AgentSkillsCN

kind:cluster

管理本地 Kagenti 测试用的 Kind 集群。创建、销毁、部署平台，并执行端到端测试。

SKILL.md
--- frontmatter
name: kind:cluster
description: Manage Kind clusters for local Kagenti testing. Create, destroy, deploy platform, and run E2E tests.

Kind Cluster Management Skill

Manage Kind (Kubernetes in Docker) clusters for local Kagenti development and testing.

When to Use

  • •Creating a new local test cluster
  • •Destroying an existing cluster
  • •Deploying Kagenti platform to Kind
  • •Running E2E tests locally
  • •User asks "create cluster", "deploy locally", or "run tests"

Quick Start

bash
# Full workflow: create cluster -> deploy platform -> run tests
./.github/scripts/local-setup/kind-full-test.sh

Individual Scripts

Create Cluster

bash
# Create with default name (kagenti)
./.github/scripts/kind/create-cluster.sh

# Create with custom name
./.github/scripts/kind/create-cluster.sh my-cluster

# Or with environment variable
CLUSTER_NAME=my-cluster ./.github/scripts/kind/create-cluster.sh

Destroy Cluster

bash
# Destroy cluster by name
./.github/scripts/kind/destroy-cluster.sh

# Destroy specific cluster
./.github/scripts/kind/destroy-cluster.sh my-cluster

Deploy Platform

bash
# Deploy full Kagenti platform to existing cluster
./.github/scripts/kind/deploy-platform.sh

This deploys:

  • •Kagenti platform via Ansible installer
  • •All platform components (Istio, Keycloak, SPIRE, etc.)
  • •Ollama LLM (pulls qwen2.5:0.5b model)
  • •Weather agent and tool demo

Run E2E Tests

bash
# Run E2E tests against deployed platform
./.github/scripts/kind/run-e2e-tests.sh

Tests include:

  • •Platform deployment health
  • •Agent conversation via A2A protocol
  • •MCP tool invocation
  • •Ollama LLM integration

Access UI

bash
# Show access information and port-forward commands
./.github/scripts/kind/access-ui.sh

Then run the suggested command and visit:

Full Test Workflows

Kind Full Test

bash
# Full test: create -> deploy -> test (keeps cluster)
./.github/scripts/local-setup/kind-full-test.sh

# Destroy cluster after test
./.github/scripts/local-setup/kind-full-test.sh --include-cluster-destroy

Manual Step-by-Step

bash
# 1. Create cluster
./.github/scripts/kind/create-cluster.sh

# 2. Deploy platform (~15-20 minutes)
./.github/scripts/kind/deploy-platform.sh

# 3. Run tests
./.github/scripts/kind/run-e2e-tests.sh

# 4. Access UI
./.github/scripts/kind/access-ui.sh

Prerequisites

  • •Docker Desktop/Rancher Desktop/Podman (12GB RAM, 4 cores minimum)
  • •Kind: brew install kind
  • •kubectl: brew install kubectl
  • •helm: brew install helm
  • •Python 3.11+ with uv

Cluster Configuration

Default Kind config: deployments/ansible/kind/kind-config-registry.yaml

Custom config:

bash
KIND_CONFIG=/path/to/config.yaml ./.github/scripts/kind/create-cluster.sh

Environment Variables

VariableDefaultDescription
CLUSTER_NAMEkagentiKind cluster name
KIND_CONFIGkind-config-registry.yamlKind configuration file
SKIP_OLLAMAfalseSkip Ollama installation
SKIP_AGENTSfalseSkip agent deployment

Debugging

Check Cluster Status

bash
# List clusters
kind get clusters

# Check nodes
kubectl get nodes

# Check all pods
kubectl get pods -A

View Logs

bash
# Agent logs
kubectl logs -n team1 deployment/weather-service --tail=100 -f

# Tool logs
kubectl logs -n team1 deployment/weather-tool --tail=100 -f

# Platform operator
kubectl logs -n kagenti-system deployment/kagenti-platform-operator --tail=100 -f

Check Events

bash
kubectl get events -A --sort-by='.lastTimestamp' | tail -50

Ollama Status

bash
# Check if running
ps aux | grep ollama

# List models
ollama list

# Test API
curl http://localhost:11434/api/tags

Troubleshooting

Cluster Creation Fails

bash
# Check Docker resources
docker info | grep -E "CPUs|Total Memory"

# Increase Docker Desktop resources to 12GB+ RAM, 4+ CPUs

Platform Deployment Timeout

bash
# Check pod status
kubectl get pods -A | grep -v Running

# Check events
kubectl get events -A --sort-by='.lastTimestamp' | tail -30

# Retry
./.github/scripts/kind/destroy-cluster.sh
./.github/scripts/kind/create-cluster.sh
./.github/scripts/kind/deploy-platform.sh

E2E Tests Fail

bash
# Check agent health
kubectl get pods -n team1

# View agent logs
kubectl logs -n team1 deployment/weather-service --tail=100

# Check Ollama
ollama list | grep qwen

UI Not Accessible

bash
# Check gateway pod
kubectl get pods -n kagenti-system -l app=http-istio

# Restart port-forward
kubectl port-forward -n kagenti-system svc/http-istio 8080:80

Related Skills

  • •local:testing: Detailed local testing guide
  • •kagenti:deploy: Alternative deployment via Python installer
  • •k8s:pods: Debug pod issues
  • •k8s:logs: Query logs

Related Documentation

  • •.github/scripts/local-setup/README.md - Local setup documentation
  • •deployments/ansible/README.md - Ansible deployment guide