On Mon, 2003-10-27 at 04:15, Daniel M. Drucker wrote:
> I keep getting messages like these in /var/log/messages:
>
> Oct 26 22:52:25 eco kernel: application bug: exim(7111) has SIGCHLD
> set to SIG_IGN but calls wait().
> Oct 26 22:52:25 eco kernel: (see the NOTES section of 'man 2 wait').
> Workaround activated.
>
>
> This seems to have been discussed here a few times before, but with
> the resolution being that it was fixed in 4.14. I am running Exim
> 4.24 with Linux kernel 2.4.20, and I'm still getting that error. Any
> ideas?
If you are not an ipv6 site, then switch off all ipv6 stuff by:-
* setting local_interfaces explicitly to 127.0.0.1 and your own
IPs
* setting "dns_ipv4_lookup = *"
I think the latter action is superfluous but have not spend time working
it out.
My theory as to why this affects things is that the listen against an
ipv6 wildcard forces a module load of the ipv6 stack which causes some
related process in the module load to call wait without the SIGCHLD
being reset. However this is unproven, and currently I consider it a
RH bug.
Nigel.
--
[ Nigel Metheringham Nigel.Metheringham@??? ]
[ - Comments in this message are my own and not ITO opinion/policy - ]
