[ 
https://issues.apache.org/jira/browse/NIFI-4490?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16206897#comment-16206897
 ] 

Matt Burgess commented on NIFI-4490:
------------------------------------

The "Error creating binlog enrichment JDBC connection" does not stop the binlog 
client from running, so events may have been received but not processed 
(because initialization failed so onTrigger is not called). If the problem is 
fixed, presumably the user would want to keep those events and have them 
processed the next time the processor starts. However if a RESET MASTER is 
issued to the DB, then all binlog events are cleared (unbeknownst to the 
processor). The fix seems to be to not start the binlog client until the JDBC 
enrichment connection has been created successfully.

> Repetitive events detected on changing the incorrect driver name in 
> CaptureChangeMySQL
> --------------------------------------------------------------------------------------
>
>                 Key: NIFI-4490
>                 URL: https://issues.apache.org/jira/browse/NIFI-4490
>             Project: Apache NiFi
>          Issue Type: Bug
>            Reporter: Matt Burgess
>
> Followed the below steps
> 1. In CaptureChangeMySQL, set the driver name to "com.mysql.some.driver"
> 2. Create a table in the test database and start the CaptureChangeMySQL 
> processor
> 3. "Error creating binlog enrichment JDBC connection" error message is thrown 
> which is correct.
> This completes the first test. Next do the following
> 1. Correct the driver name
> 2. Reset the binlog events by running "RESET MASTER"
> 3. Clear the state of the CaptureChangeMySQL processor.
> 4. Start the CaptureChangeMySQL processor.
> Result : The Create statement which was triggered when the driver was invalid 
> is detected thrice in the tests that I carried out by CaptureChangeMySQL 
> processor
> Expected : The event should not have been detected at all since the binlogs 
> have been deleted, state reset and the database pattern hardcoded to a single 
> db



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to