In the Instance Node, detecting the MIME type of Microsoft documents (doc, docx, ppt, pptx, etc.) or Zip files results in file descriptor not being closed and finally, an error of  "Too many open files"

You can recognize the issue because of the errors " <Last file opened> (Too many open files)"


Steps to Reproduce

  • Ensure the system property property is set to true (default) and validate that your Localhost Node logs are showing the error: " /proc/version (Too many open files)"
  • on your local box (e.g. Mac), run the command: "lsof | grep java | grep apache-tika" and notice the number of open tmp files
  • open a problem, and attach to it several MSFT files (doc, docx, ppt, pttx)
  • rerun the lsof command, and notice that the files remain open
  • re-attach the [same] files, and notice that each time the number of open files increases, e.g.
    /var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T%lsof | grep java | grep apache-tika
    java 43338 ivan.covdy 729r REG 1,4 1002581 15622531 /private/var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T/apache-tika-4561432680763589546.tmp
    java 43338 ivan.covdy 730r REG 1,4 344443 15631383 /private/var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T/apache-tika-1201343537274981905.tmp
    java 43338 ivan.covdy 739r REG 1,4 642385 15622931 /private/var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T/apache-tika-1668968703780498832.tmp
    java 43338 ivan.covdy 748r REG 1,4 1708609 15622533 /private/var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T/apache-tika-7038660925747838567.tmp
    java 43338 ivan.covdy 749r REG 1,4 522240 15622786 /private/var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T/apache-tika-9193332320042250766.tmp
    java 43338 ivan.covdy 750r REG 1,4 42471 15620816 /private/var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T/apache-tika-3560554309601548080.tmp
    java 43338 ivan.covdy 751r REG 1,4 2969478 15620818 /private/var/folders/bh/10qvvlh553g9l_4j051sdrlr0000gn/T/apache-tika-772828221569896927.tmp


  • Restart the affected Instance in Hi.
  • Alternatively, On-Premise customers can restart their own affected nodes.
  • The https://<<host>>:<<port>>/ would show the last restart time.

Related Problem: PRB1292566

Seen In

There is no data to report.

Fixed In

Kingston Patch 11
London Patch 3

Associated Community Threads

There is no data to report.

Article Information

Last Updated:2019-11-19 02:05:39