Technologies/PostgreSQL/postgresql.bgwriter.checkpoint.requested
PostgreSQLPostgreSQLMetric

postgresql.bgwriter.checkpoint.requested

Requested checkpoints
Dimensions:None
Available on:DatadogDatadog (1)Native (1)OpenTelemetryOpenTelemetry (1)PrometheusPrometheus (1)
Interface Metrics (4)
DatadogDatadog
The number of requested checkpoints that were performed.
Dimensions:None
Native
Number of requested checkpoints
Dimensions:None
OpenTelemetryOpenTelemetry
The number of checkpoints performed.
Dimensions:None
PrometheusPrometheus
Number of requested checkpoints that have been performed
Dimensions:None
Sources
postgresql.bgwriter.checkpoints_requestedgithub.com
checkpoints_reqgithub.com
postgresql.bgwriter.checkpoint.countgithub.com
stat_bgwriter_checkpoints_req_totalgithub.com

Technical Annotations (11)

Configuration Parameters (6)
checkpoint_completion_targetrecommended: 0.9
spreads checkpoint I/O over 90% of interval
checkpoint_timeoutrecommended: 15min
maximum time between checkpoints
checkpoint_warningrecommended: 30s
log warning if checkpoints occur more frequently
bgwriter_delayrecommended: 200ms
background writer wake interval to spread writes
max_wal_sizerecommended: 2-4GB for high-write workloads
WAL size threshold for checkpoint trigger
min_wal_sizerecommended: 1GB (OLTP) or 4GB (OLAP)
minimum WAL to maintain
CLI Commands (1)
SELECT * FROM pg_stat_bgwriter;diagnostic
Technical References (4)
checkpointconceptbgwritercomponentpg_stat_bgwritercomponentWALcomponent
Related Insights (5)
Checkpoint spikes cause I/O contention and query latencywarning
▸
Undersized max_wal_size triggers frequent checkpoints and I/O spikeswarning
▸
Frequent checkpoints cause performance spikes during write-heavy workloadswarning
▸
Checkpoint and WAL configuration mismatches cause I/O spikeswarning
▸
Checkpoint frequency exceeds sustainable ratewarning

Frequent checkpoints cause I/O spikes and performance hiccups. When requested checkpoints (checkpoint_req) significantly exceed timed checkpoints (checkpoint_timed), it indicates max_wal_size is too low or write workload is too high.

▸