What problem does it solve? When an Airflow DAG fails in a Cloud Composer environment, finding the actual cause requires digging through remote logs and comparing them against the code actually deployed in the environment. This Skill guides a methodical, evidence-based Root Cause Analysis (RCA) so you fix the real problem instead of guessing. ## Core Features & Use Cases - Evidence-Based Log Analysis: Uses gcloud logging read with severity, DAG ID, and task filters to retrieve the exact errors and stack traces behind a failure. - Remote Code Verification: Downloads the actual DAG code from the Composer GCS bucket via gcloud storage to confirm the remote version matches local files before analyzing. - RCA Report Generation: Correlates log entries with specific code lines, explains the root cause, proposes a fix, and produces a saved RCA report. - Use Case: Your daily_sales_agg DAG failed yesterday at 2pm. The Skill pulls the error logs, finds a KeyError: 'region' in the process_sales task, pinpoints line 45 in the remote DAG code, and recommends adding record.get('region', 'unknown'). ## Quick Start Ask the agent to troubleshoot why your Composer DAG failed by providing the DAG ID, environment name, and approximate failure time.
