My predecessor wrote a perl file that is run manually twice a month to send a set of emailed reports to one of our clients. My boss has requested that I look into automating this; it seemed simple enough, and I thought I had it set up, but I'm running into a rather strange problem.

Both the emailing program (in perl) and the report files themselves are kept on a Novell server reached from our Linux box via /mnt/srv/path/to/file/. From the linux box, I can see and manipulate them fine. The files are displayed as having 755 permissions across the board. The Linux box itself is running SuSE 9.3.

A manual run of the emailing program works fine; the files are sent out as expected. However, when I tried to set up a crontab entry for this, I'm running into trouble. I manually edited /etc/crontab to include a line setting the program to run at '0 21 * * 1,2,3,4,5'. The program itself runs fine, but errors out every time it tries to attach a file; the error message is one that various perl sites indicate means 'could not read the path to this file'.

What's the difference between a cron run and a manual run? What are some possible reasons the former fails and the latter works?

Possible differences:

What user is the crontab run as? Is it run as root?
Are all of the paths fully qualified? I believe when a cron job runs, it doesn't run with the environment variables you have when you are logged in the shell.

1 members found this post helpful.

can you cat the /etc/crontab file so we can see exactly what you've done?

Thank you for pointing that out.
There was a point in the code where it was looking for a relative path to a file; I've personally run that file manually from multiple directories before and had it work, so that particular path may have been in the environment variables for my login. Coding the path into place ahead of it allowed a crontab run to go ahead successfully; thank you for your assistance.