On Fri, 8 Nov 2002, Nico Erfurth wrote:
> > 2002-10-29 12:29:04 186UYR-0002vi-00 remote delivery process count got
> > out of step
> >
> > Running exim 4.10 with SUN Solaris 8. This doesn't seem to break
> > anything, but I'd rather it wasn't there. Any ideas what it might be
> > trying
> > to tell me?
>
> This happens from time to time, it just means that exim's internal
> counter of processes does not match the real count. This could happen
> because of faulty/lost signals and stuff.
That's true, but on the whole it doesn't happen. (I don't think we ever
see it here with Solaris, for instance.) Exim doesn't actually rely on
signals for detecting the end of child processes (because that is/was
unreliable on some systems). It uses waitpid() calls.
It shouldn't really happen, so I'm interested in trying to pin down the
circumstances, if possible, though I realize it is likely to be one of
those problems that is hard to track down.
> AFAIK the current snapshot has some fix for this problem, but IIRC it
> had to do with linux and ptrace.
There was a change in 4.05 that fixed some instances of this problem,
and yes, there is a fix in the snapshot for the case of strace in Linux.
However, I tested the equivalent in Solaris (truss), and it did not
have the problem.
Philip
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
