Metrics via Prometheus

Available on: Enterprise Edition

Enable metrics endpoint

By default the Prometheus metrics endpoint is disabled. To enable Prometheus metrics, configure the environment variable below:

HASURA_GRAPHQL_ENABLED_APIS=metadata,graphql,config,metrics

Secure the Prometheus metrics endpoint with a secret:

HASURA_GRAPHQL_METRICS_SECRET=<secret>

curl 'http://127.0.0.1:8080/v1/metrics' -H 'Authorization: Bearer <secret>'

Note

• The metrics endpoint should be configured with a secret to prevent misuse and should not be exposed over the internet.

Metrics exported

Name	Description	Type	Labels
hasura_http_connections	Current number of active HTTP connections (excluding WebSocket connections)	Gauge	none
hasura_websocket_connections	Current number of active WebSocket connections	Gauge	none
hasura_active_subscriptions	Current number of active subscriptions	Gauge	none
hasura_graphql_requests_total	Number of GraphQL requests received (excluding subscriptions)	Counter	• "operation_type": query|mutation|unknown • The "unknown" operation type will be returned for queries that fail authorization, parsing, or certain validations • "response_status": success|failed
hasura_graphql_execution_time_seconds	Execution time of successful GraphQL requests (excluding subscriptions)	Histogram Buckets: 0.01, 0.03, 0.1, 0.3, 1, 3, 10	• "operation_type": query|mutation
hasura_event_queue_time_seconds	Queue time for an event already in the processing queue	Histogram Buckets: 0.01, 0.03, 0.1, 0.3, 1, 3, 10, 30, 100	none
hasura_event_fetch_time_per_batch_seconds	Latency of fetching a batch of events	Histogram Buckets: 0.0001, 0.0003, 0.001, 0.003, 0.01, 0.03, 0.1, 0.3, 1, 3, 10	none
hasura_event_webhook_processing_time_seconds	The time between when an HTTP worker picks an event for delivery to the time its response is updated in the DB	Histogram Buckets: 0.01, 0.03, 0.1, 0.3, 1, 3, 10	none
hasura_event_trigger_http_workers	Current number of active event trigger HTTP workers	Gauge	none

Note

The GraphQL request execution time:

• Uses wall-clock time, so it includes time spent waiting on I/O.
• Includes authorization, parsing, validation, planning, and execution (calls to databases, remote schemas).