[
https://issues.apache.org/jira/browse/TIKA-567?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12983395#action_12983395
]
David Benson commented on TIKA-567:
-----------------------------------
We upgraded from Tika 0.8 to 0.9-SNAPSHOT because of TIKA-548, but the
temporary file leak is causing large indexing jobs to run out of file handles
(even after bumping up fs.file-max = 65535). I would suspect this would be a
major concern for anyone else indexing millions of PDFs. Can this be addressed?
Thanks
> Temporary file leak in TikaInputStream
> --------------------------------------
>
> Key: TIKA-567
> URL: https://issues.apache.org/jira/browse/TIKA-567
> Project: Tika
> Issue Type: Bug
> Components: general
> Affects Versions: 0.8
> Reporter: Jukka Zitting
> Assignee: Jukka Zitting
> Fix For: 0.9
>
>
> The TikaInputStream.get(InputStream) method can end up leaking temporary
> files in cases like the Detector classes that will not close the returned
> stream.
> There should be a mechanism to signal that a potential TikaInputStream
> wrapper will no longer be used even if the underlying stream should not be
> closed.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
