I've heard a few reports on this in IRC. The startup scripts do not check to see if the programs that they are still running when 'script start' is called. They are only checking to see if the 'script start' has been run. If 'script stop' hasn't run, then it fails to start. There can be some issues with this if it depends on having the program running. This should be fixed. The scripts should verify the PID in a/the pidfile they create using if ps PID &>/dev/null; then IT_RUNNING_SO_FAIL; else IT_IS_NOT_RUNNING_SO_START; fi That way when a program crashes, you can just run 'script start' again and it will realize that the program is dead for some reason.
Note from <roman_> "stop" is also broken, 'cause when the PID does not exists, it will die in an error, and rc will not erase /mnt/.init.d/started/<link>
That is what the "/etc/init.d/foo zap" is for. Meaning you could do for a service that died: # /etc/init.d/foo zap start Not that efficient, I know. I have been working on improving this (all local changes), but the one or two I sent the examples to did not reply as of yet.
Non issue
