-
Enhancement
-
Resolution: Done
-
Minor
-
2.5.0.Alpha2
-
None
-
False
-
None
-
False
Feature request or enhancement
For feature requests or enhancements, provide this information, please:
Which use case/requirement will be addressed by the proposed feature?
In the config for Oracle Connector, there is option:
| log.mining.transaction.retention.ms | 0 | Positive integer value that specifies the number of milliseconds to retain long running transactions between redo log switches. When set to 0, transactions are retained until a commit or rollback is detected. By default, the LogMiner adapter maintains an in-memory buffer of all running transactions. Because all of the DML operations that are part of a transaction are buffered until a commit or rollback is detected, long-running transactions should be avoided in order to not overflow that buffer. Any transaction that exceeds this configured value is discarded entirely, and the connector does not emit any messages for the operations that were part of the transaction. |
However this is not implemented for ISPN event processor and thus this option does nothing when ISPN is enabled.
Having it available for ISPN based connector main benefits will be monitoring - visibility into potential problem happening. Right now it is not easily visible that there is transaction stuck for e.g. 12 hours (our preferred configuration value). There were cases when some logminer events were missed, transaction stayed opened and it would be like that until ISPN cache refresh.
This should be enabled as part of this ticket.
Implementation ideas (optional)
It should be possible to move the abandoned transaction handling to common abstract class (already exists)
