PostgreSQL Data Unloader Complete Manual
PDU recovers PostgreSQL 10–18 data. This documentation is organized by failure type and includes command and parameter references.
Scope
PostgreSQL 10–18
EL 7/8/9 · Ubuntu · Debian
Linux x86 / ARM64
Offline file analysis
Recovery scenarios
Confirm the current database state, then select the matching procedure.
File recovery
Database files and directories
The instance is unavailable, or damaged pages must be bypassed for direct file parsing.
WAL recovery
Accidental DELETE and UPDATE
The database remains available and WAL covering the incident time range has been preserved.
Media recovery
Dropped and truncated tables
The table was dropped or truncated and no directly restorable backup is available.
Basic workflow
Every recovery follows the same evidence preservation, diagnosis, and copy-based workflow.
- 1
Preserve recovery materials
Copy PGDATA, tablespaces, WAL, and logs, and record the PostgreSQL version.
- 2
Identify the failure type
Separate instance damage, file damage, accidental DML, and dropped tables.
- 3
Recover from a copy
Choose the matching commands and verify row counts, parse failures, and results.
Reference
Find command syntax, parameter purpose, and version differences.
Supported data types
smallint, integer, bigint, numeric, real, double precision, moneychar, varchar, text, name, byteadate, time, timestamp, timestamptz, intervaljson, jsonb, jsonpath, xml, uuidinet, cidr, macaddr, oid, xid, tidenum, domain, tsvector, range, multirange, pgvector vectorarrays, PostGIS geometry / geography / raster