Skip to content

Metrics ​

The Prometheus Metrics plugin lives in plugins/metrics-prometheus.

powershell
.\gradlew.bat :server:launcher:run --args="--firefly.metrics.prometheus.enabled=true"

Or enable it through the plugin list:

powershell
.\gradlew.bat :server:launcher:run --args="--firefly.plugins=metrics-prometheus"

Default endpoint:

text
http://127.0.0.1:9711/metrics

Core Metrics ​

MetricDescription
firefly_plugin_upWhether the plugin is running
firefly_jobs_totalTotal jobs
firefly_jobs_enabledEnabled jobs
firefly_nodes_onlineOnline nodes
firefly_next_fire_time_epoch_secondsNext fire time
firefly_jobs_due_totalDue jobs total
firefly_jobs_overdue_max_secondsMaximum overdue duration
firefly_schedule_delay_secondsScheduling delay
firefly_execution_duration_secondsExecution duration
firefly_executor_connectionsExecutor connections
firefly_scheduler_owned_shardsShards owned by this scheduler
firefly_database_clock_offset_millisecondsDatabase clock offset
firefly_tenant_capacity_observations_dropped_total{reason}Capacity observations dropped because buffers are full or shutdown timed out
firefly_tenant_capacity_flush_failures_total{reason}Asynchronous capacity-sample flush failures
firefly_tenant_capacity_flush_lag_secondsDelay between bucket completion and durable flush
firefly_tenant_quota_preview_total{risk}Quota previews grouped by risk level

Capacity-collection health metrics omit tenant IDs so tenant growth does not multiply Prometheus cardinality. config/prometheus/firefly-alerts.yml contains alert templates for schedule delay, overdue jobs, hot shards, outbox stalls, lease failures, database clock drift, capacity backpressure, and executor registration errors.

Released under the Apache-2.0 License.