[cisco-voip] MWI Collisions

Wed Mar 12 20:02:08 EDT 2008

It sounds like the problem really was that the ports unregistered, and 
that's what you need to get to the bottom of (look at the link between 
unity and ccm, high cpu on either host? etc), and the MWI collisions are 
a bit of a red herring. You've got to keep in mind that, an MWI 
collision isn't really possible in a skinny integration. The term is 
really relevant to analog integrations, where it is entirely possible 
that Unity could be trying to go off hook to dial an MWI only to find 
that, in that instant, someone had just called into that same analog 
line. In a skinny integration, where the call setup and signaling is all 
done out of band, that can't happen.

That said, I've definitely seen that error before (usually, just like 
you, many many times a second), but it has always been with CME 
integrations, due to a bug where CME couldn't process requests as fast 
as Unity expected it to. Of course, that's not too relevant here.

If you've got skinny traces from Unity along with detailed callmanager 
traces, TAC should be able to give you a good idea of what happened. 
Without those, though, you're not likely to be able to find much.

-------- Original Message  --------
From: Joel P <tman701 at gmail.com>
To: Pat Hayes <pat at wcyv.com>
Cc: "'Jorge L. Rodriguez Aguila'" <jorge.rodriguez at netxar.com>, 
"cisco-voip at puck.nether.net" <cisco-voip at puck.nether.net>
Subject: Re:[cisco-voip] MWI Collisions
Date: 3/12/2008 7:39 PM

> Hey Pat,
> 
> You are right about the 25%. I never really paid too much attention to it,
> but this time I did. I was watching the port status monitor and out of 16
> ports, 4 of which are assigned for MWI, only 2 ports out of the 4 were used
> during the resync. This particular customer has about 200 users and they are
> an insurance customer so I was surprised just how fast MWi dialing was
> happening and on only 2 ports.
> 
> As for why my calls weren’t getting thru. In this case the customer wanted
> Unity to handle all inbound calls thru call handlers. Only after they have
> completed certain steps would it be handed off to a receptionist or to the
> appropriate party.
> They are on CUCM 5.1 and Unity 4.2 don’t have the version of TSP handy right
> now. While this was happening internal callers would get a fastbusy when
> trying to reach VM.
> 
> TAC still hasn’t gotten back to me with anything yet. I went ahead and
> installed the DST patches anyway.
> 
> Joel P
> 
> -----Original Message-----
> From: Pat Hayes [mailto:pat-cv at wcyv.com]
> Sent: Wednesday, March 12, 2008 7:23 PM
> To: Joel Perez
> Cc: Jorge L. Rodriguez Aguila; cisco-voip at puck.nether.net
> Subject: Re: [cisco-voip] MWI Collisions
> 
> The 25% allocation is a bit of an old recommendation, based mostly on
> PBXs where an MWI attempt take a a while to dial out. Take a look at a
> callmanager integration, watch in port status monitor during a resync,
> when things are moving as fast as they can, how long your MWIs take to
> dial out and how many ports they use. They complete in a fraction of a
> second, and even on busy 96 port systems, you'd be hard pressed to see
> Unity need more than 4 ports for MWI dialout. At 25%, you'd be
> dedicating 24 ports. The limiting factor is really how fast exchange can
> pump the data to exchange. Not trying to tell you guys you need to
> change everything, just something to consider.
> 
> These days, I usually suggest around 15-20% total, depending on usage
> patterns, but to also designate those ports for TRaP and notification
> dial out as well. That way you've got all of your outdialing happening
> in one block, which you of course exclude from your line group. I'm not
> sure what the latest design guide says, but it'd be worth a look.
> 
> As for Joel's original question, could you provide a little more detail
> on the configuration? I assume since we're talking about Unity and calls
>   through your PRI not completing that these calls route through Unity
> initially? During that same period, could internal users call Unity?
> What version of Unity? Callmanager (or CME?)? What version of TSP on Unity?
> 
> -------- Original Message  --------
> From: Joel Perez <tman701 at gmail.com>
> To: "Jorge L. Rodriguez Aguila" <jorge.rodriguez at netxar.com>
> Cc: "cisco-voip at puck.nether.net" <cisco-voip at puck.nether.net>
> Subject: [cisco-voip] MWI Collisions
> Date: 3/12/2008 12:31 PM
> 
>> Hey Jorge,
>>
>> Actually I do have 25% of my ports assigned to only MWI and they arent
> part of the VMHuntGroup.
>> That is why i opened up the TAC case. Calls should not be going to these
> ports at all.
>> Im kinda going to guess that since the ports between Unity/CUCM
> de-registered at around the same time then that would cause Unity to use any
> available registered ports for inbound calls also. But that is only a guess,
> still waiting on TAC to get back to me.
>> Joel P
>>
>>
>> On 3/12/08, Jorge L. Rodriguez Aguila
> <jorge.rodriguez at netxar.com<mailto:jorge.rodriguez at netxar.com>> wrote:
>> Do you have Dedicated MWI ports? It looks like you don't. As a rule of
> thumb, you should dedicate 25% of your ports to MWI, Meaning you do not send
> calls to these ports on the auto attendant Hunt Group. On the Unity, you
> should also set them to MWI only. This Should fix your problem. Keep in mind
> that by doing this you are effectively reducing the amount of port that will
> answer and depending on you port usage, you could be in need of adding
> ports. Set up 4 ports for mwi and 12 for Voice calls, If 12 is not enough,
> your next step should be to add ports, no to reduce the amount of MWI
> dedicated ports.
>>
>>
>> Hope this Helps;
>>
>>
>>
>>
>>
>> Jorge Rodríguez Aguila
>>
>> CCNA,CCVP
>>
>> Senior Network Consultant
>>
>> Netxar Technologies
>>
>> jorge.rodriguez at netxar.com<mailto:jorge.rodriguez at netxar.com>
>>
>> Office 787-765-0058
>>
>> PCS 787-688-8530
>>
>> [cid:image001.png at 01C88434.9CDB9060]
>>
>>
>>
>>
>>
>>
>>
>> From:
> cisco-voip-bounces at puck.nether.net<mailto:cisco-voip-bounces at puck.nether.net
> [mailto:cisco-voip-bounces at puck.nether.net<mailto:cisco-voip-bounces at puck.ne
> ther.net>] On Behalf Of Todd Franklin
>> Sent: Wednesday, March 12, 2008 9:39 AM
>> To: cisco-voip at puck.nether.net<mailto:cisco-voip at puck.nether.net>
>> Subject: [cisco-voip] MWI Collisions
>>
>>
>>
>> Yesterday we had a failure with the system whereby calls coming in on the
> PRI were not happening, but we could call out on the PRI.
>> The entire event lasted maybe 15 minutes.
>> When I looked at the Unity event viewer, I see many THOUSANDS of these
> during that timeframe (on average it looks like 60 every second were
> logged):
>> "Cisco Unity-CM TSP device 20 (Cisco Unity port 16): An attempt to turn ON
> the message waiting indicator (MWI) for extension 1212 failed because a
> collision occurred with an incoming call on the same port.
>> The MWI request will be retried. But to prevent collisions, we recommend
> that ports setting MWIs be isolated from ports handling incoming calls. If
> the MWI status remains unchanged for an extended period of time or if there
> are many of these warnings from Cisco Unity in a short period of time, there
> may be an MWI misconfiguration or another problem."
>> ---
>>
>> Any ideas anyone?  Unity looks quiet today, but this was a bizarre failure
> right out of the blue.  If I need to do something to avoid a repeat of this,
> I'm more than willing!!
>> Thanks!
>>
>> _______________________________________________
>> cisco-voip mailing list
>> cisco-voip at puck.nether.net<mailto:cisco-voip at puck.nether.net>
>> https://puck.nether.net/mailman/listinfo/cisco-voip
>>
>>
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> cisco-voip mailing list
>> cisco-voip at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-voip
> 