cantor.2 at osu.edu
Tue Sep 16 14:01:07 UTC 2008
> Exactly! As shibd does not daemonize itself, the init script has to
> do that before shibd is started. Thus its stderr gets lost. If one
> starts shibd without the init script magic, it puts out the same
> warning as shibd -t, then hangs on the console running the listener
> loop. Mystery solved.
Ok. That doesn't explain why 2.0 didn't log the same error, of course.
> Now, why doesn't shibd daemonize itself?
I don't know how. I develop on Windows. All the Unix support is ported
afterwards without any particular adaptations other than a small amount of
signal handling that I didn't actually write.
There's a Red Hat script function that's supposed to daemonize things, and
it didn't work when I tried it either.
> Would you accept patches
> which make it do so? Do you expect any complication?
Yes, and I have no idea, since I don't know what the difference is.
More information about the Pkg-shibboleth-devel