PostgreSQL transaction metadata recovery

PostgreSQL pg_multixact corruption affects row-lock and visibility history, so resetting it blindly can expose inconsistent tuples.

The pg_multixact directory records multi-transaction membership used for shared row locks. Missing segments, wraparound conditions or storage corruption can prevent startup, VACUUM or normal queries even when relation files still exist.

Understand the symptom

MultiXact IDs connect tuple headers to offset and member records in cluster-wide state.

WHY

What is happening

A recovery plan distinguishes missing pg_multixact files from damaged heap pages, WAL gaps, an interrupted upgrade and underlying storage failure. Because tuple lock and update metadata is distributed across relation pages and pg_multixact segments, replacing or resetting one side without the other can make rows appear valid or obsolete incorrectly.

What you may see

Symptoms that change the recovery path.

01

Could not access MultiXact member

A required member segment is missing or unreadable.

02

Startup or VACUUM stops

The server cannot advance or validate MultiXact state safely.

03

Wraparound warnings escalate

The cluster cannot freeze tuples before old MultiXact IDs become unsafe.

04

Errors follow a storage incident

Lost files, partial restores or filesystem damage affected cluster metadata.

Protect the source

Actions that can make the situation worse.

01

A reset can hide inconsistencies

Changing counters does not reconstruct missing membership records referenced by heap tuples.

02

Foreign files are not compatible

Another cluster's pg_multixact contents do not match this cluster's tuple history.

03

Continued writes expand uncertainty

New transactions and vacuum activity alter the evidence needed to map affected rows.

Do this now

Safe steps before evaluation.

01

Stop the cluster cleanly if possible

Prevent restart loops, autovacuum, application writes and automatic failover.

02

Record the precise error

Save segment names, MultiXact IDs, relation context, logs, version and recent maintenance history.

03

Preserve the full cluster

Copy PGDATA, tablespaces, WAL, archives, configuration, extensions and base backups together.

04

Investigate storage

Capture filesystem, kernel, RAID and SMART events before replacing files or repairing volumes.

NEPFOSYS process

How a data-first evaluation proceeds.

01

Metadata scope analysis

NEPFOSYS maps the missing segments, cluster counters, affected relations and available WAL or backups.

02

Copy-based recovery testing

Restore, replay, controlled extraction and narrow metadata repair options are tested away from production.

03

Logical validation

Priority schemas and rows are exported and checked for visibility and application consistency.

A clear next step

Preserve the original device before another repair or recovery attempt.

Start a case with the exact symptoms, incident timeline, prior attempts, encryption details, and the files that matter most. NEPFOSYS will route the case to the appropriate DFW intake or nationwide mail-in path.

Start with a case review

Questions, answered

What customers ask before starting.

What does pg_multixact store?

It stores groups of transaction IDs used when multiple transactions lock or update the same tuple.

Will pg_resetwal fix pg_multixact corruption?

It can reset control values in limited emergency cases, but it does not recreate missing member data and can leave logical inconsistencies.

Can I restore only the pg_multixact folder?

Only from a recovery source proven consistent with the same cluster checkpoint and WAL history; mixing states can make corruption worse.

DFW local · Nationwide mail-in

Protect the device. Start with the facts.

Start my recovery case