When using "/etc/rc.d/init.d/sendmail restart" which is basically a stop
followed by a start it fails.

The fault can be looked at in one of two ways.

1. The stop being at fault:
It only kills the original sendmail server process and will leave other
sendmail proccess around to complete sending/receiving mail.  This is
actually desired behaviour IMHO.

This means the start process finds a sendmail process running and refuses
to start the server.


2. The start proccss being at fault:
The start process perhaps needs to pass an optional extra argument or
employ some method such that it can search for a certain instance of a
process.

Eg. in the case of sendmail the process that is relevant to the restart is
the one accepting connections which appears as follows when doing a "ps
auxww":
sendmail: accepting connections

Hence there exists the opportunity to test for a process that has both the
process name of sendmail and contains the key information "accepting
connections"

I hope this makes sense as it is certainly a feature I'd love to see fixed
as we regularly restart our mail server after various configuration
changes.