Warren Togami wrote: > Wouldn't it be preferable to have a solution where spamd itself returns > when it is finished stopping rather than relying on an asynchronous kill > signal?
Warren, what do you mean by "spamd itself returns when it is finished processing"? Isn't the problem how to identify the correct process that you want to stop, and not anything to do with how it stops once you ask it to? This seems to be a general problem with services that are not started up as root, and a deficiency is relying on pidof to do it. Or a deficiency in pidof. Since you have the same problem with ISC Bind named, that demonstrates that a general solution applicable to all services that are started as non-root is necessary. I could think of one alternative in which the script that starts up spamd and does run as root copies or moves the pid file to /var/run. But that is not much different than the solution in the bug #161785 in which you create a directory to put the pid file in and create a softlink to it in /var/run. Is there a way to create child processes so that they appear sufficiently different to ps so that it is easier to get pidof to work? -- sidney
