu: "4"    # 4 GPUs
  storage: "500Gi"       # 500 GB storage

Team Balance Management​

Set initial balance and configure auto-recharge:

{
  "balance": 1000.00,
  "autoRecharge": {
    "enabled": true,
    "amount": 500.00,
    "schedule": "monthly",
    "threshold": 100.00
  }
}

Alert Configuration​

Configure multi-channel alerts for low balance and quota warnings.

Webhook Alerts​

{
  "type": "webhook",
  "enabled": true,
  "url": "https://your-webhook-endpoint.com/alerts",
  "headers": {
    "Authorization": "Bearer YOUR_TOKEN"
  },
  "template": {
    "title": "Bison Alert",
    "message": "Team {{.TeamName}} balance is {{.Balance}}"
  }
}

DingTalk Alerts​

{
  "type": "dingtalk",
  "enabled": true,
  "webhook": "https://oapi.dingtalk.com/robot/send?access_token=YOUR_TOKEN",
  "secret": "YOUR_SECRET"
}

WeChat Work Alerts​

{
  "type": "wechat",
  "enabled": true,
  "corpid": "YOUR_CORP_ID",
  "corpsecret": "YOUR_CORP_SECRET",
  "agentid": 1000001
}

OpenCost Integration​

Configure OpenCost connection:

Check OpenCost Connectivity​

# Test OpenCost API
kubectl port-forward -n opencost-system svc/opencost 9003:9003
curl http://localhost:9003/healthz

# Test allocation API
curl http://localhost:9003/allocation/compute?window=1d

Update OpenCost URL​

If OpenCost is deployed in a different namespace or with a different service name:

helm upgrade bison bison/bison \
  --set opencost.url=http://my-opencost.custom-namespace.svc:9003 \
  --namespace bison-system

Authentication & OIDC​

Enable authentication and integrate with your SSO provider:

Basic Authentication​

auth:
  enabled: true
  admin:
    username: admin
    password: SecurePassword123

OIDC Integration​

auth:
  enabled: true
  oidc:
    enabled: true
    issuerURL: https://your-oidc-provider.com
    clientID: bison-client-id
    clientSecret: your-client-secret
    redirectURL: https://bison.example.com/callback

Environment Variables​

Additional configuration can be provided via environment variables:

VariableDescriptionDefault
KUBECONFIGPath to kubeconfig fileIn-cluster config
OPENCOST_URLOpenCost API URLhttp://opencost.opencost-system.svc:9003
AUTH_ENABLEDEnable authenticationfalse
LOG_LEVELLogging levelinfo
BILLING_INTERVALBilling calculation interval10m

Set environment variables in Helm values:

apiServer:
  env:
    - name: LOG_LEVEL
      value: debug
    - name: BILLING_INTERVAL
      value: 5m

Advanced Configuration​

Custom Resource Pricing​

Price any Kubernetes resource:

{
  "pricing": {
    "cpu": 0.05,
    "memory": 0.01,
    "nvidia.com/gpu": 2.50,
    "amd.com/gpu": 2.00,
    "ephemeral-storage": 0.001,
    "custom.io/fpga": 5.00
  }
}

Multi-Cluster Support​

Deploy Bison in each cluster with shared billing:

# Cluster A
apiServer:
  clusterName: prod-us-west

# Cluster B
apiServer:
  clusterName: prod-us-east

Next Steps​