data_corrupted

What this means

PostgreSQL detected corruption in on-disk data pages or heap tuples. This is a critical error indicating the stored data does not pass internal consistency checks.

Why it happens

1. 1Storage hardware failure (bad disk sectors, failing SSD)
2. 2Incomplete or interrupted writes due to power failure without battery-backed write cache
3. 3Kernel or filesystem bug causing corrupt writes
4. 4RAM errors causing corrupted data before it was written to disk
5. 5pg_filedump or direct file manipulation corrupted a data page
6. 6Bug in PostgreSQL version (rare)

How to reproduce

Any read operation against a table or index with a corrupted page

SELECT * FROM my_table;  -- if a data page is corrupted

Fix 1: Restore from a known-good backup

Data corruption detected — this is the safest recovery path

-- Stop PostgreSQL, restore from backup, replay WAL to the point before corruption

Fix 2: Use zero_damaged_pages as a temporary workaround

Backup is unavailable and data loss is acceptable for some rows

SET zero_damaged_pages = on;
SELECT * FROM my_table;  -- corrupted pages return zeros instead of erroring

Fix 3: Run pg_dump to extract salvageable data

Attempting to recover as much data as possible before restore

-- pg_dump -h localhost -U postgres -d mydb -t my_table > salvage.sql

What not to do

Dangerous variant

Sources

Confidence assessment

See also