You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Zsihovszki Krisztina (Jira)" <ji...@apache.org> on 2022/08/29 14:51:00 UTC

[jira] [Assigned] (NIFI-10404) TailFile processor's persistent state not cleaned up

     [ https://issues.apache.org/jira/browse/NIFI-10404?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Zsihovszki Krisztina reassigned NIFI-10404:
-------------------------------------------

    Assignee: Zsihovszki Krisztina

> TailFile processor's persistent state not cleaned up
> ----------------------------------------------------
>
>                 Key: NIFI-10404
>                 URL: https://issues.apache.org/jira/browse/NIFI-10404
>             Project: Apache NiFi
>          Issue Type: Bug
>            Reporter: Zsihovszki Krisztina
>            Assignee: Zsihovszki Krisztina
>            Priority: Major
>
> TailFile processor's persistent state map is not maintained properly, entries are added to this map but the references to non existent files are not moved from the map.
> At NiFi restart the states map (storing the TailFileObjects) is empty and it is [filled|https://github.com/hortonworks/nifi/blob/CFM-2.0.4.0/nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/TailFile.java#L360] with entries based on the persistent state map. 
> Each TailFileObject contains a byte buffer with the size of 65536 bytes. Since the persistent state map is not cleaned up and it may contain thousands of file references, a lot of TailFileObject may be created at startup, each consuming 65536 bytes which can end up in out of memory error. 
> The persistent state map needs to be actualized based on the actual files as it is done for the [states map|https://github.com/hortonworks/nifi/blob/CFM-2.0.4.0/nifi-nar-bundles/nifi-standard-bundle/nifi-standard-processors/src/main/java/org/apache/nifi/processors/standard/TailFile.java#L374].
> The allocated byte buffer size (65536 bytes) should be configurable to be able to use smaller buffer size (default value would be 65536 bytes for backward compatibility).
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)