debug_io_direct

debug_io_direct — Use direct I/O for file access. Observed in PG16–19 Beta 3; its last measured boot default is empty string in PG19 Beta 3, with postmaster context. This is a beta-snapshot fact and can change before PostgreSQL 19 GA.

Note

Fact — official short description: “Use direct I/O for file access.”

Identity

Type , Valuestring: Upstream pg_settings type
Context , Valuepostmaster: Requires a server restart
Unit , Value—: Raw unit
Range , Value—: Raw limits in the last observed version
Enum values , Value—: — for non-enum types
Category , ValueDeveloper Options: Upstream classification
Latest boot value , Value"": empty string

Lifecycle

Fact	Value
First observed	PG16
Present in	PG16–19 Beta 3
Removed in	No
Introduction commit	`319bae9a8da6` — Rename io_direct to debug_io_direct.
Commit date	2023-05-15
Discussion	thread 1

Default history

Measured PG9.0–19 Beta 3 boot defaults
Versions	Raw `boot_val`	Unit	Human value
PG16–19 Beta 3	`""`	—	empty string

How it works

debug_io_direct requests cache-bypassing file access for selected operations: data, wal, and wal_init. PostgreSQL maps this to O_DIRECT, F_NOCACHE, or FILE_FLAG_NO_BUFFERING according to platform.

The empty string disables direct I/O. Unsupported kernels or file systems can reject startup or fail operations, and alignment constraints differ across platforms.

PostgreSQL documentation explicitly describes current nondefault use as performance-reducing developer testing. It is distinct from the PG18 asynchronous-I/O method and should not be presented as a production cache policy. Its postmaster context fixes the value at server start; changing it requires a restart.

Tuning advice

Tip

Advice. These are workload-specific starting points and must be validated with measurements.

Workload	Guidance
OLTP	Do not tune production OLTP with debug_io_direct. Enable it only for a bounded reproduction with an owner, log/disk budget, rollback condition, and evidence-capture plan; restore the default immediately afterward.
OLAP	Long analytical runs can amplify debug_io_direct’s debug overhead and artifacts. Prefer standard EXPLAIN and statistics first, and isolate any developer experiment from normal users.
Small nodes	Keep debug_io_direct at its upstream default. A small host has less spare CPU, disk, connection, and log capacity for developer instrumentation.

Pigsty

Values use the fixed 8-vCPU, 32-GiB, 100-GiB SSD fixture and render the current Pigsty templates for PG19 Beta 3; this does not assert current Pigsty support for that historical or beta release.

Template	Effective value	Versus upstream boot	Source expression
OLTP	Unmodified	—	—
OLAP	Unmodified	—	—
CRIT	Unmodified	—	—
TINY	Unmodified	—	—

Caution

Advice — pending human review. Fact from the current Pigsty template projection: OLTP: PG16–19 Beta 3 unmodified; OLAP: PG16–19 Beta 3 unmodified; CRIT: PG16–19 Beta 3 unmodified; TINY: PG16–19 Beta 3 unmodified. No Pigsty-specific rationale is inferred from an absent override.

Common pitfalls

Leaving debug_io_direct enabled after the bounded diagnostic or recovery task.
Running the experiment on the only copy of production data.
Underestimating log, core-file, temporary-file, WAL, CPU, or connection-slot amplification.
Treating a server that merely starts or completes a query as proof that data and behavior are correct.

io_method · io_max_concurrency · wal_sync_method · shared_buffers · backend_flush_after