Gaby Vanhegan:
> > There is a delay of up to $trigger_timeout seconds when the Postfix
> > postdrop command tries to notify the pickup daemon that new mail is
> > ready for delivery.
> >
> > PHP -> sendmail -> postdrop -> pickup
> >
> > Perhaps your pickup daemon is very busy.
>
> The system itself is very low traffic right now, about 1 email/hour,
> practically zero load average, the only other traffic is a mostly
> static website.
>
> > Perhaps you use Linux, and have Selinux brain damage turned on.
>
> I'm using CentOS 4.2, which is what came installed as part of my
> hosting package. I can't find any references to SELinux that are
> obvious but that doesn't mean it's not in use somewhere. I will
> investigate further.
>
> > You can ignore the problem and specify "trigger_timeout = 1" in
> > main.cf, or you can try to solve the problem and find out what
> > keeps the pickup server from reading its FIFO.
>
> I'd prefer to fix the root cause of the problem. I have eliminated
> some of the delay by installing my own DNS resolver, the ones supplied
> by my host were slow for some reason. How would I go about
> investigating why pickup is taking a while to kick in? Is there a
> debuggering switch I can turn on somewhere?
The pickup daemon is supposed to be running all the time, and it
is supposed to react immediately. For debugging, see the "-v" option
in the master(5) manual page.
If you submit mail as a non-httpd user, then you will very likely
find that mail is delivered immediately.
Unfortunately, some Selinux configurations are known to prevent
the postdrop command from working properly when invoked by a web
daemon. Many people have found that turning off Selinux will
mysteriously solve problems.
Needless to say, I do not offer any warranties for damage done
by Selinux brain damage.
Wietse
