Overview

This n8n workflow provides automated CI/CD testing for Kubernetes applications using KinD (Kubernetes in Docker). It creates temporary infrastructure, runs tests, and cleans up everything automatically.

---

Three-Phase Lifecycle

INIT Phase - Infrastructure Setup

• Installs dependencies (sshpass, Docker, KinD)
• Creates KinD cluster
• Installs Helm and Nginx Ingress
• Installs HAProxy for port forwarding
• Deploys ArgoCD
• Applies ApplicationSet

TEST Phase - Automated Testing

• Downloads Robot Framework test script from GitLab
• Installs Robot Framework and Browser library
• Executes automated browser tests
• Packages test results
• Sends results via Telegram

DESTROY Phase - Complete Cleanup

• Removes HAProxy
• Deletes KinD cluster
• Uninstalls KinD
• Uninstalls Docker
• Sends completion notification

---

Execution Modes

Full Pipeline Mode (progress_only = false) > Automatically progresses through all phases: INIT → TEST → DESTROY

Single Phase Mode (progress_only = true) > Executes only the specified phase and stops

---

Prerequisites

Local Environment (n8n Host)

• n8n instance version 1.0 or higher
• Community node n8n-nodes-robotframework installed
• Network access to target host and GitLab
• Minimum 4 GB RAM, 20 GB disk space

Remote Target Host

• Linux server (Ubuntu, Debian, CentOS, Fedora, or Alpine)
• SSH access with sudo privileges
• Minimum 8 GB RAM (16 GB recommended)
• 20 GB free disk space
• Open ports: 22, 80, 60080, 60443, 56443

External Services

• GitLab account with OAuth2 application
• Repository with test files (test.robot, config.yaml, demo-applicationSet.yaml)
• Telegram Bot for notifications
• Telegram Chat ID

---

Setup Instructions

Step 1: Install Community Node

1. In n8n web interface, navigate to Settings → Community Nodes
2. Install n8n-nodes-robotframework
3. Restart n8n if prompted

Step 2: Configure GitLab OAuth2

Create GitLab OAuth2 Application

1. Log in to GitLab
2. Navigate to User Settings → Applications
3. Create new application with redirect URI: https://your-n8n-instance.com/rest/oauth2-credential/callback
4. Grant scopes: read_api, read_repository, read_user
5. Copy Application ID and Secret

Configure in n8n

1. Create new GitLab OAuth2 API credential
2. Enter GitLab server URL, Client ID, and Secret
3. Connect and authorize

Step 3: Prepare GitLab Repository

Create repository structure:

your-repo/
├── test.robot
├── config.yaml
├── demo-applicationSet.yaml
└── .gitlab-ci.yml

Upload your:

• Robot Framework test script
• KinD cluster configuration
• ArgoCD ApplicationSet manifest

Step 4: Configure Telegram Bot

Create Bot

1. Open Telegram, search for @BotFather
2. Send /newbot command
3. Save the API token

Get Chat ID

For personal chat:

• Send message to your bot
• Visit: https://api.telegram.org/bot<YOUR_TOKEN>/getUpdates
• Copy the chat ID (positive number)

For group chat:

• Add bot to group
• Send message mentioning the bot
• Visit getUpdates endpoint
• Copy group chat ID (negative number)

Configure in n8n

1. Create Telegram API credential
2. Enter bot token
3. Save credential

Step 5: Prepare Target Host

Verify SSH access:

• Test connection: ssh -p <port> <username>@<host_ip>
• Verify sudo: sudo -v

The workflow will automatically install dependencies.

Step 6: Import and Configure Workflow

Import Workflow

1. Copy workflow JSON
2. In n8n, click Workflows → Import from File/URL
3. Import the JSON

Configure Parameters

Open Set Parameters node and update:

| Parameter | Description | Example | |-----------|-------------|---------| | target_host | IP address of remote host | 192.168.1.100 | | target_port | SSH port | 22 | | target_user | SSH username | ubuntu | | target_password | SSH password | your_password | | progress | Starting phase | INIT, TEST, or DESTROY | | progress_only | Execution mode | true or false | | KIND_CONFIG | Path to config.yaml | config.yaml | | ROBOT_SCRIPT | Path to test.robot | test.robot | | ARGOCD_APPSET | Path to ApplicationSet | demo-applicationSet.yaml |

> Security: Use n8n credentials or environment variables instead of storing passwords in the workflow.

Configure GitLab Nodes

For each of the three GitLab nodes:

• Set Owner (username or organization)
• Set Repository name
• Set File Path (uses parameter from Set Parameters)
• Set Reference (branch: main or master)
• Select Credentials (GitLab OAuth2)

Configure Telegram Nodes

1. Send ROBOT Script Export Pack node:

   • Set Chat ID
   • Select Credentials

2. Process Finish Report node:

   • Update chat ID in command

Step 7: Test and Execute

1. Test individual components first
2. Run full workflow
3. Monitor execution (30-60 minutes total)

---

How to Use

Execution Examples

Complete Testing Pipeline

• progress = "INIT"
• progress_only = "false"
• Flow: INIT → TEST → DESTROY

Setup Infrastructure Only

• progress = "INIT"
• progress_only = "true"
• Flow: INIT → Stop

Test Existing Infrastructure

• progress = "TEST"
• progress_only = "false"
• Flow: TEST → DESTROY

Cleanup Only

• progress = "DESTROY"
• Flow: DESTROY → Complete

Trigger Methods

1. Manual Execution

• Open workflow in n8n
• Set parameters
• Click Execute Workflow

2. Scheduled Execution

• Open Schedule Trigger node
• Configure time (default: 1 AM daily)
• Ensure workflow is Active

3. Webhook Trigger

• Configure webhook in GitLab repository
• Add webhook URL to GitLab CI

Monitoring Execution

In n8n Interface:

• View progress in Executions tab
• Watch node-by-node execution
• Check output details

Via Telegram:

• Receive test results after TEST phase
• Receive completion notification after DESTROY phase

Execution Timeline: | Phase | Duration | |-------|----------| | INIT | 15-25 minutes | | TEST | 5-10 minutes | | DESTROY | 5-10 minutes |

Understanding Test Results

After TEST phase, receive testing-export-pack.tar.gz via Telegram containing:

• log.html - Detailed test execution log
• report.html - Test summary report
• output.xml - Machine-readable results
• screenshots/ - Browser screenshots

To view:

1. Download .tar.gz from Telegram
2. Extract: tar -xzf testing-export-pack.tar.gz
3. Open report.html for summary
4. Open log.html for detailed steps

Success indicators:

• All tests marked PASS
• Screenshots show expected UI states
• No error messages in logs

Failure indicators:

• Tests marked FAIL
• Error messages in logs
• Unexpected UI states in screenshots

---

Configuration Files

test.robot

Robot Framework test script structure:

• Uses Browser library
• Connects to http://autotest.innersite
• Logs in with autotest/autotest
• Takes screenshots
• Runs in headless Chromium

config.yaml

KinD cluster configuration:

• 1 control-plane node
• 1 worker node
• Port mappings: 60080 (HTTP), 60443 (HTTPS), 56443 (API)
• Kubernetes version: v1.30.2

demo-applicationSet.yaml

ArgoCD Application manifest:

• Points to Git repository
• Automatic sync enabled
• Deploys to default namespace

gitlab-ci.yml

Triggers n8n workflow on commits:

• Installs curl
• Sends POST request to webhook

---

Troubleshooting

SSH Permission Denied

Symptoms:

Error: Permission denied (publickey,password)

Solutions:

• Verify password is correct
• Check SSH authentication method
• Ensure user has sudo privileges
• Use SSH keys instead of passwords

Docker Installation Fails

Symptoms:

Error: Package docker-ce is not available

Solutions:

• Check OS version compatibility
• Verify network connectivity
• Manually add Docker repository

KinD Cluster Creation Timeout

Symptoms:

Error: Failed to create cluster: timed out

Solutions:

• Check available resources (RAM/CPU/disk)
• Verify Docker daemon status
• Pre-pull images
• Increase timeout

ArgoCD Not Accessible

Symptoms:

Error: Failed to connect to autotest.innersite

Solutions:

• Check HAProxy status: systemctl status haproxy
• Verify /etc/hosts entry
• Check Ingress: kubectl get ingress -n argocd
• Test port forwarding: curl http://127.0.0.1:60080

Robot Framework Tests Fail

Symptoms:

Error: Chrome failed to start

Solutions:

• Verify Chromium installation
• Check Browser library: rfbrowser show-trace
• Ensure correct executablePath in test.robot
• Install missing dependencies

Telegram Notification Not Received

Symptoms:

• Workflow completes but no message

Solutions:

• Verify Chat ID
• Test Telegram API manually
• Check bot status
• Re-add bot to group

Workflow Hangs

Symptoms:

• Node shows "Executing..." indefinitely

Solutions:

• Check n8n logs
• Test SSH connection manually
• Verify target host status
• Add timeouts to commands

---

Best Practices

Development Workflow

1. Test locally first

   • Run Robot Framework tests on local machine
   • Verify test script syntax

2. Version control

   • Keep all files in Git
   • Use branches for experiments
   • Tag stable versions

3. Incremental changes

   • Make small testable changes
   • Test each change separately

4. Backup data

   • Export workflow regularly
   • Save test results
   • Store credentials securely

Production Deployment

1. Separate environments

   • Dev: Frequent testing
   • Staging: Pre-production validation
   • Production: Stable scheduled runs

2. Monitoring

   • Set up execution alerts
   • Monitor host resources
   • Track success/failure rates

3. Disaster recovery

   • Document cleanup procedures
   • Keep backup host ready
   • Test restoration process

4. Security

   • Use SSH keys
   • Rotate credentials quarterly
   • Implement network segmentation

Maintenance Schedule

| Frequency | Tasks | |-----------|-------| | Daily | Review logs, check notifications | | Weekly | Review failures, check disk space | | Monthly | Update dependencies, test recovery | | Quarterly | Rotate credentials, security audit |

---

Advanced Topics

Custom Configurations

Multi-node clusters:

• Add more worker nodes for production-like environments
• Configure resource limits
• Add custom port mappings

Advanced testing:

• Load testing with multiple iterations
• Integration testing for full deployment pipeline
• Chaos engineering with failure injection

Integration with Other Tools

Monitoring:

• Prometheus for metrics collection
• Grafana for visualization

Logging:

• ELK stack for log aggregation
• Custom dashboards

CI/CD Integration:

• Jenkins pipelines
• GitHub Actions
• Custom webhooks

---

Resource Requirements

Minimum

| Component | CPU | RAM | Disk | |-----------|-----|-----|------| | n8n Host | 2 | 4 GB | 20 GB | | Target Host | 4 | 8 GB | 20 GB |

Recommended

| Component | CPU | RAM | Disk | |-----------|-----|-----|------| | n8n Host | 4 | 8 GB | 50 GB | | Target Host | 8 | 16 GB | 50 GB |

---

Useful Commands

KinD

• List clusters: kind get clusters
• Get kubeconfig: kind get kubeconfig --name automate-tst
• Export logs: kind export logs --name automate-tst

Docker

• List containers: docker ps -a --filter "name=automate-tst"
• Enter control plane: docker exec -it automate-tst-control-plane bash
• View logs: docker logs automate-tst-control-plane

Kubernetes

• Get all resources: kubectl get all -A
• Describe pod: kubectl describe pod -n argocd <pod-name>
• View logs: kubectl logs -n argocd <pod-name> --follow
• Port forward: kubectl port-forward -n argocd svc/argocd-server 8080:80

Robot Framework

• Run tests: robot test.robot
• Run specific test: robot -t "Test Name" test.robot
• Generate report: robot --outputdir results test.robot

---

Additional Resources

Official Documentation

• n8n: https://docs.n8n.io
• KinD: https://kind.sigs.k8s.io
• ArgoCD: https://argo-cd.readthedocs.io
• Robot Framework: https://robotframework.org
• Browser Library: https://marketsquare.github.io/robotframework-browser

Community

• n8n Community: https://community.n8n.io
• Kubernetes Slack: https://kubernetes.slack.com
• ArgoCD Slack: https://argoproj.github.io/community/join-slack
• Robot Framework Forum: https://forum.robotframework.org

Related Projects

• k3s: Lightweight Kubernetes distribution
• minikube: Local Kubernetes alternative
• Flux CD: Alternative GitOps tool
• Playwright: Alternative browser automation