script not working in crontab

Question

I noticed a number of similar questions ask and I also tried the suggestions posted, but I just can't seem to get this working. Below is my code.

#!/bin/bash

while IFS="" read -d "" -r filename;
do
filearray[i++]="$filename"
done < <(find . -maxdepth 1 -type f -iname "*.txt" -print0)
printf '%s\n' "${filearray[0]}"

Simply, I just want to search for all txt type files in a specific directory and put them into an array, which is displayed at the end. When I run from the command line, no problem. It works well. As soon as I execute this via crontab, I get the following error:

syntax error near unexpected token `<'
`done < <(find . -maxdepth 1 -type f -iname "*.txt" -print0)'

Here's the cron entry itself:

* * * * * . /usr/online/scripts/test.sh 2>> /usr/online/scripts/log/test.log

Why is the script working on command line, but not crontab? I am declaring the shell, so surely crontab should use the shell? Is there another why I can do this which works?

Jakkie

Answer 1

command < <(other command) is a bashism. Since you are specifying a shebang line, you should not specify a shell in your crontab; that's just confusing. You should do one or the other - I very much prefer a shebang line because you're being explicit about the syntax that follows, rather than pretending to be a "generic" shell script.

Oh, and make sure you're making your script executable and putting only /path/to/your/script.sh in the crontab command field, not sh /path/to/your/script.sh, sh < /path/to/your/script.sh or even . /path/to/your/script.sh. The last three ignore the shebang line, and instead runs the script in the context of the cron shell, whatever that is set up to be.

See redirection instructions for more info on saving script output to a file.

Answer 2

Assuming the script is executable and is doing the right thing when invoked from the command line, write your crontab entry like this (to run every 15th minute):

*/15 * * * * /path/to/script.sh >/path/to/output 2>/path/to/errors