Shell Programming and Scripting

Hi,

I have a shell script as below:

ORACLE_HOME=/usr/local/opt/oracle/product/dev
export ORACLE_HOME
PATH=$PATH:$ORACLE_HOME/bin:/usr/bin
export PATH

OUTFILE=/export/home/`basename $0`.out
export OUTFILE
export IDEN

df -k . | tail -1 | read a b c d e f
echo $a >> $OUTFILE
echo $b >> $OUTFILE
echo $c >> $OUTFILE
echo $d >> $OUTFILE
echo $e >> $OUTFILE
echo $f >> $OUTFILE

When i try to execute this script manually, I get the value for all a,b,c,d,e,f populated. However, on trying to execute it thru the cron, all these values come blank.

Can anyone help me with the reason behind this?

Thanks in advance.

cron's default PATH is much more barebones than one you get from a login. Set PATH to a sane default, not just $PATH:otherstuff

Also, this:

Code:

df -k . | tail -1 | read a b c d e f

I'm amazed this ever worked. You must have been using a very strange shell to get those variables read and assigned in a usable way behind three pipes. In nearly every shell that'd be done in a subshell, and save no results to the main shell. Redirect the output of that into a tempfile instead, and read a b c d e f < tempfile, then rm -f tempfile

So you tell everything, env is not same as you login. Define the shell, PATH and so on.
Ksh can read from pipe, no problem.

Code:

#!/usr/ksh
PATH=/bin:/usr/bin:...
...

This syntax has always worked for me in SunOS ksh.

In my cron jobs, I always have a special cron_env.sh that contains all of my required variables like PATH and whatnot. Then the cron entry looks something like:

Code:

* * * * * /bin/ksh /home/quirk/bin/cron_env.sh /fullpath/to/my/cron_job.sh

... keeping all complicated looking garbage in the cron_job.sh itself.