[cisco-voip] Inter cluster trunk calls dropping at 10 minutes since upgrade to 7.0.2

[Wes Sisk]

That should work around the issue but will not "fix" it.  Issue is the 
h225 TCP session is getting aborted by TCP FIN or RST.  Allowing h.323 
call preservation will mask the issue by allowing calls to stay up.  
However, there will be no supplementary services for those calls such as 
hold/transfer.  Call will be in the same status as when a call is active 
and phone changes to "cm down, features disabled".

Additionally, preserving h.323 calls is likely to leave abandoned active 
RTP session active between endpoints in the network.  Calls can be 
nailed up indefinitely until endpoints are rebooted or calls forcibly 
cleared.  If the call goes out a gateway on remote site as an LD call 
to, for example, a conference service, you could run up one whopping 
phone bill.  When call is "preserved" CM gives up on all management of 
the call so even CM's "max call duration" does not tear the call down 
after 24 hours.

/Wes

On Tuesday, September 08, 2009 9:41:55 PM , Erick Bergquist 
<erickbee at gmail.com> wrote:
> FYI,
>
> Been awhile, but setting the H323 Peer Preserve Service parameter on
> both clusters fixed this issue for me.
>
>
> On Thu, Aug 13, 2009 at 11:47 AM, Erick Bergquist<erickbee at gmail.com> wrote:
>   
>> FYI,
>>
>> This is line from the CCM Detailed trace where it disconnects due to
>> abnormal TCP Session disconnect.
>>
>> 08/10/2009 11:08:21.108
>> CCM|H225Cdpc(0000476)::active10_TcpStopSessionInd: H225 Tcp session
>> terminated abnormally
>> |<CLID::StandAloneCluster><NID::10.12.5.2><CT::4,100,65,1.
>> 11216924><IP::10.12.5.2><DEV::SEP00DE735C201><LVL::Error><MASK::0100>
>>
>> Cisco TAC also suggested to set the H323 Peer Preserve value to True
>> on both clusters which we changed today. The H323 TCP Keepalive is
>> also to true on both sides (was).
>>
>>
>> On Wed, Aug 12, 2009 at 1:35 PM, Erick Bergquist<erickbee at gmail.com> wrote:
>>     
>>> No, there is no timeout set to 10 minutes for media exchange timeout I
>>> could find.
>>>
>>> I have opened a TAC Case on this with the CCM traces I collected on a
>>> problem call.
>>>
>>> Mike O, any success on your similar issue?
>>>
>>> On Fri, Aug 7, 2009 at 10:44 AM, Wes Sisk<wsisk at cisco.com> wrote:
>>>       
>>>> check your CM service parameters to see if you have anything like media
>>>> exchange timeout set to 10 minutes.
>>>>
>>>> alternatively you can look at CM SDI and SDL traces to see what timer is
>>>> firing.
>>>>
>>>> Do you have any firewall or NAT between clusters?  Have you enabled
>>>> H.225/h245 TCP keepalives (CM service parameter)
>>>>
>>>> /Wes
>>>>
>>>> On Thursday, August 06, 2009 9:42:42 AM , Erick Bergquist
>>>> <erickbee at gmail.com> wrote:
>>>>         
>>>>> Anyone know of any issues with calls dropping across a Intercluster
>>>>> trunk (Non Gatekeeper controlled) between a CUCM 6.1.2 cluster and a
>>>>> 7.0.2 cluster? The calls stay connected for 10 mins and then
>>>>> disconnect. The disconnect occurs when a user on the 6.1.2 cluster
>>>>> calls someone on the 7.0.2 cluster and  not the other way. This
>>>>> problem did not exist before the one side was updated to 7.0.2. I've
>>>>> compared the trunk configuration on both sides, etc and all that looks
>>>>> good.
>>>>>
>>>>> Anything change in 7.0.2 trunk-wise I may need make adjustments for?
>>>>> Just posting here quick before I gather traces and open a TAC case.
>>>>>
>>>>>
>>>>> Thanks, Erick
>>>>> _______________________________________________
>>>>> cisco-voip mailing list
>>>>> cisco-voip at puck.nether.net
>>>>> https://puck.nether.net/mailman/listinfo/cisco-voip
>>>>>
>>>>>           
>>>>