KBI 310091 Unix Monitoring Scripts Can Fill Up /Tmp Filesystem




19 Feb 2008


In some cases UNIX monitoring scripts are not properly removed from /tmp on the Monitored Node.

Customers have reported running out of space on /tmp as a result.

Technical Background

The Argent Guardian uses PSCP to securely copy the monitoring script to the Monitored Node. The file is placed in /tmp.

If the script fails, times out, or is otherwise interrupted, the script is not properly removed by the ‘rm‘ command specified in c:\Argent\ArgentManagementConsole\ArgentGuardian\unixssh.ini.

These files can build up over time, eventually filling up the file system containing /tmp and causing more serious issues.


An Enhancement Request has been submitted to allow the user to specify on a per-Node basis (in Node Properties, Unix Tab) an alternate location for the copied scripts.

This would allow for more control over where scripts are deposited and tools like the Argent Guardian, the Argent Job Scheduler or even ‘cron‘ could automate the task of cleaning out dead script files.

To work around this issue, setup a cron job to run a shell script that deletes SCP* from /tmp on a daily basis (preferably around midnight to avoid deleting currently running scripts).