NT port link activation issues

Hi,

it seems I have some misunderstanding what I can expect regarding NT ports and link activation (layer 1 & 2).

I try to implement a Asterisk based ISDN voice gateway / proxy to extract
some data from the D-channel. I want to intercept a ISDN line like this:
existing setup:
NTBA -> PTP mode -> TE Port Alcatel 4400 PBX
with the ISDN gateway/proxy:
NTBA -> PTP mode -> TE Port Asterisk PBX <-same machine-> NT Port Asterisk PBX -> PTP mode -> TE Port Alcatel 4400 PBX

The ISDN gateway should be integrated with a very short service disruption, so neither the Asterisk PBX nor the Alcatel can be rebooted after unplugging/plugging the ISDN connection cables.

I made testing with two different ISDN cards:
Sangoma A500
Junghanns Quad-BRI

I use Debian 9 and the included Asterisk, libpri and Dahdi for the Junghanns card. And I use a Ubuntu 16.04 with the Debian Asterisk, libpri and Dahdi versions. The Wanpipe driver for the Sangoma ISDN card is self compiled.

I have another PBX system for some cross testing in our lab. It is a Cisco voice gateway with dual BRI ISDN cards.

What I see about the link activation between NT port Asterisk PBX and TE port Cisco voice gateway:

  • when I hot plug the cable to the Sangoma card, the port does not come up. When I restart the wanrouter service the link comes up and the proxy works as designed. When I remove the cable and keep it disconnected for longer than 10 seconds, the link does not come up again. When I trigger a call from TE side, the link comes up and stays up.
  • when I hot plug the cable to the Junghanns card, the port does not come up. When I restart the server or activate the link via a Call from the TE side, the link comes up an stays up. So even when I disconnect the cable longer then 10 seconds, the link always come up again.

When I try to generate a call on Asterisk to go out on NT port or trigger it with a phone or soft-phone I get busy and the NT port isn’t activated.

When I do basic tests with the Cisco NT port, the port always come up directly and stays up. No matter what kind of TE port I connect.

This is all for PTP mode.

Indeed for PTMP mode, the NT port on the Linux side comes up and stays up. No problems seen so far. But actually I cannot change anything on the existing PBX and reconfiguring it to PTMP is no option.

So I am wondering, does it works as the standards define or is here some bug between libpri/chan_dahdi and dahdi kernel modules?

Thanks in advance for any hints,
Waldemar