Topic: Dialup stopped working - Help!

[Sasha]

After upgrading a v4b3 server to the latest rpm's the dialup got messed up.

the rpm versions are e-smith-base 4.0.11-6, email-4.0.6-1, proxy-4.0.6-2.noarch
diald-0.99.4-0

At first it just kept dialing and dropping the line every 15-30 seconds. Then
it started to dial and just holds the line. The web access is working and the
mail is sending and receiving when the link is active.

What could have gone wrong?

Below is a portion of the messeges log.

It seems that it gets stuck on:
in.identd[685]: request_thread: read(0, ..., 1023) failed: Connection reset by peer

and just does not drop the line.

Jun  1 20:47:42 hat-server diald[372]: Calling site 192.168.1.2
Jun  1 20:47:43 hat-server chat[657]: abort on (BUSY)
Jun  1 20:47:43 hat-server chat[657]: abort on (ERROR)
Jun  1 20:47:43 hat-server chat[657]: abort on (NO CARRIER)
Jun  1 20:47:43 hat-server chat[657]: abort on (NO DIALTONE)
Jun  1 20:47:43 hat-server chat[657]: abort on (Invalid Login)
Jun  1 20:47:43 hat-server chat[657]: abort on (Login incorrect)
Jun  1 20:47:43 hat-server chat[657]: send (ATZ^M)
Jun  1 20:47:43 hat-server chat[657]: expect (OK)
Jun  1 20:47:44 hat-server chat[657]: ATZ^M^M
Jun  1 20:47:44 hat-server chat[657]: OK
Jun  1 20:47:44 hat-server chat[657]:  -- got it
Jun  1 20:47:44 hat-server chat[657]: send (ATL0M0DT0,4837200^M)
Jun  1 20:47:44 hat-server chat[657]: expect (CONNECT)
Jun  1 20:47:44 hat-server chat[657]: ^M
Jun  1 20:48:11 hat-server chat[657]: ATL0M0DT0,4837200^M^M
Jun  1 20:48:11 hat-server chat[657]: CONNECT
Jun  1 20:48:11 hat-server chat[657]:  -- got it
Jun  1 20:48:11 hat-server chat[657]: send (^M)
Jun  1 20:48:11 hat-server diald[372]: Connected to site 192.168.1.2
Jun  1 20:48:11 hat-server diald[372]: Running pppd (pid = 664).
Jun  1 20:48:12 hat-server kernel: PPP: version 2.3.7 (demand dialling)
Jun  1 20:48:12 hat-server kernel: PPP line discipline registered.
Jun  1 20:48:12 hat-server kernel: registered device ppp0
Jun  1 20:48:12 hat-server pppd[664]: pppd 2.3.11 started by root, uid 0
Jun  1 20:48:12 hat-server pppd[664]: Using interface ppp0
Jun  1 20:48:12 hat-server pppd[664]: Connect: ppp0 <--> /dev/ttyS0
Jun  1 20:48:21 hat-server kernel: PPP BSD Compression module registered
Jun  1 20:48:21 hat-server kernel: PPP Deflate Compression module registered
Jun  1 20:48:21 hat-server pppd[664]: local  IP address 196.2.27.233
Jun  1 20:48:21 hat-server pppd[664]: remote IP address 196.2.20.2
Jun  1 20:48:21 hat-server diald[372]: New addresses: local 196.2.27.233, remote 196.2.20.2, broadcast 0.0.0.0
Jun  1 20:48:22 hat-server diald[372]: start ppp0: SIOCSIFMTU: Invalid argument
Jun  1 20:48:31 hat-server in.identd[685]: started
Jun  1 20:48:31 hat-server in.identd[685]: request_thread: read(0, ..., 1023) failed: Connection reset by peer

[sasha]

It does not seem to be the updates at all.

I re-installed the server with the latest beta 7
and it still seems to be forcing the link up. It is not specific to this site as
the same problem occurs at another site.

The common link is the isp. The users can consistently connect to the Internet and the server collects mail (using ETRN and a fixed ip addr on link), however the diald script repeatedly gets

Jun  5 10:48:20 hat-server in.identd[5141]: request_thread: read(0, ..., 1023) failed: Connection reset by peer

and then re-dials.

I suspected it might have to do with the ETRN settings, since removing these did seem to make the link quiet, but not consistenly. The weird think is that the server still collects its mail even if the ETRN servers DNS name is blank.

[sasha]

Problem solved.

It is the broken DNS in the latest RPMS's that is causing the modem to dial.
In this case Outlook clients were trying to connect to mail.mydomain.xxx and
this would force the link up.
Setting Outlook to connect to the ip addr of the server stopped the mad dialing.