[exim] Strange timeout problem on port 113

Hi all,

I got a strange timeout problem: I put my linux box with exim behind a
router few days ago instead of connecting it directly to the internet.
It had a real IP address before and now it's got only an address of my
internal network. Packets to port 25 are now forwarded from the external
address (router with firewall) to the internal (exim server).

The strange thing is that while trying to connect to the external
address on port 25 exim opens a session and spawns a new child which is
now trying to establish a connection to the external address on port 113:

[root@zebehn root]# strace -p980
connect(0, {sa_family=AF_INET, sin_port=htons(113),
sin_addr=inet_addr("217.9.31.15")}, 16) = ? ERESTARTSYS (To be restarted)
--- SIGALRM (Alarm clock) @ 0 (0) ---
...

This connection hangs just before "ERESTARTSYS" and then fails since the
firewall does not answer requests to this port. After that timeout the
SMTP request is normally processed. When trying to "telnet 217.9.31.15
113" my telnet session hangs also, but "telnet localhost 25" from the
exim box I get a "connection refused" immediatly. This seems to be the
reason why I didn't have this behavior before.

I inspected my exim box and found no place where the external address is
still configured. But the exim configuration if course still uses the
name of the external address as primary_hostname. The question is, how I
can tell exim not to connect to 113 (which seems to be authd, right?)
and from where does exim take the external address to try to connect to
this port?

Hope I didn't forget a detail. Thanks in advance for any help.

Stefan