gcp-composer-troubleshooting

Start by gathering the DAG ID, run ID, and task ID, then use gcloud logging read with severity ERROR filters on the cloud_composer_environment resource. Download the actual remote DAG code from the environment's GCS bucket and correlate log tracebacks with specific code lines to identify the root cause.

Follow a structured RCA process: collect context, retrieve error logs with gcloud logging, download the running DAG code via gcloud storage, pinpoint the failing line, then document the root cause with cited log entries and code snippets. Save the findings as an RCA report file.

The remote environment runs whatever code was last synced to its GCS bucket, which may differ from your local files. Always download the remote DAG with gcloud storage cp before analysis, since the deployed code is the source of truth for the failure.

Use gcloud logging read with a filter like resource.type=cloud_composer_environment combined with the environment name, dag_id label, and severity>=ERROR. Add --limit and a table format with timestamp, severity, task_id, and textPayload for readable output.

No, it operates read-only by default and only analyzes logs and code to produce an RCA report with a proposed fix. It modifies code only when you explicitly ask it to apply the fix after the analysis is complete.