[
https://issues.apache.org/jira/browse/NIFI-5228?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16487180#comment-16487180
]
ASF GitHub Bot commented on NIFI-5228:
--------------------------------------
Github user markap14 commented on a diff in the pull request:
https://github.com/apache/nifi/pull/2733#discussion_r190229672
--- Diff:
nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/ListFile.java
---
@@ -255,43 +262,47 @@ public void onScheduled(final ProcessContext context)
{
final Path absPath = filePath.toAbsolutePath();
final String absPathString = absPath.getParent().toString() +
File.separator;
+ final DateFormat formatter = new
SimpleDateFormat(FILE_MODIFY_DATE_ATTR_FORMAT, Locale.US);
--- End diff --
We don't recommend ever using ThreadLocal for processors in NiFi. This is
because each time a Processor is run, it is done in a potentially different
thread. For a large deployment you could have hundreds of threads, and the
threads stay around for the life of the instance, so the cleanup is a little
awkward. The pattern that we commonly follow is to use a BlockingQueue and poll
from that, then create if necessary, and put back. I.e., a simple Object Pool.
And I did consider it but decided that the complexity that it adds to the code
was not worth it, given the cost of creating the DateFormat.
> Allow user to choose whether or not to add File Attributes as FlowFile
> Attributes when using ListFile
> -----------------------------------------------------------------------------------------------------
>
> Key: NIFI-5228
> URL: https://issues.apache.org/jira/browse/NIFI-5228
> Project: Apache NiFi
> Issue Type: Improvement
> Components: Extensions
> Reporter: Mark Payne
> Assignee: Mark Payne
> Priority: Major
> Fix For: 1.7.0
>
>
> The FetchFile processor adds several FlowFIle attributes such as the file's
> owner, last accessed time, creation time, etc. While these certainly can be
> useful pieces of information and do serve a purpose, they can be expensive to
> determine in some configurations. In my use case, I have an Azure File Store
> mounted to an Ubuntu system with CIFS using SMB 3.0. The remote directory
> that I am listing has 7,000-8,000 files and takes about 3 minutes to perform
> the listing with ListFile.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
