Currently, when the Oracle connector throws an ORA-00310 error during mining, the connector will stop & restart as this is considered a retriable exception.

This specific error can often mean that the redo state is changing faster than Debezium can consume it, and this means that the log state in LogMiner has become outdated. In such cases, we should see if we can treat it as "inconsistent log state" and regather the logs rather than kicking this up to the runtime.