bug in perfdataproc.php mv with too many files

[jvestrum]

Minor bug report, I don't need support at this time but wanted to pass this along:

In /usr/local/nagiosxi/cron/perfdataproc.php, files are moved from /usr/local/nagios/var/spool/xidpe/ to /usr/local/nagios/var/spool/perfdata/ with a simple "mv *" command. Normally that works fine, but our perfdata cronjob was broken for a few days (for unrelated reasons). So thousands of files built up in the xidpe dir, and then when we got the perfdata job going again, the mv failed with "Argument list too long". So perfdata wasn't getting processed, and the xidpe dir kept growing.

I moved all the files over manually and perfdata is now working fine. It might be worth replacing the mv with something that can handle large numbers of files - many ways to do it: a find|xargs mv combo, rsync with --remove-source-files, tar/untar pipe, etc.

[abrist]

Good to know, though it is odd that npcd did not fail on parsing the perfdata directory due to the number of files after your manual move.

[jvestrum]

Good to know, though it is odd that npcd did not fail on parsing the perfdata directory due to the number of files after your manual move.

Nope, it took an hour but it didn't fail .

[abrist]

Nice, the power of C!
EDIT: I created a feature request:

Code: Select all

NEW TASK ID 5198 created - Nagios XI Feature Request: XI: Change the xidpe move commands to use a method that does not fail stat() when there are too many files