Yandex Monitoring metric reference

Written by

Updated at April 28, 2025

This section describes Managed Service for Apache Airflow™ metrics delivered to Monitoring.

The name label contains the metric name.

Labels shared by all Managed Service for Apache Airflow™ metrics:

Label	Value
service	Service ID: `managed-airflow`
cluster_id	Cluster ID

Service metrics:

Name Type, units	Description Labels
`airflow.dag_processing.file_path_queue_size_value` `GAUGE`, count	Number of DAG files to include in the next scan
`airflow.dag_processing.file_path_queue_update_count.value` `COUNTER`, count	Number of file system scans that resulted in all existing DAGs being enqueued.
`airflow.dag_processing.import_errors.value` `GAUGE`, count	Number of DAG file parsing errors
`airflow.dag_processing.total_parse_time.value` `GAUGE`, seconds	Time taken to scan and import all DAG files to include the next scan
`airflow.dagbag_size.value` `GAUGE`, count	Number of DAG files found in the latest DAGBag scan based on scheduler configuration
`airflow.dataset.orphaned.value` `GAUGE`, count	Number of datasets labeled as orphans since they are no longer linked to DAG schedule parameters or task outputs.
`airflow.job_start.value` `COUNTER`, count	Number of runs of different types of tasks, such as `SchedulerJob`, `LocalTaskJob`, etc.
`airflow.pool.deferred_slots.default_pool.value` `GAUGE`, count	Number of deferred slots in default pools
`airflow.pool.deferred_slots.value` `GAUGE`, count	Number of deferred slots in all pools. This metric features the `pool_name` label containing the pool name.
`airflow.pool.open_slots.default_pool.value` `GAUGE`, count	Number of open slots in default pools
`airflow.pool.open_slots.value` `GAUGE`, count	Number of open slots in all pools. This metric features the `pool_name` label containing the pool name.
`airflow.pool.queued_slots.default_pool.value` `GAUGE`, count	Number of queued slots in default pools
`airflow.pool.queued_slots.value` `GAUGE`, count	Number of queued slots in all pools. This metric features the `pool_name` label containing the pool name.
`airflow.pool.running_slots.default_pool.value` `GAUGE`, count	Number of running slots in default pools
`airflow.pool.running_slots.value` `GAUGE`, count	Number of running slots in all pools. This metric features the `pool_name` label containing the pool name.
`airflow.scheduler.critical_section_duration.50_percentile` `TIMING`, milliseconds	Time spent on the critical section of the scheduler loop¹, 50th percentile
`airflow.scheduler.critical_section_duration.95_percentile` `TIMING`, milliseconds	Time spent on the critical section of the scheduler loop¹, 95th percentile
`airflow.scheduler.critical_section_duration.99_percentile` `TIMING`, milliseconds	Time spent on the critical section of the scheduler loop¹, 99th percentile
`airflow.scheduler.critical_section_duration.count` `TIMING`, count	Number of time measurements for the critical section of the scheduler loop¹
`airflow.scheduler.critical_section_duration.lower` `TIMING`, milliseconds	Minimum time spent on the critical section of the scheduler loop¹
`airflow.scheduler.critical_section_duration.mean` `TIMING`, milliseconds	Average time spent on the critical section of the scheduler loop¹
`airflow.scheduler.critical_section_duration.median` `TIMING`, milliseconds	Median time spent on the critical section of the scheduler loop¹
`airflow.scheduler.critical_section_duration.stddev` `TIMING`, milliseconds	Standard deviation of the time spent on the critical section of the scheduler loop¹
`airflow.scheduler.critical_section_duration.sum` `TIMING`, milliseconds	Total time spent on the critical section of the scheduler loop¹
`airflow.scheduler.critical_section_duration.upper` `TIMING`, milliseconds	Maximum time spent on the critical section of the scheduler loop¹
`airflow.scheduler.critical_section_query_duration.50_percentile` `TIMING`, milliseconds	Time spent running a query in the critical section¹, 50th percentile
`airflow.scheduler.critical_section_query_duration.95_percentile` `TIMING`, milliseconds	Time spent running a query in the critical section¹, 95th percentile
`airflow.scheduler.critical_section_query_duration.99_percentile` `TIMING`, milliseconds	Time spent running a query in the critical section¹, 99th percentile
`airflow.scheduler.critical_section_query_duration.count` `TIMING`, count	Number of query run time measurements in the critical section¹
`airflow.scheduler.critical_section_query_duration.lower` `TIMING`, milliseconds	Minimum time spent running a query in the critical section¹
`airflow.scheduler.critical_section_query_duration.mean` `TIMING`, milliseconds	Average time spent running a query in the critical section¹
`airflow.scheduler.critical_section_query_duration.median` `TIMING`, milliseconds	Median time spent running a query in the critical section¹
`airflow.scheduler.critical_section_query_duration.stddev` `TIMING`, milliseconds	Standard deviation of the time spent running a query in the critical section¹
`airflow.scheduler.critical_section_query_duration.sum` `TIMING`, milliseconds	Total time spent running a query in the critical section¹
`airflow.scheduler.critical_section_query_duration.upper` `TIMING`, milliseconds	Maximum time spent running a query in the critical section¹
`airflow.scheduler.heartbeat.value` `COUNTER`, count	Number of heartbeat messages from the scheduler indicating its active state
`airflow.scheduler.load_serializers.50_percentile` `TIMING`, milliseconds	Time spent serializing data in the scheduler², 50th percentile
`airflow.scheduler.load_serializers.95_percentile` `TIMING`, milliseconds	Time spent serializing data in the scheduler², 95th percentile
`airflow.scheduler.load_serializers.99_percentile` `TIMING`, milliseconds	Time spent serializing data in the scheduler², 99th percentile
`airflow.scheduler.load_serializers.count` `TIMING`, count	Number of time measurements for data serialization in the scheduler²
`airflow.scheduler.load_serializers.lower` `TIMING`, milliseconds	Minimum time spent serializing data in the scheduler²
`airflow.scheduler.load_serializers.mean` `TIMING`, milliseconds	Average time spent serializing data in the scheduler²
`airflow.scheduler.load_serializers.median` `TIMING`, milliseconds	Median time spent serializing data in the scheduler²
`airflow.scheduler.load_serializers.stddev` `TIMING`, milliseconds	Standard deviation of the time spent serializing data in the scheduler²
`airflow.scheduler.load_serializers.sum` `TIMING`, milliseconds	Total time spent serializing data in the scheduler²
`airflow.scheduler.load_serializers.upper` `TIMING`, milliseconds	Maximum time spent serializing data in the scheduler²
`airflow.scheduler.orphaned_tasks.adopted.value` `COUNTER`, count	Number of tasks adopted by the scheduler that were marked as orphans.
`airflow.scheduler.orphaned_tasks.cleared.value` `COUNTER`, count	Number of tasks cleared by the scheduler that were marked as orphans.
`airflow.scheduler.scheduler_loop_duration.50_percentile` `TIMING`, milliseconds	Time spent running one scheduler loop, 50th percentile
`airflow.scheduler.scheduler_loop_duration.95_percentile` `TIMING`, milliseconds	Time spent running one scheduler loop, 95th percentile
`airflow.scheduler.scheduler_loop_duration.99_percentile` `TIMING`, milliseconds	Time spent running one scheduler loop, 99th percentile
`airflow.scheduler.scheduler_loop_duration.count` `TIMING`, count	Number of time measurements for one scheduler loop
`airflow.scheduler.scheduler_loop_duration.lower` `TIMING`, milliseconds	Minimum time spent running one scheduler loop
`airflow.scheduler.scheduler_loop_duration.mean` `TIMING`, milliseconds	Average time spent running one scheduler loop
`airflow.scheduler.scheduler_loop_duration.median` `TIMING`, milliseconds	Median time spent running one scheduler loop
`airflow.scheduler.scheduler_loop_duration.stddev` `TIMING`, milliseconds	Standard deviation of time spent running one scheduler loop
`airflow.scheduler.scheduler_loop_duration.sum` `TIMING`, milliseconds	Total time spent running one scheduler loop
`airflow.scheduler.scheduler_loop_duration.upper` `TIMING`, milliseconds	Maximum time spent running one scheduler loop
`airflow.scheduler.tasks.executable.value` `GAUGE`, count	Number of tasks that are ready to run based on pool limitations, DAG concurrency, worker status, and priority levels.
`airflow.scheduler.tasks.starving.value` `GAUGE`, count	Number of tasks that cannot be scheduled due to no free slots in the pool.
`CeleryExecutor` `GAUGE`, count	Number of Celery workers at different task running stages. The `status` label value means the number of workers with the following statuses: `open`: Workers ready to take on tasks. `queued`: Workers with tasks in queue but not yet running. `running`: Workers running tasks.

¹ Only a single scheduler can enter this loop at a time.
² Data serialization for exchange between tasks and for web server and scheduler security.

Yandex Monitoring metric reference

Was the article helpful?