[TIKA-3208] tika-server Detect when using fileUrl header does not close the file handle

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.23, 1.24, 1.24.1
• Fix Version/s: 1.25
• Component/s: server

Environment

Centos 7

java 1.8

 

• Assignee: Unassigned
• Reporter: Darren Cooper
• Created: 09/Oct/20 20:38
• Updated: 09/Oct/20 22:43
• Resolved: 09/Oct/20 21:07

Description

When using the fileUrl header to detect a file on localdisk the java process does not release the file handle.

When the detected file is deleted, the disk space is not released because of the open file handle.

Test

curl -X PUT -H "fileUrl: file:///root/test.pdf" http://localhost:9998/detect/stream

rm /root/test.pdf

lsof of (pid of java tika-server)--

file handle open to /root/test.pdf (deleted)

 

As a test fix tika-server/src/main/java/org/apache/tika/server/resource/DetectorResource.java was modified to add

try

{ tis.close(); }

catch (Exception e) {}

 

to the finally blcok in public detect and the problem seemed to be resolved.