[
https://issues.apache.org/jira/browse/NIFI-4836?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16351904#comment-16351904
]
ASF GitHub Bot commented on NIFI-4836:
--------------------------------------
Github user MikeThomsen commented on a diff in the pull request:
https://github.com/apache/nifi/pull/2447#discussion_r165436273
--- Diff:
nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/QueryDatabaseTable.java
---
@@ -123,8 +124,22 @@
public static final PropertyDescriptor MAX_ROWS_PER_FLOW_FILE = new
PropertyDescriptor.Builder()
.name("qdbt-max-rows")
.displayName("Max Rows Per Flow File")
- .description("The maximum number of result rows that will be
included in a single FlowFile. " +
- "This will allow you to break up very large result
sets into multiple FlowFiles. If the value specified is zero, then all rows are
returned in a single FlowFile.")
+ .description("The maximum number of result rows that will be
included in a single FlowFile. This will allow you to break up very large "
+ + "result sets into multiple FlowFiles. If the value
specified is zero, then all rows are returned in a single FlowFile.")
+ .defaultValue("0")
--- End diff --
When I did something similar on GetMongo, I think I chose to make it
optional and have "blank" be the equivalent. Thoughts?
> Allow QueryDatabaseTables to send out batches of flow files while result set
> is being processed
> -----------------------------------------------------------------------------------------------
>
> Key: NIFI-4836
> URL: https://issues.apache.org/jira/browse/NIFI-4836
> Project: Apache NiFi
> Issue Type: Improvement
> Components: Extensions
> Reporter: Matt Burgess
> Assignee: Matt Burgess
> Priority: Major
>
> Currently QueryDatabaseTable (QDT) will not transfer the outgoing flowfiles
> to the downstream relationship(s) until the entire result set has been
> processed (regardless of whether Max Rows Per Flow File is set). This is so
> the maxvalue.* and fragment.count attributes can be set correctly for each
> flow file.
> However for very large result sets, the initial fetch can take a long time,
> and depending on the setting of Max Rows Per FlowFile, there could be a great
> number of FlowFiles transferred downstream as a large burst at the end of QDT
> execution.
> It would be nice for the user to be able to choose to have FlowFiles be
> transferred downstream while the result set is still being processed. This
> alleviates the "large burst at the end" by replacing it with smaller output
> batches during processing. The tradeoff will be that if an Output Batch Size
> is set, then the maxvalue.* and fragment.count attributes will not be set on
> the outgoing flow files.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)