io_error could not read block N in file “…” — 58030

Symptoms

PostgreSQL read fewer bytes than expected for a data block, indicating a truncated/short read. PostgreSQL raises SQLSTATE 58030 (io_error).

• A block read returned fewer bytes than a full page.
• Often a truncated file or failing storage.
• Signals possible corruption or hardware trouble.

What the server log shows

ERROR:  could not read block 12345 in file "base/16384/24576": read only 0 of 8192 bytes

Why PostgreSQL raises this — what the manual says

As Section 28.6 WAL Internals explains:

A read of a specific data or WAL block failed, typically signalling underlying storage corruption or an I/O fault; the block could not be retrieved from the relation file at the expected offset.

Data files are read in fixed 8 KB pages. A short read (fewer bytes than a page) means the file is truncated or the storage failed, so PostgreSQL reports 58030.

Common causes

• A truncated relation file (e.g. from a botched restore/copy).
• Failing disk or filesystem returning short reads.
• Out-of-band file tampering in the data directory.

How to fix it

1. Investigate storage/filesystem health immediately.
2. Restore the affected relation/database from a known-good backup.
3. Never manually edit files in the data directory.

Related & next steps

• SQLSTATE 58030 — I/O Error (full reference)
• Message: could not write block
• Glossary: Page

Reference: PostgreSQL 18 Section 30.5 “WAL Internals”.