Cisco MTP resets needed sometimes 2x per week? Is this necessary?

[johngalek]

Hello,

We have a call manager 10.5 and several MTP routers, We are needing to stop any call on each MTP router, reset the router and enable the router, to clear any MTP issues.  We were told at one point by Cisco to login to the CUCM and uncheck the MTP resources required box and that should alleviate the need to reset the MTP routers on a weekly basis?  Can anyone clarify? We have multiple CUCM 10.5 clusters utilizing Multiple Cisco 3825/3945 MTP routers. Any advice or documentation would be greatly appreciated.

[Roger Kallberg]

Normally there should not be a need to have the require MTP check box set. Do you know for what reason it has been set from the start?

On your question about the need to reset the MTPs. That’s not something that is normally needed. If I where to venture on a guess I would suspect that there would be a defect/bug in your version of IOS that makes this needed.

[johngalek]

We have no idea why it was setup this way, it was a previous regime. I know it is checked for our SIP trunks (Cubes) – and when we get busy with heavy traffic the routers cant handle the call volume. I am trying to eliminate the process but I don’t want to take down the business in trying to do so, I am unsure what is using MTP resources- is there a way to find out?

[Les Pruden · Accepted solution]

There are a couple of useful IOS commands you can try.

show call active media compact <-- will show the IP info and the age of the call

show sccp connections  <-- there are a few sub commands that allow you to dig in if needed but this is a good summary command

As noted, you should not need to check MTP on all your SIP trunks. If you do need an MTP you can set up a SIP Profile and apply that to the trunk. Look for this field in the profile.  Big improvements to MTP use in newer versions of Call Manager but this might help you get off the hard MTP check for any trunk that does need an MTP from time to time.

[johngalek]

Hello Les,

Thanks for the information!

I was able to show approx.. 2800 active calls per cube.

No Active sccp connections on any of the cubes