[
https://issues.apache.org/jira/browse/NIFI-5828?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16689819#comment-16689819
]
Colin Dean commented on NIFI-5828:
----------------------------------
I've tracked this down.
The number of rows put into {{executesql.row.count}} comes from {{nrOfRows}},
an AtomicLong that's created early and set
[here|https://github.com/apache/nifi/blob/102a5288efb2a22cd54815dd7331dfc5826aee91/nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/AbstractExecuteSQL.java#L270]::
{code:java}
nrOfRows.set(sqlWriter.writeResultSet(resultSet, out, getLogger(), null));
{code}
and the attribute is set
[here|https://github.com/apache/nifi/blob/102a5288efb2a22cd54815dd7331dfc5826aee91/nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/AbstractExecuteSQL.java#L280]
10 lines later:
{code:java}
attributesToAdd.put(RESULT_ROW_COUNT, String.valueOf(nrOfRows.get()));
{code}
It's pretty clear that {{SqlWriter#writeResultSet}} returns the number of rows
it wrote. Normally, this would be the whole set. The number of max rows is
passed to the DefaultAvroSqlWriter
[here|https://github.com/apache/nifi/blob/102a5288efb2a22cd54815dd7331dfc5826aee91/nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/ExecuteSQL.java#L145].
I think that the {{executesql.row.count}} should be in the same boat as the
fragment count and added [near the end of the
onTrigger|https://github.com/apache/nifi/blob/102a5288efb2a22cd54815dd7331dfc5826aee91/nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/AbstractExecuteSQL.java#L334].
> ExecuteSQL executesql.row.count meaning changes when Max Rows Per Flow File > > 0
> -------------------------------------------------------------------------------
>
> Key: NIFI-5828
> URL: https://issues.apache.org/jira/browse/NIFI-5828
> Project: Apache NiFi
> Issue Type: Bug
> Components: Extensions
> Affects Versions: 1.8.0
> Environment: Linux, MSSQL 2016
> Reporter: Colin Dean
> Priority: Major
> Labels: regression
>
> When *Max Rows Per Flow File* ({{esql-max-rows}}) is set greater than 0 to
> enable it, the {{executesql.row.count}} attribute on the resulting FlowFiles
> is not the number of rows in the result set but rather the number of rows in
> the FlowFile.
> This is a deviation from documented behavior, which is "Contains the number
> of rows returned in the select query".
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
