Why does service stop after RPM is updated

Question

I have a software package for which I created an RPM. I can't paste the entire RPM here for IP reasons, but here is the gist of the problem:

%pre
/sbin/pidof program
if [ "$?" -eq "0" ]
then
  /sbin/service program stop
fi

%post
/sbin/chkconfig program on
/sbin/service program start

%preun
/sbin/service program stop
/sbin/chkconfig program off

%postun
rm -rf /program_folder

Everytime I try to upgrade the package, it stops the program service, installs everything, starts the service, and then stops it again and deletes the folder...any ideas?

Answer 1

This has to do with the order in which the scripts are ran :

%pre of new package
(package install)
%post of new package
%preun of old package
(removal of old package)
%postun of old package

So in your case, the %preun of the old rpm is ran last, which shuts off the service.

This can be solved by looking at the argument to the post script. Here is a table for the value:

          install    upgrade  uninstall
%pre      $1 == 1   $1 == 2   (N/A)
%post     $1 == 1   $1 == 2   (N/A)
%preun    (N/A)     $1 == 1   $1 == 0
%postun   (N/A)     $1 == 1   $1 == 0

So, you'd want your %postun script to instead be this:

%preun
if [ "$1" = "0" ]; then
    /sbin/service program stop
    /sbin/chkconfig program off
fi
exit 0

That will make it stop the program only on complete uninstall (ie; rpm -e) and should do it for you.

NOTE: you'll want to do the same with your %postun script as well

NOTE: the exit 0 at the end; I like to specifically put that there just in case, as a bad exit code from the last command will carry over and cause the script to exit with that status, causing problems with the rpm installation.

NOTE: fixing this in the new rpm doesn't fix it in the currently installed rpm. You'll want to rpm -e the old rpm with the broken script, and you will be fine moving forward.

This just covers the pre/post scripts; a more detailed version with triggers and such can be found here.