Eve Deploy Debugging

What it does

This skill, Eve Deploy Debugging, assists with deploying applications and diagnosing issues within a testing or quality assurance workflow. It provides tools for managing deployments through an environment (staging), monitoring job progress, and troubleshooting failures using diagnostic commands and log analysis. The tool emphasizes controlled infrastructure changes via Terraform and uses the Eve CLI for application deployment management.

When to use it

• Deploying applications to a staging environment.
• Troubleshooting failed deployments or jobs.
• Monitoring the health of an environment during deployment.
• Diagnosing why a job won't start, is stuck, or has failed.
• Investigating system-level issues impacting deployments and jobs.

Key capabilities

• Environment Management: Create and use profiles for different environments (e.g., staging).
• Deployment via CLI: Deploy applications using eve env deploy with options to bypass pipelines (--direct) and pass inputs.
• Job Monitoring: Track job progress in real-time with eve job follow, poll status updates with eve job watch, and retrieve diagnostic information with eve job diagnose.
• Log Analysis: Access raw worker logs (eve job runner-logs), system logs (eve system logs), and thread messages (eve thread messages).
• Diagnostic Commands: Provides commands to check dependencies, environment readiness, orchestrator status, and secret lists.

Example prompts

• "Deploy the application from the main branch to the staging environment."
• "Show me the diagnostic information for job ID 12345."
• "What is the current health of the staging environment?"
• "Follow the progress of job ID 67890 in real-time."

Tips & gotchas

• Infrastructure Changes: Direct Kubernetes resource modifications (like kubectl apply) are prohibited; use the Eve CLI (eve env, eve env deploy) for deployments.
• Pipeline Bypass: Use --direct with caution, as it bypasses automated pipeline steps.
• Polling: After a pipeline deployment, poll the pipeline endpoint (/pipelines/{name}/runs/{id}) until all steps complete and active_pipeline_run is null to confirm successful completion.