Apologies for the delay. I had quite a mixture of responces from people about this : - Renaming /tmp to /tmp-old, recreating /tmp with newfs and then deleteing /tmp-old. - Use find but replace the "| xargs rm -f" with "-exec rm {} \;" - Use the mtime option to find to only remove file that are over x days old. - (rm -rf /tmp ; mkdir /tmp ; chown root.root /tmp ; chmod 1777 /tmp) in a startup script. A couple of people drew my attention to possible security implications when using a "find .. | xargs rm -f" as root. See the following page for more details : http://www.securityfocus.com/archive/1/4494 I am still in two minds about what to use. I think it will either be creating a new file system at boot time but I'm slightly concerned about the delay that this will cause, or use the following as a cronjob : "/usr/bin/find /tmp ! -name 'lost+found' ! -name 'ps_data' ! -type l ! -user root -atime +7 -exec rm -rf {} \;" Thanks go to Andrew Brennan, Casey Jones, Sean Quaint, Vince Merrell, Lieven Marchand, Arun Gurung, Rick Francis and James R Grinter for their responces. Original Mail: -------------------------------------------------------------------------- I have a few E10k domains that have standard UFS /tmp's which obviously don't get emptied upon reboot as tmpfs /tmp's do. I've been using a simple startup script that does the following : /usr/bin/find /tmp ! -name 'lost+found' ! -name 'ps_data' | xargs rm -f I was just wondering if anyone had a more elegant script that provides better functionality. -------------------------------------------------------------------------- Cheers, Adam.Received on Thu May 24 18:28:56 2001
This archive was generated by hypermail 2.1.8 : Wed Mar 23 2016 - 16:24:55 EDT