do we have to reset MGCP gateway after T1 down/up

[Eric.Wang]

hi! we have a T1 CAS circuit (on 2821 MGCP GW) went down but came up shortly. But the problem is on CallManager, under Gateway configuration, the end point associated with this T1 is

Registration: Unknown

and there is snmp trap:

Message: snmp trap ccmGatewayFailed

1.should we have to perform"reset gateway" at end point level (not the whole router reset)

2. is it normal MGCP behavior?

[brandoncheah]

I would check the MGCP gateway's name and make sure it matches exactly with Call Manager. If the names do not match exactly, it will never register.

[Eric.Wang]

there are other ISDN PRI on the SAME router and they are all registered with CM properly. and there is no change on hostname and domain.

It seemed the GW is not trying to register with CM for this T1 after it's up again.

[brandoncheah]

If you do a "sh isdn stat", does that T1 that went down show "mulitple frame established"?

[Eric.Wang]

but it's not PRI isdn. it's T1 CAS

[brandoncheah]

I apologize...can you attach a sh run?

[Nicholas Matthews]

A couple things to check:

-Does the pots dial peer for the CAS circuit have 'service mgcp' under it

-Does 'show ccm | i Name' match what you've defined in CUCM

-Do 'show voice call summary' and 'show voice port summary' show that the T1 is idle and up?

-Is 'show controllers t1' showing the controller up?

Otherwise, 'debug mgcp packet' may be helpful.

-nick

[Eric.Wang]

from CM, I "reset gateway" on this T1 only. Now this T1 is registered with CM.

but this is not normal. The GW is not trying to register with CM for a T1 that has been restored..

Aug 11 17:11:00.451 EDT: %LINK-3-UPDOWN: Interface recEive and transMit0/0/1:1(24), changed state to up

GW#

Aug 11 17:11:08.059 EDT: %CMAPP-6-CONFIG_DONE: Configuration by CCM is done

[Nicholas Matthews]

Hi Eric,

It would most likely be because the T1 was in a bad state, which is why you would want the show commands to see which state it was in.

"Only from CUCM" is very misleading. When you do this, it re-configures the voice ports and shuts/no shuts them via TFTP. So you only touch CUCM, but you're still resetting things on the gateway.

-nick

[Eric.Wang]

no. this T1 CAS was fine because later we were able to make LD calls and debug vpm signal showed activities.

yes. you're correct CM just instructed GW to reset the T1.

my concern now is that if it takes manual reset to register the T1 to CM, is it a normal behavior?

[Nicholas Matthews]

No it's not. I would get the show commands and open a TAC case if necessary. If you're on the latest version you'll get less hassle, because there's a pretty good chance someone will ask you to reproduce it on the latest version.

-nick

[Eric.Wang]

Yes. I opened TAC case and TAC is looking into it.

[paolo bevilacqua]

Why don't you take the easier route and use H.323 instead.

You will find that it works more predictably and has more features than MCGP.

MGCP was invented to shift intelligence from gateway to server, unfortunately in the cisco implementation it seems like the intelligence has been lost during the move.

[veerappan]

Hi,

We have encountered a similar issue, what were TAC's findings and conclusion on this issue.

Thnaks,

VV