Amazon DocumentDB adds retryable writes in engine 8.0.2

Amazon DocumentDB supports retryable writes starting with engine version 8.0.2, AWS announced on September 28, 2026. The capability lets MongoDB-compatible drivers automatically retry eligible write operations once after a transient error, such as a network interruption or primary failover.

Availability and version boundary

According to AWS, retryable writes are available on DocumentDB 8.0.2 instance-based and serverless clusters, in all AWS Regions where DocumentDB 8.0 is available. Engine versions earlier than 8.0.2 remain unsupported and require retryable writes to be disabled.

Most current MongoDB-compatible drivers enable retryable writes by default, AWS says. Applications connecting to 8.0.2 no longer need the retryWrites=false connection-string setting to avoid errors. AWS also says enabling the capability requires no additional code changes.

Retry limits

This is a bounded retry mechanism, not an unlimited retry loop. The driver retries an eligible operation only once. AWS explicitly excludes updateMany and deleteMany, so applications using those multi-document commands should not assume that retryable writes cover them.

Rollout checks

For teams operating mixed engine versions, review configuration by destination cluster rather than changing every connection string at once. Check both connection strings and driver settings before removing the disabling flag: the new capability does not change the requirement to disable retryable writes on older engines.

In a test environment, exercise an eligible write during a transient failure and verify how the application handles a failure that persists after the driver’s single retry. Keep application-level error handling, and assess commands outside the retryable-write scope separately. The feature can simplify handling of eligible transient failures, but should not be treated as a guarantee that every write will succeed.