Checkpoint Completion Target | PG Pilot PostgreSQL Monitoring

What is this issue?

PostgreSQL periodically writes dirty pages to disk (checkpoints). The
checkpoint_completion_target controls how this I/O is spread:

- 0.5 (old default): Finish writes in 50% of the checkpoint interval
- 0.9 (recommended): Spread writes across 90% of the interval

A low value causes:
- I/O spikes at checkpoint time
- Query latency jitter
- Uneven disk utilization

Bursty writes compete with query I/O

Query response times become unpredictable

SSDs prefer steady write patterns

I/O bandwidth not fully utilized

```sql
SELECT name, setting
FROM pg_settings
WHERE name = 'checkpoint_completion_target';
```

1

This is the recommended value:

ALTER SYSTEM SET checkpoint_completion_target = 0.9;
SELECT pg_reload_conf();

2

Check checkpoint frequency and timing:

SELECT * FROM pg_stat_bgwriter;

Watch for checkpoints_timed vs checkpoints_req.