Evgeniy Berdnikov <bd4@???> (Mi 08 Mai 2013 09:39:13 CEST):
> On Tue, May 07, 2013 at 03:13:56PM +0200, Heiko Schlittermann wrote:
> > <nitpicking>
> > If the pipe blocks, new exim processes will startup and try to accept
> > messages, resulting in a huuuge amount of blocked exim processes, until
> > the connections run into a timeout. If exim checks the log space, and
> > the space is insufficient, it will 4xx on new connection attempts.
> > </nitpicking>
> >
> > :)
>
> Good point for digging. Result really depends on implementation details.
>
> When exim runs as a listener, it forks for every new incoming connection.
> Each child, as well as parent, do fd=open("mainlog",O_WRONLY|O_APPEND...),
> write to fd adn then close it. On writing to blocked pipe each process
> would block independently, regardless is it parent or child.
>
> With default log selectors parent process accepts incoming connections
> silently and all looks as you describe. With log_selector="+all" parent
> process writes log record about new connection, and would block if pipe
> is blocked. So exim listener would hang and stop spawning childs.
:) Thank you for clarification. So, both we're right, in case that was
the point (for me, it wasn't :))
Best regards from Dresden/Germany
Viele Grüße aus Dresden
Heiko Schlittermann
--
SCHLITTERMANN.de ---------------------------- internet & unix support -
Heiko Schlittermann, Dipl.-Ing. (TU) - {fon,fax}: +49.351.802998{1,3} -
gnupg encrypted messages are welcome --------------- key ID: 7CBF764A -
gnupg fingerprint: 9288 F17D BBF9 9625 5ABC 285C 26A9 687E 7CBF 764A -
(gnupg fingerprint: 3061 CFBF 2D88 F034 E8D2 7E92 EE4E AC98 48D0 359B)-
