question about find command

[nirvanaforu]

Hi All:
I would like to write a small shell script to be added to cronjobs, the main purpose is for the shell script is to delete the outdated files and outdated sub-directories under a certain folder, however I do not want to delete the top folder.
For example: the top folder is /proj/tmp
the command I wrote was:
“ find /proj/tmp/ -mtime +10 | xargs rm -rf ”

The problem is: once there are some files inside satisfying the find condition, "/proj/tmp/" will be returned as well.
I can't add -type f, because I would like to remove those subfolders as well.

if I revise the command as the following:
“ find /proj/tmp/* -mtime +10 | xargs rm -rf ”
it works as I expected. However, as the top folder is empty, it always report some errors like: "find /proj/tmp/*: no such file or directory". If it's added to cronjob, which will be emailed to root again and again.

Any suggestions to this problem? Thanks a lot!!!

btw: I tried with >/dev/null, which didn't help with the error message. Someone suggested tmpwatch, the problem is that I don't know which files or directories returned by tmpwatch, so that I can't test with tmpwatch.

[vsemaska]

Error messages go to stderr not stdout. So if you want to ignore error messages the redirection would be 2>/dev/null

[secondmouse]

find /proj/tmp/* -mtime +10 | xargs rm -rf

find /proj/tmp ! -regex ^/proj/tmp$ -mtime +10 -print0 | xargs -0 rm -rf

-regex is being used here to exclude the root directory itself. -print0 to print the find result in one line and pass it as the parameter of the rm command. good luck.

[sathiya]

use the mindepth option, and eliminate the search in the parent directory....

hope this helps ?

[secondmouse]

The mindepth approach is definitely more elegant than any other methods mentioned in this thread so far. Thanks for sharing.