k8s-game-2048/docs/WORKFLOWS.md

🔄 CI/CD Pipeline Documentation

This document describes the complete automated deployment pipeline for the Knative 2048 Game on k3s.

📋 Table of Contents

• Pipeline Overview
• Workflow Details
• Manual Actions
• Environment Configuration
• Troubleshooting

🎯 Pipeline Overview

Complete Automatic Flow

graph TD
    A[Push to develop] --> B[Build & Push Image]
    B --> C[Deploy to Development]
    C --> D[Smoke Tests Dev]
    D --> E[Auto-Promote to Staging]
    E --> F[Build & Push Staging Image]
    F --> G[Deploy to Staging]
    G --> H[Smoke Tests Staging]
    H --> I[Auto-Promote to Production]
    I --> J[Push to main]
    J --> K[Build & Push Prod Image]
    K --> L[Deploy to Production]
    L --> M[Smoke Tests Production]
    
    N[Manual Deploy Prod] -.-> L
    O[Manual Promote Prod] -.-> I
    P[Manual Smoke Tests] -.-> D
    P -.-> H
    P -.-> M

Key Principles

• Fully Automatic: Zero manual intervention from develop to production
• No Race Conditions: Each step waits for the previous to complete
• Test After Deploy: Smoke tests run on newly deployed versions
• Commit-Specific Images: Each environment uses exact commit-tagged images
• Automatic Promotion: Successful tests trigger automatic promotion
• Manual Override: Emergency manual deployment still available

🔧 Workflow Details

1. Build and Push Container Image (build-image.yml)

Triggers:

• Push to main, develop, staging
• Pull requests to these branches

What it does:

• Builds Docker image from current commit
• Creates commit-specific tags: {branch}-{commit-hash}
• Pushes to GitHub Container Registry (GHCR)
• Provides foundation for all deployments

Tags created:

• develop-abc1234 (for develop branch)
• staging-def5678 (for staging branch)
• main-ghi9012 (for main branch)

2. Deploy to Development (deploy-dev.yml)

Triggers:

• After "Build and Push Container Image" completes successfully on develop
• Manual dispatch

What it does:

• Waits for build to complete (no race conditions)
• Uses exact commit-tagged image that was just built
• Deploys via webhook to k3s development namespace
• Sets up development environment

Dependencies:

• Requires successful build completion
• Uses environment secrets: DEV_WEBHOOK_URL, WEBHOOK_SECRET

3. Smoke Tests (smoke-test.yml)

Triggers:

• After any deployment completes ("Deploy to Development", "Deploy to Staging", "Deploy to Production")
• Scheduled every 6 hours
• Manual dispatch

What it does:

• Tests the newly deployed version (not previous)
• Validates canonical Knative domains
• Checks content, performance, SSL certificates
• Runs environment-specific tests

Environments tested:

• 🧪 Development: Your configured development domain
• 🎭 Staging: Your configured staging domain
• 🚀 Production: Your configured production domain

4. Auto-Promote Pipeline (auto-promote.yml)

Triggers:

• After "Smoke Tests" complete successfully on develop branch

What it does:

• Verifies development smoke tests passed
• Merges develop → staging automatically
• Triggers staging deployment pipeline
• Creates promotion summary

Safety features:

• Only runs if smoke tests pass
• Handles "already up to date" scenarios gracefully

5. Deploy to Staging (deploy-staging.yml)

Triggers:

• Push to staging branch (triggered by auto-promotion)
• After "Auto-Promote Pipeline" completes
• Manual dispatch

What it does:

• Builds and deploys staging-specific image
• Uses staging-{commit} tagged image
• Deploys via webhook to k3s staging namespace

6. Auto-Promote to Production (promote-to-production.yml)

Triggers:

• After "Smoke Tests" complete successfully on staging branch (AUTOMATIC)
• Manual dispatch (emergency override only)

What it does:

• Verifies staging smoke tests passed
• Merges staging → main automatically
• Triggers production deployment immediately
• Creates production promotion summary

Automation features:

• Runs automatically after staging tests pass
• No manual confirmation required
• Seamless promotion from staging to production

7. Deploy to Production (deploy-prod.yml)

Triggers:

• Push to main branch (triggered by auto-promotion) - AUTOMATIC
• Manual dispatch (requires typing "DEPLOY" for emergencies)

What it does:

• Automatically deploys when main branch is updated
• Uses main-{commit} tagged image
• Deploys via webhook to k3s production namespace
• Blue-green deployment strategy for zero downtime

Automation features:

• No manual confirmation required for automatic deployments
• Immediate deployment after staging promotion
• Manual override still available for emergencies

8. Deployment Status Check (deployment-status.yml)

Triggers:

• Manual dispatch
• Scheduled every 4 hours

What it does:

• Checks health of all environments
• Shows current versions deployed
• Provides manual action options
• Creates comprehensive status report

🎮 Manual Actions (Emergency Use Only)

Note

: The pipeline is fully automatic. Manual actions are only for emergency situations or debugging.

Emergency Actions

Action	Workflow	Required Input	Use Case
Check Status	Deployment Status Check	None	Monitor all environments
Test Environment	Smoke Tests	Environment (dev/staging/prod/all)	Debug specific environment
Emergency Deploy	Deploy to Production	Type "DEPLOY"	Emergency production fix
Force Promotion	Auto-Promote to Production	None	Skip normal promotion flow

Emergency Procedures

Emergency Production Deployment

Use only if automatic pipeline is broken

1. Go to Actions → "Deploy to Production"
2. Click "Run workflow"
3. Type "DEPLOY" in confirmation field
4. Optionally specify image tag
5. Click "Run workflow"

Force Production Promotion

Use only if auto-promotion fails

1. Go to Actions → "Auto-Promote to Production"
2. Click "Run workflow"
3. Optionally skip tests if staging already validated
4. Click "Run workflow"

3. Check Deployment Status

1. Go to Actions → "Deployment Status Check"
2. Click "Run workflow"
3. View results in workflow summary

4. Run Smoke Tests

1. Go to Actions → "Smoke Tests"
2. Click "Run workflow"
3. Select environment to test
4. Click "Run workflow"

⚙️ Environment Configuration

Required Secrets

Secret	Purpose	Used By
GH_TOKEN	GitHub Container Registry access	Build workflows
WEBHOOK_SECRET	Webhook signature validation	All deployment workflows
DEV_WEBHOOK_URL	Development deployment endpoint	Deploy to Development
STAGING_WEBHOOK_URL	Staging deployment endpoint	Deploy to Staging
PROD_WEBHOOK_URL	Production deployment endpoint	Deploy to Production
DEV_DOMAIN	Development domain suffix	Smoke Tests
STAGING_DOMAIN	Staging domain suffix	Smoke Tests
PROD_DOMAIN	Production domain suffix	Smoke Tests

Environment URLs

Environment	Canonical Domain
Development	https://${DEV_CANONICAL_DOMAIN}
Staging	https://${STAGING_CANONICAL_DOMAIN}
Production	https://${PROD_CANONICAL_DOMAIN}

Image Tagging Strategy

Branch	Tag Format	Example	Environment
develop	develop-{commit}	develop-abc1234	Development
staging	staging-{commit}	staging-def5678	Staging
main	main-{commit}	main-ghi9012	Production

🔍 Troubleshooting

Common Issues

Pipeline Not Triggering

Symptoms: New commit pushed but no workflows start Causes:

• Workflow file syntax error
• Missing required secrets
• Branch protection rules blocking

Solutions:

1. Check workflow syntax in .github/workflows/
2. Verify all secrets are set in repository settings
3. Check Actions tab for error messages

Deployment Fails

Symptoms: Deployment workflow fails Causes:

• Webhook endpoint unreachable
• Invalid webhook signature
• k3s cluster issues
• Image not found

Solutions:

1. Check webhook handler logs: kubectl logs -n webhook-system deployment/webhook-handler
2. Verify webhook secret matches between GitHub and cluster
3. Confirm image exists in GHCR
4. Check k3s cluster health

Smoke Tests Fail

Symptoms: Tests report environment unreachable Causes:

• DNS resolution issues
• SSL certificate problems
• Application not responding
• Ingress configuration issues

Solutions:

1. Test domains manually: curl -I https://${DEV_CANONICAL_DOMAIN}
2. Check Knative service status: kubectl get ksvc -A
3. Verify ingress configuration: kubectl get ingress -A
4. Check certificate status: kubectl get certificates -A

Auto-Promotion Not Working

Symptoms: Tests pass but promotion doesn't happen Causes:

• Workflow permission issues
• No new commits to merge
• Dependency chain broken

Solutions:

1. Check workflow permissions in repository settings
2. Verify branch protection rules
3. Check workflow run logs in Actions tab
4. Manual promotion as fallback

Debug Commands

# Check all environments
kubectl get all -A | grep game-2048

# Check webhook handler
kubectl logs -n webhook-system deployment/webhook-handler --tail=50

# Check Knative services
kubectl get ksvc -A

# Check ingress
kubectl get ingress -A

# Test webhook endpoint
curl -X POST -H "Content-Type: application/json" \
  -d '{"test": "true"}' \
  https://your-webhook-url/webhook

# Check DNS resolution
dig ${DEV_CANONICAL_DOMAIN}

# Test SSL certificate
openssl s_client -servername ${DEV_CANONICAL_DOMAIN} \
  -connect ${DEV_CANONICAL_DOMAIN}:443

Emergency Procedures

Rollback Production

1. Identify last known good commit/tag
2. Run "Deploy to Production" manually
3. Specify the good image tag
4. Type "DEPLOY" to confirm

Skip Failed Tests

1. Run "Promote to Production" manually
2. Type "PROMOTE" to confirm
3. Enable "Skip tests" if staging already validated

Force Promotion

1. Manually merge branches using git
2. Push to trigger deployments
3. Monitor via "Deployment Status Check"

---

📚 Related Documentation

• Environment Setup
• Webhook Deployment
• Setup Guide
• Branching Strategy

---

Last updated: 2025-01-01 16:00:00 UTC