out_of_memory out of memory — 53200

Symptoms

A backend could not allocate the memory it needed for an operation. PostgreSQL raises SQLSTATE 53200 (out_of_memory) and usually logs a memory-context dump showing where memory went.

• Often during large sorts, hash joins, or huge result materialization.
• The log includes a TopMemoryContext breakdown.
• May be a single oversized query rather than a global shortage.

What the server log shows

ERROR:  out of memory
DETAIL:  Failed on request of size 1073741824 in memory context "ExecutorState".

Why PostgreSQL raises this — what the manual says

Section 19.4.1 Memory (work_mem):

“Sets the base maximum amount of memory to be used by a query operation (such as a sort or hash table) before writing to temporary disk files.”

Each backend allocates from process memory governed partly by work_mem per operation. A query that demands more than the OS can give (or an over-large work_mem multiplied across operations) exhausts memory and aborts with 53200.

Common causes

• A single query with a huge sort, hash join, or result set.
• An over-large work_mem multiplied across many concurrent operations.
• OS memory pressure, tight cgroup limits, or disabled overcommit.

Relevant GUC parameters

How to fix it

1. Lower a too-aggressive work_mem, or set it per-session for big queries only.
2. Reduce result size with filtering/pagination; avoid materializing huge sets client-side.
3. Check OS memory and overcommit settings; ensure swap and limits are sane.

Diagnostic query

-- Inspect current memory-related settings
SELECT name, setting, unit FROM pg_settings
WHERE name IN ('work_mem','maintenance_work_mem','hash_mem_multiplier','shared_buffers');

Related & next steps

• SQLSTATE 53200 — Out Of Memory (full reference)
• Message: temporary file size exceeds temp_file_limit
• Glossary: work_mem

Reference: PostgreSQL 18 Section 20.4 “Resource Consumption”.