qa3019, autofs-3.1.7-14 If autofs dies and the lock file remains there, there is no way you can restart autofs! Since the process is not running the returnvalue of killproc is 1 and the lock file does not get removed upon "stop". Hence "start" wil do nothing and exit with 0. I think this should be fixed! Maybe other initscripts are affected as well.
This problem is greatly exacerbated by the recent change to autofs commenting out the /misc entry from /etc/auto.master. Because of this change, the default configuration of autofs is that there are no mount points, i.e., /usr/sbin/automount won't be started at all when "autofs start" is run. Nevertheless, the lock file will get created, but then when "autofs stop" is run later, it'll fail because there are no automount processes to kill. This on a properly configured system with no autofs mounts configured but with the service enabled anyway, stopping autofs will always fail. I will attach a patch.
Created attachment 29347 [details] Patch to make "autofs stop" behave cleanly when none were started
This fix should be integrated into 3.1.7-20 and later. Thanks!