Cisco IOS ISDN Serious Problem

[Maondas]

We recently encountered a serious problem with Cisco IOS 12.2(19) and ISDN connections. Our setup was a 7206vxr router with 2 inbound PRIs.

When a remote site would fail over to ISDN backup (from frame relay) and call in to the 7206, the connection would come up correctly. When frame relay service returns, the ISDN connection is supposed to drop after 15 minutes and the ISDN interface on the remote router should return to a standby mode.

After 15 minutes what we saw was the remote ISDN interface indeed going into standby mode, and the frame interface up/up. However looking at the local 7206, we still saw the ISDN connection in a connected state. Thinking that was quite odd, we bounced the PRI interface, which killed all connections (sh isdn serv, etc showed no active lines).

Or so we thought.

When we got a large bill for ISDN usage the next month, we started investigating.

It seems what happens is even though the local and remote routers see no sign of an ISDN connection, it is STILL THERE (confirmed by ISDN provider). So basically we had calls up 24x7 that were not being used.

The reason for this post is to query the user community if anyone else has ever experienced this problem before or heard of anything like it. Cisco has not yet found this is a bug in their IOS 12.2(19).

This bug started occuring right after we upgraded to 12.2(19), and stopped as soon as we rolled back to our old IOS. This is how we know that it is an IOS bug.

So, any thoughts?

Thanks!

 

[vipergg]

Don't have anything that high but we have not seen a problem with 12.2.15 T9 with isdn , basically a setup like yours .

 

[techkiwi]

It would be interesting to see if it's possible for your ISDN provider to give you more details. Can they see calls being cleared down correctly? Which end is initiating the CLEAR REQUEST or whatever they call it nowadays? Is the 7206 responding to the CLEAR REQUEST and then making a new call for some reason, or is it just ignoring the CLEAR REQUEST?

Graham

 

[Maondas]

I'm not sure which side is issuing the clear request on the ISDN line, but I believe it would be the remote side. It is the remote routers that call in to the [local] 7206.

Sometimes the remote would go into standby mode and apparently the 'clear request' wasn't processed properly by the local 7206, leaving the connection hanging in a state that even shut/no-shutting the PRI lines on the 7206 could not correct.

In other scenarios it seemed the remote router would keep calling back every few seconds after it was kicked off ISDN (again there was no need to call back - the primary link was up).

Im doing some troubleshooting today with Cisco in our current [working] environment so I hope that may reveal if we do anything different than other ISDN implementations. Also I hope to gain a better understanding of what exactly is happening behind the scenes.

Thanks - Mao