Skip to main content
Version: 0.0.1

Administrator Guide

This guide is for platform administrators who deploy, configure, and manage the Bison platform.

Responsibilities​

As a platform administrator, you are responsible for:

  • ✅ Deploying and configuring Bison
  • ✅ Creating and managing teams
  • ✅ Setting global billing configuration
  • ✅ Monitoring cluster-wide metrics
  • ✅ Responding to alerts and recharge requests

Getting Started​

1. Deploy Bison​

Follow the Installation Guide to deploy Bison in your Kubernetes cluster.

2. Configure Billing​

Set up billing rules and pricing:

  1. Access the Web UI
  2. Navigate to Settings > Billing Configuration
  3. Configure:
    • Currency: USD, CNY, EUR, etc.
    • CPU Price: Cost per core-hour
    • Memory Price: Cost per GB-hour
    • GPU Price: Cost per GPU-hour
  4. Click Save

3. Create First Team​

Create a team for your users:

  1. Navigate to Teams page
  2. Click Create Team
  3. Fill in:
    • Team Name: e.g., "ml-team"
    • Description: Team purpose
    • Resource Quota:
      • CPU: e.g., "20" cores
      • Memory: e.g., "64Gi"
      • GPU: e.g., "4"
    • Initial Balance: e.g., 1000.00
  4. Click Create

Common Tasks​

Managing Teams​

View All Teams​

# Via kubectl
kubectl get tenants

# Via API
curl http://localhost:8080/api/v1/teams

Update Team Quota​

  1. Navigate to Teams page
  2. Click Edit on the team row
  3. Modify quotas
  4. Click Save

Recharge Team Balance​

  1. Navigate to Teams page
  2. Click Recharge on the team row
  3. Enter amount
  4. Add notes (optional)
  5. Click Confirm

Monitoring​

View Dashboard​

Access real-time cluster metrics:

  • Total teams and projects
  • Resource utilization
  • Cost trends
  • Top consumers
  • Balance status

Check Alerts​

Monitor low-balance and quota alerts:

  1. Navigate to Alerts page
  2. Review active alerts
  3. Take action as needed

Billing Configuration​

Update Pricing​

curl -X PUT http://localhost:8080/api/v1/billing/config \
  -H "Content-Type: application/json" \
  -d '{
    "pricing": {
      "cpu": 0.06,
      "memory": 0.012,
      "nvidia.com/gpu": 3.00
    }
  }'

Configure Alert Thresholds​

{
  "lowBalanceThreshold": 20,
  "suspendThreshold": 5,
  "alertChannels": ["webhook", "dingtalk"]
}

Best Practices​

Team Naming​

  • Use lowercase, alphanumeric characters and hyphens
  • Example: ml-team, data-science, dev-team

Quota Allocation​

  • Start with conservative quotas
  • Monitor usage for 1-2 weeks
  • Adjust based on actual needs

Balance Management​

  • Set up auto-recharge for critical teams
  • Monitor balance trends weekly
  • Respond to low-balance alerts promptly

Security​

  • Enable authentication in production
  • Use OIDC/SSO for enterprise deployments
  • Regularly audit user permissions

Troubleshooting​

Team Creation Failed​

Check Capsule operator logs:

kubectl logs -n capsule-system deployment/capsule-controller-manager

Billing Not Working​

Verify OpenCost connectivity:

kubectl port-forward -n opencost-system svc/opencost 9003:9003
curl http://localhost:9003/healthz

High Resource Usage​

Check resource consumption:

kubectl top pods -n bison-system

Next Steps​