Skip to content
BoilStream

Metrics reference

reference
TL;DR

Look up the public Prometheus metric names and types used for service monitoring. Counters reset with the process; histograms expose buckets for latency analysis. Inspect the deployed endpoint’s HELP and TYPE lines for the exact release. These measurements describe activity and pressure, not durable row counts or query-completion guarantees.

ComparisonMatch the query to the metric type
Match the query to the metric typeCounters accumulate events, gauges describe current values and histograms summarize distributions. Prometheus queries must respect those types.MeaningTypical queryCounterCumulative eventsrate() over a time rangeGaugeCurrent levelCurrent value or trendHistogramObserved distributionRate of buckets; quantileMatch the query to the metric typeCounters accumulate events, gauges describe current values and histograms summarize distributions. Prometheus queries must respect those types.MeaningTypicalqueryCounterCumulativeeventsrate() overa time rangeGaugeCurrentlevelCurrentvalue ortrendHistogramObserveddistributionRate ofbuckets;quantile
Use the documented metric names and labels. A counter value is not an instantaneous throughput rate.
Jump to reference ↓

Useful metrics

Section titled “Useful metrics”

These names are exported by the current server. Counters reset when a process restarts; use rate() for trends rather than interpreting them as durable accounting records.

Metric Type Meaning
flight_requests_total Counter Flight API requests received
flight_errors_total Counter Flight API errors
flight_request_duration_seconds Histogram Flight request duration
kafka_requests_total Counter Kafka requests received
kafka_bytes_received_total Counter Bytes received from Kafka clients
kafka_request_duration_seconds Histogram Kafka request duration
pgwire_connections Gauge Active PGWire connections
pgwire_active_queries Gauge Running PGWire queries
storage_writes_total Counter Storage writes
storage_errors_total Counter Storage errors
storage_bytes_written_total Counter Bytes written to storage
storage_operation_duration_seconds Histogram Storage operation duration
window_backpressure Gauge Ingestion backpressure, from 0 to 100
memory_current_bytes Gauge Current process memory usage

Inspect the endpoint’s HELP and TYPE lines for the metrics available in your deployed version. A request or byte counter does not establish row durability or query visibility.