Scenario
You see wait_event_type = 'LWLock' in pg_stat_activity — lightweight locks protect PostgreSQL's internal shared memory structures. Here is how to read the specific event and what to do. Diagnose it -- Current LWLock waiters grouped by…
Investigation Path
You see wait_event_type = 'LWLock' in pg_stat_activity — lightweight locks protect PostgreSQL’s internal shared memory structures. Here is how to read the specific event and what to do.
Diagnose it
-- Current LWLock waiters grouped by event:
SELECT wait_event,
count(*) AS waiters,
count(*) FILTER (WHERE state = 'active') AS active_waiters
FROM pg_stat_activity
WHERE wait_event_type = 'LWLock'
GROUP BY wait_event
ORDER BY waiters DESC;Why it happens
Key LWLock events and their typical causes:
- BufferMapping — contention on the buffer pool hash table; usually seen with many
concurrent small-relation queries (many tables, each accessed briefly). Reduce connection count or
consolidate workloads. - WALWrite / WALBufMapping — WAL buffer contention under very
high write rates. Increasewal_buffers. - ProcArrayLock — contention on the process array (read during snapshot
acquisition). Seen with very high connection counts. A connection pooler helps. - CLogControlLock / XactSLRULock — commit log SLRU contention.
Correlates with high subtransaction usage. - LockManager — the heavyweight lock table itself is contended. Reduce the
number of concurrent lock acquisitions. - RelationMappingLock — shared catalog file mapping; usually transient.
How to fix it
There is no single fix for LWLock contention — the right response depends on the specific
event. The general strategies are: reduce connection count (use a pooler), reduce concurrency
on the specific subsystem, or tune the relevant GUC (e.g., wal_buffers for
WAL-related LWLocks).
Prevent it next time
Sample pg_stat_activity for LWLock events every 15 seconds over a 10-minute
production window. Aggregate by wait_event. If any single event accounts for
more than 5% of all backend samples, it is worth a deeper investigation of the corresponding
subsystem.
Related & next steps
Career Impact
This scenario builds production judgment and operational confidence under pressure.