writing System V init scripts 1

[nix45]

I'm writing a System V init script for a custom application. I'm basically just reverse engineering an existing script. I have a question on the Postfix init script on RHEL AS 3.0. I'm just using Postfix as an example, a lot of them do the same thing. I know Perl scripting much better than bash scripting.

start() {
        # Start daemons.
        echo -n "Starting postfix: "
        /usr/sbin/postalias /etc/postfix/aliases
        /usr/sbin/postfix start 2>/dev/null 1>&2 && success || failure
        RETVAL=$?
        [ $RETVAL -eq 0 ] && touch /var/lock/subsys/postfix
        echo
        return $RETVAL
}

In the line "/usr/bin/postfix start 2>/dev/null 1>&2 && success || failure", what does the "&& success || failure" mean? I know that && means to only execute if the previous command was successful and I know that || means "or", but whats with the 'success' and 'failure'? They aren't system commands or anything.

At the end where it says...
[ $RETVAL -eq 0 ] && touch /var/lock/subsys/postfix
echo
return $RETVAL

...they are saying "if the return value is 0, then create the file /var/lock/subsys/postfix". Why do they create this file? It doesn't seem to serve any purpose. Also, it then says "echo" on a line by itself, and then "return $RETVAL". What is the purpose of that?

Thanks,
Chris

 

[lgarner]

The "success" and "failure are functions called to display status. In the original postfix script, look for a line like:

. /etc/init.d/functions

This loads some generic functions , like those that show "OK" or "FAILED" in green & red when Linux starts.

The /var/lock/subsys/postfix file is used by the script to determine if postfix is running. There's a startup option "condrestart" which only works if it's running, and has the line:

[ -f /var/lock/subsys/postfix ] && restart || :

which says "if this file exists, call the restart() function". I'm not sure about the ":" in "|| :", though.

I think whether you need this depends on your script. I don't know if there is a rule of thumb or standard for creating these files.

I'm no expert, so this reflects my interpretation in simple terms that even I can understand...

 

[nix45]

thanks dude....anyone know about the last part? I know that in a Perl script, if you have a subroutine you might end it with something like "return $variable". Is that along the same lines?

Thanks,
Chris

 

[lgarner]

Oops, I missed that part. Yes, that's what it'll do- it's the return value for the start() function.

i.e.:
testfunc()
{ return 5 }

statuscode=`testfunc`
exit $statuscode

Then at the OS prompt:
linux# ./testscript
linux# echo $?
5
linux#

$? is kind of magic- it's the result of the last operation. So here, "exit $?" would do the same thing. That's why RETVAL is used to capture $? in the postfix script. Otherwise, $? keeps changing.

 

[nix45]

What part of the script reads the return value though? I know that if $? is equal to 0 the the last command was a success, but where does it read this variable? Do the success and failure functions use it? It seems like the command will run just fine without the "return $RETVAL" at the end. Thats why its confusing me a bit.

 

[lgarner]

It would be just fine. Continuing to use "postfix" as an example, the very last line of the script is "exit $?". So this value could be used by a script either calling "postfix" or run right after it.

The return value isn't required if you don't need it.

 

[nix45]

thanks lgarner