Do this before and after you restart those two processes:

Are the process owners (first column) the same both times?

The only other thing I can think of that might be different is your environment variables... This is a common problem with scripts which don't run properly from crontab, or remotely using ssh, or at start up, ... but which work fine when you test them as yourself. I usually debug this kind of problem by having the script redirect stdout and stderr to a logfile and dump it's environment variables first thing. By the way, this is why I was careful to use full pathnames in the example, since PATH is one of the variables that can differ. Make the script look more like this:



OBTW - When you set the "External Channel Change Command" under your "Input Connection" did you use the full path to the script? If you created it as /usr/local/bin/my_channel_changer.sh you need to use that full name in myttv-setup too.

I added the logging ,but the PATH and other variables seem equal. I also check the owner of the 2 processes and they are 'mythtv' for mythfrontend and 'root' for the backend.

On the other hand the logging did show me the problem :



The above logging happens when I change channel in mythfrontend. When I open an xterm at that point (mythfrontend still running) and execute the script, it switches fine.

So I figured the only possibility was that mythtv didn't pass on the channel correctly. I added logging for $1, and saw the channel was passed correct...

When mythfrontend refuses to switch the channel, I do notice this in the frontend output :

I don't know if it has anything to do with the problem.

So again I'm in the dark since I don't see the link between the segmentation fault, and fixing it with a simple restart of the backend.

For now I got around the problem, by adding a backend restart in the Knoppmyth-run script prior to starting the frontend.

Having the backend run as root isn't normal, it's supposed to run as the mythtv user. At least mine always have across more versions of KnoppMyth than I can count.

OK, that's ugly, the SEGV looks like it's coming from the shell script itself. Try redirecting stdin from /dev/null. Change the exec line to be:

You might also adding this so it is using the right HOME directory:


No idea how that error message is related, although that might become clear over time. The restart runs the code in a different context than what it runs in at start up, a different controlling console, potentially different environment variables, ... Those are the main things we need to look at. I've got another thread going with someone about mythfilldatabase where some similar issues have come up.

I'm going abroad for a week, I'll try the suggestions when I get back.
Thanks for the help so far