composer/_ Log Check Step
Base class for performing a log query and reporting results.
All steps available in composer
Base class for performing a log query and reporting results.
Check for Redis authentication failures.
Check for Celery broker publishing timeouts.
Check the deleted airflow-scheduler deployment via Cloud Logging.
Check for worker pod evictions due to ephemeral storage exhaustion.
Check for Firewall rules blocking Redis traffic.
Check for gcs-syncd container hitting its ephemeral limit.
Check if worker pods were preempted by GKE.
Checks for failed Liveness probes on worker pods.
Check for Redis OOM on Container.
Check for Redis OOM on Node.
Check the scheduler exceeded 5,000 runs via Cloud Logging.
Check for worker pods failing to schedule due to quota issues.
Check for Out-of-Memory (OOM) events in worker pods.
Check for Zombie tasks or SIGKILL signals in environment logs.
Checks the deleted environment service account via Cloud Logging.
Check the disabled/deleted environment service account via Cloud Logging.
End step for the runbook.
Check if the GKE cluster has IP exhaustion.
Composite step for GKE and Pod stability investigations.
Base class for performing a log query and reporting results for a given step.
Check if Cloud Monitoring reports the environment as unhealthy.
Check the stability of the Composer environment’s Monitoring Pod via cloud logging.
Check if the Composer environment’s plugins directory exists.
End step for Redis troubleshooting.
Validates environment existence and extracts GKE metadata.
End step for Airflow worker restart runbook.
Check the scheduler CPU utilization via Monitoring.
Check the scheduler health via Monitoring.
Validates environment existence and extracts GKE metadata.
Check the liveness probe logs for the scheduler via Cloud Logging.
Check the status of the Composer environment’s service accounts.
Check the shared VPC org policy constraint via Cloud Logging.
Check the stability of the Composer environment’s Snapshot Load via cloud logging.
End of the runbook.
Start step for the unhealthy Composer environment diagnostic tree.
Check the CPU utilization of the Composer environment’s web server.
Validates environment existence and extracts GKE metadata.