ORA-01555 Debezium Snapshot Error During Production Deployment

[Ramesh K]

Hi Chris,

During the deployment of 160 tables (~200 GB) in production using initial snapshot mode on Debezium 2.5, we encountered the following error:

ORA-01555: snapshot too old: rollback segment too small
java.sql.SQLRecoverableException: Closed Connection
Connector restarted due to RetriableException

Based on the Debezium FAQ, this is due to Oracle undo tablespace limitations during long-running queries.

Proposed Approach (without increasing UNDO retention):

• Deploy the connector with snapshot.mode=schema_only to skip the full initial snapshot.
• Trigger incremental snapshots via the signaling table

Request for Guidance

Could you please confirm the exact steps we should follow for this approach in production?

• Should we first deploy with schema_only mode and then trigger incremental snapshots via the signaling table?
• Any additional best practices you recommend for large datasets?
  
  
  
  

Regards,

Ramesh

[Chris Cranford]

Hi Ramesh -

Assuming you have correctly configured the connector for signaling and incremental snapshots, yes using `schema_only` snapshot mode paired with incremental snapshots is the one and only way to avoid ORA-01555 for large tables but limited undo retention.

-cc