Re: checkLocation service dying

[Chris Warburton]

Looks like this is caused by pollingService failing to satisfy the
requirements of monitoredService.

monitoredService takes 'shouldRun' and 'isRunning' scripts. If these are
inconsistent (i.e. we're not running when we should be, or vice versa)
then a start or stop action is taken.

After taking an action, we check whether we're consistent, and 'fail' if
not.

The 'pollingService' is passing 'false' as the 'isRunning' command. This
ensures that the 'start' script always gets called whenever 'shouldRun'
gives true, which implements the polling behaviour quite nicely.

However, forcing things to always be inconsistent in this way will cause
the sanity check to fail, and hence we'll crash with a call to 'fail'.

I've tried to fix this now, by adding an optional 'allGood' script to
monitoredService. If given, this will be used as the sanity check
instead of checking for consistency. This way, we can pass 'true' and
avoid this problem.