I am experiencing a behavior that is not what I expected when I initially configured my DLQ.
Basically I have error handling enabled (errors.tolerance: all) in the DLQ with the expectation that "connectivity" issues would not be errors that the connector can handle.
So basically for example if I have tablespace issues in the JDBC target all my messages to the DLQ. This for me is not ideal :D and I would much rather that my connector fails in this case.
Going back to my original questions.
Is there a way I can configure the JDBC sink so that if there is a connectivity issue (for example a table space is full) the connector tasks go into error status, but if there is an issue in the deserializing/transformation chain they go into the DLQ?
Hi all,
I am experiencing a behavior that is not what I expected when I initially configured my DLQ.
Basically I have error handling enabled (errors.tolerance: all) in the DLQ with the expectation that "connectivity" issues would not be errors that the connector can handle. So basically for example if I have tablespace issues in the JDBC target all my messages to the DLQ. This for me is not ideal :D and I would much rather that my connector fails in this case.
I've found this discussion: https://github.com/confluentinc/kafka-connect-jdbc/issues/721 And this discussion: https://forum.confluent.io/t/dead-letter-queue-support-for-jdbc-connector-exceptions/499 but I could not find a way to address/understand how this can be configured.
I also tried the forums but I was not so lucky so I'll try here also. https://forum.confluent.io/t/on-sink-connector-is-it-possible-to-avoid-having-messages-going-to-dlq-when-target-systems-are-un-reachable/7094
I also ready this topic but I did not fully understand if this is a configuration or just an explanation on how this works https://www.javierholguera.com/2020/04/13/error-tolerance-in-kafka-connect-i/
Going back to my original questions. Is there a way I can configure the JDBC sink so that if there is a connectivity issue (for example a table space is full) the connector tasks go into error status, but if there is an issue in the deserializing/transformation chain they go into the DLQ?