UNIX for Dummies Questions & Answers

Hi gurus,

I'm having a strange problem and I hope you can help me solving it.

I'm working in Unix Solaris, version 5.10, ksh.
I have a script with environment variables which I have to execute prior to other scripts. The script belongs to userA and when I log in, and cd to its directory. It doesn't work and the following message appears:

Code:

$ . load_env.sh
ksh: load_env.sh:  not found

The strange thing is that if I log in with a different user, userB, and then I do:

Code:

$ su - userA

Then cd to my script's directory and execute it, it works.

The command env returns similar results in both cases.

Thanks in advance for your help.

Regards

ermur

It sounds like your PATH environmental variable may be the culprit.

What happens if you do?

Code:

$ . ./load_env.sh

Hi fpmurphy,

In both cases the PATH variable is the same:

Code:

$ echo $PATH
/usr/bin::/opt/EMCpower/bin:/etc/emc/bin:/etc

if I use . ./ it works. However, I would like to have it working the other way (<dot><space>), since, this is not the only machine (the other 4 machines are working) with this configuration and the operations personnel have already been trained.

Thanks a lot for your help

The highlighted blank entry in $PATH is the equivalent of having a dot "." in $PATH (which is a known security issue). When your current working directory is the same as the script it will execute if the script is executable and the user has permission to execute the script.

Code:

Example.  See the effects of an extra colon in $PATH.

myscript.sh
echo "working"

PATH=/usr/bin
myscript.sh
sh: myscript.sh:  not found.

PATH=/usr/bin:
myscript.sh
working

PATH=:/usr/bin
myscript.sh
working

It is more secure to create a central scripts directory with the scripts owned by root and each script with permissions 755. Then append the scripts directory to $PATH in each user's .profile. Where there is a mixture of user groups on a machine it can be necessary to set the permissions to 750 and use secondary groups to control who can execute a script.

A central script can then be executed with the command you describe. We use this technique extensively for ease of maintenance. Also if a user owns a script they can change or delete it !

Thanks a lot for your valuable help.

ermur