Vacuum Metrics

This page documents Prometheus metrics related to vacuum operations in Yellowbrick — both autovacuum and manual vacuum — which help manage table bloat, reclaim dead tuples, and prevent transaction ID wraparound.

Purpose

Vacuum metrics are critical for monitoring the health and efficiency of background maintenance processes. They help:

• Track how often tables are vacuumed across databases
• Distinguish between automatic and manual vacuums
• Detect stale or unvacuumed tables (e.g. through long “time since last vacuum”)
• Measure the duration and frequency of vacuum operations

These metrics support bloat management, query performance tuning, and alerting for vacuum starvation or regressions in autovacuum behavior.

Metrics

Name	Type	Freq	Labels	Version Introduced	Version Deprecated	Description
yb_active_autovacuum_count	gauge	1m	-	7.3.0	7.3.1	Number of currently active autovacuum processes
yb_active_manual_vacuum_count	gauge	1m	-	7.3.0	7.3.1	Number of currently active manual vacuum processes
yb_time_since_last_vacuum_seconds_bucket	counter	1m	le, database_oid	7.3.0	7.3.1	Cumulative count of tables since last vacuum (manual or auto), grouped by database
yb_time_since_last_vacuum_seconds_count	counter	1m	database_oid	7.3.0	7.3.1	Total number of vacuumed tables per database
yb_time_since_last_vacuum_seconds_sum	gauge	1m	database_oid	7.3.0	7.3.1	Total time since last vacuum across all tables per database
yb_vacuum_time_seconds_bucket	counter	1m	le	7.3.0	7.3.1	Cumulative count of vacuum operations running ≤ each duration bucket
yb_vacuum_time_seconds_count	counter	1m	-	7.3.0	7.3.1	Total number of currently running vacuum operations
yb_vacuum_time_seconds_sum	gauge	1m	-	7.3.0	7.3.1	Total execution time of all currently running vacuum operations