ctlinnd oddness

[Dave Williams]

Dave Williams <dave at clues.com> wrote:
> Alex Kiernan <alexk at demon.net> wrote:
> > 
> > Permissions on the control socket?
 
> I should've said - sorry.  Not the permissions on the control socket.
> Looked there almost first thing. :)

It looks like (at least for 'pause') any argument passed to cltinnd
is being thrown away - probably somewhere inside ctlinnd - trying
to pause or throttle means the SC_THROTTLE gets passed to innd and
innd gives us CCnoreason as a result. 

I draw the conclusion (possibly on superficial evidence) that there's
a problem with that bit of code (and anything related) because using
the old ctlinnd binary (well, there's been no protocol change) means
I can successfully pause or throttle the server.

The only real differences between the code in question seem to be
the style changes (TRUE+FALSE/true,false, ENDOF/ARRAY_END (and
ARRAY_END's use of sizeof(X) / sizeof(X[0])), xmalloc rather than
new) and some minor structural changes to accomodate using xmalloc
rather than NEW. 

I could be missing something though. :)

Dave
-- 
Dave Williams
dave at clues.com