SIP/Cube router locks up after a week

Unanswered Question
Feb 10th, 2010

We have a Verizon SIP trunk terminated on a 3845 router running 12.4(22)T IOS and sending calls to CUCM 7.0.

The trunk is SIP from 3845 to CUCM. We are transcoding most of the inbound calls from 729 to 711.

After a number of days (no set pattern), the router locks up and there not even console access.

I can see the ethernet and serial ports are showing up, and ethernet ports are showing activity, but there is no access to the box at all.

Only access is to reload.

This same router was working fine for months with no problem before the upgrade from Call Manager 4.x to CUCM 7.0.

Same number of calls same everything.

Prior to the upgrade, the SIP was the same, same router hardware, same config.

Per Verizon they will only support IOS 12.4(20)T, so we had that prior to CUCM upgrade, but have tried that IOS since upgrade and still router locks up.

Both Verizon and Cisco are saying they have no idea why this is happening

Once thing we have noticed on "sh sccp stat" is that over time the StopMediaTransmission counters are showing 0 sucessful stops, and transmissions accumulate over time.

StartMediaTransmission rx 86108, successful 86108, failed 0

StopMediaTransmission rx 86062, successful 0, failed 0

I happened to take the above snapshot a couple of hours prior to the lock up, and then looked after reboot and all counters are showing 0.

Are the sccp channels staying open and possible causing a memory leak over time?

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
markbatts Wed, 02/10/2010 - 06:56

i think you have a couple of options.

Swop the hardware , its quite unusual for router to lock completely , normally when they have trouble they just reload themselves after a crashdump so if it was me i would do that first.

Second if it is a bug and cisco don't know if it yet i would try and find an iOS that does work , its pain but try each release going back one by one, hopefully eventually you should find one that works.



matt.rossmeisl Fri, 02/12/2010 - 11:55

We are having the same issue with a customer, do keep us posted.

The latest IOS supported by Verizon is 12.4(20)T4. Rolling back to this will not help you. We are still expiriencing the exact same problem with this IOS as well as with the previous IOS we were running 12.4(20)T2. Customer had been running this setup with absolutely no reported issues on UCM 4.x for months (almost a year). The only thing that has changed here is the upgrade to UCM7.0. I'm glad I came across this post since we were informed to upgrade to 22T but obviously this is not going to help us out. When the router does lock up and become unresponsive we are unable to capture any information by way of EEM script due to high CPU utilization, so we have absolutely no idea what is going on. We see the SCCP application is overwelming the CPU resources, but this is it. Memory swap on the router has not resolved the issue so we do not believe it is a hardware issue.

We have also noticed the skinny output you have reported.

Another interesting issue is that not only are we using the CUBE as a DSP farm and for software MTP resources but we also have another 3845 sharing the skinny resource load as well. This gateway has NOT crashed, so it seems to me that:

1. UCM 7.0.x is sending a message that is causing the gateway to crash since 4.x never caused this problem.

2. The SIP stack must be involved here, because our other DSP farm hasnt crashed ever (and its been over 2 months)

Anyone else having the issue? Cisco has filed a bug ID for our issue stating a 15.x release fixes the issue, however, I dont see how that can be determined given such limited evidence.

Paolo Bevilacqua Fri, 02/12/2010 - 13:08

Can you post the actual bug ID?

Also I think you know already that if impacting business, cisco must give you an interim image with a bug fix.


This Discussion