AgentSkillsCN

local:testing

利用 Kind 在本地测试 Kagenti 平台。提供部署、测试与访问的脚本。

SKILL.md
--- frontmatter
name: local:testing
description: Test Kagenti platform locally using Kind. Scripts for deploy, test, and access.

Local Testing Skill

This skill helps you test Kagenti platform locally using Kind (Kubernetes in Docker).

Overview

The local-testing/ directory contains scripts that mirror the GitHub Actions CI workflow, allowing you to:

  • •Deploy the full Kagenti platform locally
  • •Run E2E tests with Ollama LLM integration
  • •Access the Kagenti UI and services
  • •Debug agent and tool deployments

Prerequisites

Ensure you have:

  • •Docker Desktop/Rancher Desktop/Podman (12GB RAM, 4 cores minimum)
  • •Kind, kubectl, helm installed
  • •Python 3.11+ with uv
  • •jq for JSON parsing

Available Scripts

1. Cleanup Cluster

Wipes the existing Kind cluster completely:

bash
./local-testing/cleanup-cluster.sh

Use this when you want a fresh start or cluster is in a bad state.

2. Deploy Platform

Deploys the full Kagenti platform (takes ~15-20 minutes):

bash
./local-testing/deploy-platform.sh

This script:

  • •Creates Kind cluster via Ansible installer
  • •Deploys all platform components (Keycloak, PostgreSQL, Istio, SPIRE)
  • •Installs Ollama and pulls qwen2.5:0.5b model
  • •Deploys weather agent and tool

3. Run E2E Tests

Runs the E2E test suite:

bash
./local-testing/run-e2e-tests.sh

Validates:

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

4. Access UI

Shows access information and port-forward commands:

bash
./local-testing/access-ui.sh

Then run the suggested command:

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

Visit: http://kagenti-ui.localtest.me:8080

Typical Workflow

Full Redeploy and Test

bash
# 1. Clean up existing cluster
./local-testing/cleanup-cluster.sh

# 2. Deploy platform (wait ~15-20 minutes)
./local-testing/deploy-platform.sh

# 3. Run E2E tests
./local-testing/run-e2e-tests.sh

# 4. Access UI
./local-testing/access-ui.sh

Quick Test Iteration

If platform is already deployed and you just want to rerun tests:

bash
./local-testing/run-e2e-tests.sh

Development Cycle

  1. •Make code changes
  2. •Run ./local-testing/deploy-platform.sh to redeploy
  3. •Run ./local-testing/run-e2e-tests.sh to validate
  4. •Use ./local-testing/access-ui.sh to get UI access

Debugging Commands

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 Pod Status

bash
# All pods
kubectl get pods -A

# Specific namespace
kubectl get all -n team1
kubectl get all -n kagenti-system

View Events

bash
# Recent events in team1 namespace
kubectl get events -n team1 --sort-by='.lastTimestamp' | tail -30

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

Ollama Status

bash
# Check if running
ps aux | grep ollama

# Check models
ollama list

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

Troubleshooting

Platform Not Ready

If deployment times out or components aren't ready:

bash
# Check pod status
kubectl get pods -A

# Check recent events
kubectl get events -A --sort-by='.lastTimestamp' | tail -50

# Retry deployment
./local-testing/cleanup-cluster.sh
./local-testing/deploy-platform.sh

Test Failures

If E2E tests fail:

  1. •Check agent logs: kubectl logs -n team1 deployment/weather-service --tail=100
  2. •Check tool logs: kubectl logs -n team1 deployment/weather-tool --tail=100
  3. •Verify Ollama is running: ps aux | grep ollama
  4. •Check Ollama has model: ollama list | grep qwen2.5

UI Access Issues

If UI doesn't load:

  1. •Verify pod is running: kubectl get pods -n kagenti-system -l app=kagenti-ui
  2. •Check Istio gateway: kubectl get pods -n kagenti-system -l app=http-istio
  3. •Restart port-forward: kubectl port-forward -n kagenti-system svc/http-istio 8080:80

Notes

  • •These scripts are for local development only (not committed to repo)
  • •Scripts are in .gitignore to avoid accidental commits
  • •Deployment uses same Ansible installer as CI
  • •Cluster name is kagenti (matches CI workflow)
  • •Secrets used are test values (not production)

Files Location

All scripts are in: local-testing/

  • •cleanup-cluster.sh - Wipe cluster
  • •deploy-platform.sh - Deploy platform
  • •run-e2e-tests.sh - Run tests
  • •access-ui.sh - Access information
  • •README.md - Full documentation

Related Skills

  • •local:full-test - Complete test workflows for Kind and HyperShift
  • •kind:cluster - Create/destroy Kind clusters
  • •kagenti:deploy - Deploy platform