Restart Test Containers

Restart Test Containers Skill

Purpose: Restart Docker TEST containers the RIGHT way - this is the ONLY correct procedure.

CRITICAL: This skill is for TEST containers only. For DEVELOPMENT containers, use restart-dev-containers skill.

When to Use:

• Before running E2E tests (via test-environment skill)
• When test containers are unhealthy or stopped
• After code changes that need fresh test container build
• When test database needs reset (containers are recreated)

When NOT to Use:

• For dev containers - use restart-dev-containers instead
• If you just want to run tests and containers are healthy - use --skip-rebuild flag

CONTAINER ISOLATION

CRITICAL: Dev and test containers are isolated using different project names:

• Dev containers: -p witchcityrope-dev
• Test containers: -p witchcityrope-test

This prevents operations on one environment from affecting the other.

SINGLE SOURCE OF TRUTH

This skill is the ONLY place where test container restart procedure is documented.

If you find container restart instructions elsewhere:

1. They are outdated or wrong
2. Report to librarian for cleanup
3. Use THIS skill instead

DO NOT duplicate this procedure in:

• Agent definitions
• Lessons learned (reference this skill instead)
• Process documentation (reference this skill instead)

The Correct Process

DON'T Do This:

# WRONG - Missing test overlay
docker-compose up -d

# WRONG - Missing project name (will interfere with dev containers)
docker-compose -f docker-compose.yml -f docker-compose.test.yml up -d

# WRONG - Not using --no-cache for clean test builds
docker-compose -p witchcityrope-test -f docker-compose.yml -f docker-compose.test.yml up -d --build

DO This:

Use this skill - it handles everything correctly including project isolation and --no-cache.

How to Use This Skill

Executable Script: execute.sh

# From project root - clean rebuild with --no-cache (default, ensures fresh test files)
bash .claude/skills/restart-test-containers/execute.sh

# Skip confirmation prompt (for automation)
SKIP_CONFIRMATION=true bash .claude/skills/restart-test-containers/execute.sh

# Skip rebuild - just restart existing containers (fastest, when code hasn't changed)
bash .claude/skills/restart-test-containers/execute.sh --skip-rebuild

# Use Docker cache for faster rebuilds (when only non-test code changed)
bash .claude/skills/restart-test-containers/execute.sh --use-cache

# Combine flags for automation without rebuild
SKIP_CONFIRMATION=true bash .claude/skills/restart-test-containers/execute.sh --skip-rebuild

What the script does:

1. Shows pre-flight information (purpose, when/when NOT to use)
2. Requires confirmation before proceeding (skippable with env var)
3. Validates prerequisites (Docker running, project root, test overlay exists)
4. Stops existing test containers (using -p witchcityrope-test)
5. Builds and starts containers with test overlay (docker-compose.yml + docker-compose.test.yml)
   • Default: Builds with --no-cache for clean test images
   • With --use-cache: Uses Docker cache for faster builds
   • With --skip-rebuild: Skips build entirely, just starts containers
6. Waits for containers to initialize
7. Verifies health endpoints (Web, API, Database, Test Runner)
8. Reports status summary

Script includes safety checks - it will not run blindly without showing you what it's about to do.

Quick Reference Commands

Using the Skill (Recommended)

# From project root - clean rebuild with --no-cache (default)
bash .claude/skills/restart-test-containers/execute.sh

# For quick restart without rebuild (code hasn't changed)
bash .claude/skills/restart-test-containers/execute.sh --skip-rebuild

# For faster rebuild using Docker cache (when only non-test files changed)
bash .claude/skills/restart-test-containers/execute.sh --use-cache

# For automation - skip confirmation
SKIP_CONFIRMATION=true bash .claude/skills/restart-test-containers/execute.sh

Manual Steps (If execute.sh unavailable)

# CRITICAL: Always use -p witchcityrope-test to avoid affecting dev containers!

# Stop containers
docker-compose -p witchcityrope-test -f docker-compose.yml -f docker-compose.test.yml down -v

# Start with test overlay and --no-cache
docker-compose -p witchcityrope-test -f docker-compose.yml -f docker-compose.test.yml up -d --build --no-cache

# Wait and check health
sleep 20
docker exec witchcity-api-test curl -f http://localhost:8080/health
docker exec witchcity-web-test curl -f http://localhost:5173

Common Issues & Solutions

Issue: Containers start but tests fail immediately

Cause: Compilation errors in container or health not ready

Solution: Skill automatically checks container health. If issues persist, check container logs for errors in witchcity-web-test and witchcity-api-test.

Issue: Network overlap error

Cause: Old networks from previous runs or different subnets conflicting

Solution:

# Remove old test networks
docker network ls | grep witchcity | awk '{print $1}' | xargs docker network rm 2>/dev/null

# Try again
bash .claude/skills/restart-test-containers/execute.sh

Issue: Port already in use

Cause: Dev containers or other process using same ports

Note: Test containers do NOT expose ports externally (internal communication only). This error usually means dev containers are on conflicting networks.

Solution: Check if dev containers are running (they should be fine to run alongside). If network conflicts occur, prune unused Docker networks.

Issue: Dev containers were deleted when restarting test

Cause: Missing -p witchcityrope-test flag

Solution: Always use this skill or ensure the project flag is included

Integration with Other Skills

test-environment skill

The test-environment skill calls restart-test-containers with --skip-rebuild when containers are already healthy, or without the flag when a fresh build is needed.

# test-environment internally calls:
bash .claude/skills/restart-test-containers/execute.sh --skip-rebuild

restart-dev-containers skill

For development work, use the dev container skill which uses -p witchcityrope-dev project isolation.

Output Format

When run via Claude Code, skill returns:

{
  "skill": "restart-test-containers",
  "status": "success",
  "timestamp": "2025-12-01T15:30:00Z",
  "containers": {
    "running": 5,
    "expected": 5,
    "healthy": true
  },
  "build": {
    "mode": "full",
    "no_cache": true
  },
  "healthChecks": {
    "web": "healthy",
    "api": "healthy",
    "database": "healthy",
    "test_runner": "healthy"
  },
  "readyForTesting": true,
  "message": "Test environment ready"
}

On failure:

{
  "skill": "restart-test-containers",
  "status": "failure",
  "error": "Container health check failed",
  "details": "API service did not respond within timeout",
  "action": "Check API container logs for errors"
}

Related Skills

• restart-dev-containers: For restarting DEVELOPMENT containers (uses -p witchcityrope-dev)
• test-environment: For running E2E tests (calls restart-test-containers internally)

Maintenance

This skill is the single source of truth for TEST container restart.

To update the restart procedure:

1. Update THIS file only
2. Test the new procedure
3. DO NOT update process docs, lessons learned, or agent definitions
4. They should reference this skill, not duplicate it

Version History

• 2025-12-01: Created as new skill for test container isolation
  • Uses -p witchcityrope-test project isolation
  • Default build with --no-cache for clean test images
  • Added --skip-rebuild flag for faster restarts
  • 5 containers (postgres, api, web, test-runner, db-test-helper)
  • Internal network only (no exposed ports)

Remember: This skill is executable automation. Run it, don't copy it.