[
https://issues.apache.org/jira/browse/NIFI-2861?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15571999#comment-15571999
]
ASF GitHub Bot commented on NIFI-2861:
--------------------------------------
Github user markap14 commented on a diff in the pull request:
https://github.com/apache/nifi/pull/1127#discussion_r83218107
--- Diff:
nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/ControlRate.java
---
@@ -381,6 +392,14 @@ public boolean tryAdd(final long value) {
private class ThrottleFilter implements FlowFileFilter {
+ private final long flowFilesPerTrigger;
+ private final AtomicLong flowFilesFiltered = new AtomicLong(0L);
--- End diff --
This filter is not thread-safe... don't think we need an AtomicLong here.
Can just use an int.
> ControlRate should accept more than one flow file per execution
> ---------------------------------------------------------------
>
> Key: NIFI-2861
> URL: https://issues.apache.org/jira/browse/NIFI-2861
> Project: Apache NiFi
> Issue Type: Bug
> Components: Core Framework
> Affects Versions: 1.0.0, 0.7.0
> Reporter: Joe Skora
> Assignee: Joe Skora
>
> The {{ControlRate}} processor implements a {{FlowFileFilter}} that returns
> the {{FlowFileFilter.ACCEPT_AND_TERMINATE}} result if the {{FlowFile}} fits
> with the rate limit, affectively limiting it to one {{FlowFile}} per
> {{ConrolRate.onTrigger()}} invocation. This is a significant bottleneck when
> processing very large quantities of small files making it unlikely to hit the
> rate limits.
> It should allow multiple files, perhaps with a configurable maximum, per
> {{ControlRate.onTrigger()}} invocation by issuing the
> {{FlowFileFilter.ACCEPT_AND_CONTINUE}} result until the limits are reached.
> In a preliminary test this eliminated the bottleneck.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
