Redis RDB recovery

A corrupt Redis RDB may still contain recoverable key records, but it should never replace a healthier replica or AOF without comparison.

An RDB file is a compact point-in-time snapshot of the dataset. Disk faults, incomplete copying, incompatible versions or process crashes can damage its header, serialized objects, compression blocks or checksum.

Understand the symptom

RDB recovery is a sequential parse of typed key records ending in a versioned checksum.

WHY

What is happening

The safest source may be an intact replica, a newer AOF or an earlier RDB. Recovery compares snapshot timestamps, replication IDs and offsets, AOF history and the first invalid RDB record. Partial extraction must document missing keys rather than presenting a parsed file as complete.

What you may see

Symptoms that change the recovery path.

01

Wrong RDB checksum

The file content does not match its stored integrity value.

02

Short read or unexpected EOF

The snapshot was truncated during write or transfer.

03

Unsupported RDB version

The server reading the file is incompatible with its format.

04

Redis starts empty after failure

A missing or rejected snapshot was replaced by a new dataset.

Protect the source

Actions that can make the situation worse.

01

Restart can write a new RDB

An empty or partial dataset may overwrite the best remaining snapshot.

02

Disabling checksum hides corruption

It does not make damaged serialized values trustworthy.

03

Promoting a stale replica loses newer keys

Compare offsets and persistence timestamps before choosing a source.

Do this now

Safe steps before evaluation.

01

Stop persistence writes

Disable restart loops, SAVE, BGSAVE, failover and rewrite activity.

02

Copy every persistence source

Preserve RDB, AOF, manifests, replicas and filesystem snapshots.

03

Record version and timeline

Save Redis release, replication IDs, offsets, file times and the exact load error.

04

Protect application evidence

Keep client logs or upstream sources that can validate priority keys.

NEPFOSYS process

How a data-first evaluation proceeds.

01

Persistence-source comparison

NEPFOSYS maps RDB points, AOF history, replicas and server compatibility.

02

Copy-based parsing

Readable RDB records are evaluated without changing the original file.

03

Dataset validation

Recovered keys, types, TTLs and application records are compared against the selected endpoint.

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.

Can redis-check-rdb repair the file?

It primarily validates and reports RDB structure; recovery may require an alternate persistence source or controlled extraction.

Should I disable RDB checksums?

No. That can allow damaged content to load without proving it is complete or correct.

Is AOF always newer than RDB?

Often, but not necessarily. Compare timestamps, manifests, fsync policy and replica state before selecting it.

DFW local · Nationwide mail-in

Protect the device. Start with the facts.

Start my recovery case