Skip to content

IO: DsmAllocate

Waiting for a dynamic shared memory segment to be allocated
PostgreSQL wait event dossier
ClassIO EventDsmAllocate VersionsPG 17-18 Evidence4 source location(s)

Official description

Waiting for a dynamic shared memory segment to be allocated

PG 13 PG 14 PG 15 PG 16 PG 17 PG 18

Earlier names: IO/DSMAllocate (PG 16)

Trigger mechanism

WAIT_EVENT_DSM_ALLOCATE at src/backend/storage/ipc/dsm_impl.c:366 is the grep-verified reporting path. The public identity/resource is mapped at src/backend/utils/activity/wait_event.c:604. The instrumented operation is: Waiting for a dynamic shared memory segment to be allocated. PostgreSQL reports DsmAllocate around the instrumented file or asynchronous-I/O operation named by this event. The backend resumes after the kernel, storage stack, or I/O worker completes that step.

Normal or trouble?

  • Normal: The wait is normal when the workload is expected to perform this read, write, sync, allocation, or completion operation at the observed rate.
  • Investigate: Investigate when it persists with foreground latency, many concurrent waiters, storage tail latency, throttling, or errors.

Diagnostic SQL

Sessions waiting on IO/DsmAllocate
SELECT pid, backend_type, usename, datname, application_name,
       state, now() - query_start AS query_age,
       now() - xact_start AS xact_age,
       wait_event_type, wait_event,
       pg_blocking_pids(pid) AS blocking_pids,
       left(query, 160) AS query
FROM pg_stat_activity
WHERE wait_event_type = 'IO'
  AND wait_event = 'DsmAllocate'
ORDER BY query_age DESC NULLS LAST;
Current IO cohort
SELECT wait_event, count(*) AS waiting_sessions,
       count(*) FILTER (WHERE state = 'active') AS active_waiters,
       max(now() - query_start) AS oldest_query
FROM pg_stat_activity
WHERE wait_event_type = 'IO'
GROUP BY wait_event
ORDER BY waiting_sessions DESC, wait_event;
Lock and relation context for these sessions
SELECT a.pid, l.locktype, l.mode, l.granted, l.fastpath,
       d.datname, n.nspname, c.relname,
       l.page, l.tuple, l.virtualxid, l.transactionid,
       l.classid, l.objid, l.objsubid
FROM pg_stat_activity AS a
LEFT JOIN pg_locks AS l ON l.pid = a.pid
LEFT JOIN pg_database AS d ON d.oid = l.database
LEFT JOIN pg_class AS c
  ON c.oid = l.relation
 AND l.database = (
       SELECT oid FROM pg_database WHERE datname = current_database()
     )
LEFT JOIN pg_namespace AS n ON n.oid = c.relnamespace
WHERE a.wait_event_type = 'IO'
  AND a.wait_event = 'DsmAllocate'
ORDER BY a.pid, l.granted, l.locktype, l.mode;

Response

  1. Confirm the workload phase should touch this file class.
  2. Correlate with pg_stat_io where available and per-device latency.
  3. Fix the access path, burst shape, or affected storage tier before tuning unrelated memory settings.

Source evidence