Claudio Mattioni created NIFI-15935:
---------------------------------------
Summary: AMQP processors should clean up resources after
unexpected processor failures
Key: NIFI-15935
URL: https://issues.apache.org/jira/browse/NIFI-15935
Project: Apache NiFi
Issue Type: Bug
Components: Extensions
Affects Versions: 2.9.0
Reporter: Claudio Mattioni
AMQP processors based on AbstractAMQPProcessor should clean up resources
consistently after unexpected processor failures.
Current behavior:
- When processResource() throws an Exception, AbstractAMQPProcessor logs the
failure and yields the context.
- The AMQP resource is not explicitly closed in this generic failure path.
- The ProcessSession is not explicitly rolled back in this catch block.
- If resource creation fails after the executor has been created, the AMQP
connection is closed, but the executor service is not explicitly shut down.
This can leave AMQP client resources in an inconsistent state after unexpected
processor failures and can also leave executor resources running when AMQP
resource creation fails.
Proposed changes:
- Roll back the ProcessSession after unexpected processor failures.
- Close the AMQP resource after unexpected processor failures.
- Shut down the AMQP executor service when resource creation fails.
- Update the executor thread naming pattern to use a generic AMQP client name.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
