[NIFI-12526] QueryCassandra should not output FlowFiles as soon as the "available rows without fetching" is reached

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.25.0, 2.0.0-M2
• Component/s: Extensions
• Assignee: Matt Burgess
• Reporter: Matt Burgess
• Created: 19/Dec/23 21:51
• Updated: 02/Jan/24 14:19
• Resolved: 02/Jan/24 14:19

Description

NIFI-5642 introduced the Max Rows Per Flow File and Output Batch Size properties to QueryCassandra, but still uses the internal "rowsAvailableWithoutFetching" variable (whose value comes from the Cassandra ResultSet and defaults to 5000) as a trigger to stop processing rows for a FlowFile. This can cause unexpected behavior, such as getting multiple FlowFiles when only one is expected.

Also since those properties were added, the fragment.* attributes should be added and populated accordingly (such as ExecuteSQL does).

NIFI-5642 also removes the Compression Type property, which might be ok for 2.x but will cause all flows using this property in 1.x to become invalid. On the support (1.x) branch we need to add the property back in, perhaps we can keep it removed for 2.x but we'd want to remove the Cassandra Connect Points property and such to force the user to use a Cassandra Connection controller service. The reason to add the property back is if the Cassandra Contact Points property is used instead of the Cassandra Connection controller service, there is no way to set the Compression Type.

Issue Links

• is caused by: NIFI-5642 QueryCassandra processor : output FlowFiles as soon fetch_size is reached
• links to: GitHub Pull Request #8183