Skip to main content
  • Subscribe by RSS
  • Ex Libris Knowledge Center

    Large # of files in /tmp [error: /bin/grep: Arg list too long]

    • Article Type: General
    • Product: MetaLib
    • Product Version: 4

    This morning I logged into our server and cd'ed into /exlibris/metalib/m4_2 in order to "grep string */*".
    The system came back with this error: /bin/grep: Arg list too long.
    My sysadmin informed me that this happened because there are too many files for the system to expand.
    It turns out that directory /exlibris/metalib/m4_2/tmp [$TMPDIR] has 76760 files and growing all the time.
    Util X-9 is working properly and the files are only from the last 3 days (as expected).
    Is there cause to worry that there are 76,000 files in the $TMPDIR?
    I know the files get created with standard ML activity, but is that number of files excessive?
    If yes, what should I check?

    The files of the format set.XXXX, where XXXXX is a number, are the majority of the files in $TMPDIR.
    Those files are generated by z39_gate, which cretes a file for each committed search.
    We compared the number of files created daily (approximentally 25,000) to other large-scaled customers, and it appears rather normal.
    Please note that the QuickSets contain about 9-10 resources on average, and so it means that the number of actual searches needed to create 25,000 files of this format is not very large.

    Unless this large number of files is posing a problem beyond grepping/searching; we recommend simply deleting the contents of the $TMPDIR before grepping.

    • Article last edited: 10/8/2013