So, it's a bug in the framework, not in Rainlendar self, right? I think that it's not easy to implement such routines that would work clean on many different distributions. What about something unique like file access locking or something that you can be sure about a current running instance? Maybe that would be a bit dirty, but it may fix the problem (at least as a workaround).

Until that you can add a "Rainlendar won't start on Linux" with the "./rainlendar -m" command as a temporary solution, because I'm not the only one with this problem and sometimes it's annoying.

By the way: great software you wrote, thanks

As far as I can tell it uses file locking and I would think that once the process dies the lock is also released and the new instance should be able to acquire the lock again. I have no idea why this would sometimes fail.