CME Redundancy?

Answered Question
Aug 28th, 2009

In my setup I have Cisco3845 CME routers(C3845-VSEC-CCME/K9), one with CUE module, other with non-CUE(The customer is o.k. not to have voicemails if the CUE-module CME router fails). Both the CME are on same localtion/Lan.

Each router will have one E1 line from PSTN.The Telco sends call to both CMEs in round robin.

How can we achieve CME redundancy in this setup?

1)configuring Redundant Router: Reference CME Admin Guide-Page 117



>>For CME Admin guide, it says on page 144 For configuration information, see the "SCCP: Configuring a Redundant Router" section on page 117.

But under the section:SCCP: Configuring a Redundant Router as a pre requisite it mentions following:

The physical configuration of the secondary router must be as described in the "Redundant Cisco Unified CME Router" section on page 104. >> which means fxo port with Splitter:

2)HSRP>> I heard some issues with phone registration to standby router if active router fails. Also what would happen to incoming calls coming from PSTN>> will they be able to go to IP Phones, since this router is in Standby mode?

Bearing in mind that the Telco is sending calls to both CMEs in round robin, since each CME has one E1 line from Telco.


I have this problem too.
0 votes
Correct Answer by Nicholas Matthews about 7 years 2 months ago

The only reason to run MGCP or SCCP on an FXS port is to have increased line-side features (hold, conference, park, etc) or ease of management (usually with CUCM).

In the case of a fax machine, you will never use the advanced features, and with CME SCCP FXS ports are harder to manage. For a CME system, you're correct and should just use normal Router controlled FXS ports instead of SCCP.

This is an example of a working fax configuration:

dial-peer voice 1 pots

destination-pattern 9.T

port 0/0/0

(outgoing pots dial peer)

dial-peer voice 2 pots

destination-pattern 1000

port 0/1/0

(fxs port with fax machine)


  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4.5 (11 ratings)
Paolo Bevilacqua Fri, 08/28/2009 - 10:41

Yes, you can have redundacy in this manner, however you will need to configure standby CME to send calls to active CME via a lower preference voip DP.

HSRP is not necessary.

Note if you never done this it may take a while for the configuration to be perfect, and you will notice that in practice, router almost never fail.

ra_jeshkalra_2 Fri, 08/28/2009 - 15:09


Can you please elaborable, may be tell the related command if possible for

"you can have redundacy in this manner, however you will need to configure standby CME to send calls to active CME via a lower preference voip DP".

1.My also worry is about incoming calls from E1 on standby CME, will they not get busy tone, or these will go to IP Phones without any problem.

2.What about voicemails for such calls if phone doesn't answer.


Nicholas Matthews Fri, 08/28/2009 - 11:37

Use the 'secondary' keyword on the ip source-address command.

It looks like this:


interface 1


ip source-address secondary

CME 2:

interface 1


ip source-address secondary

This way all the phones register to the secondary by default.


ra_jeshkalra_2 Fri, 08/28/2009 - 15:01

So Nick, all the phones will register to CME1 first, if CME1 fails it will register to CME2. >> Ok

Scenario: The phones are registered to CME1:

Q1.For outgoing calls, will it use CME1 E1 PRI for all calls, or it will load-balance across gateways?

Q2.What happens to the calls coming on CME2 E1 PRI(refer: my setup explaination initially), will they go perfectly to IP Phones directly(as the IP Phones registered to CME1).

Q3.If in Q2 end phone doesn't answer, what about voicemail, will the go to CUE of CME1 or it will try to look for CUE on CME2?? >>Note CME1 has CUE , CME2 has no CUE.

Q4.Suppose CME2 ethernet down(CME1 is still up), what happens to calls coming in from E1 on CME2.

Nicholas Matthews Fri, 08/28/2009 - 15:08

Good questions.

1 - this depends on your dial peer configuration. If multiple dial peers have the same destination-pattern configured, it will load balance. If you prefer one to the other, you can place a 'preference' statement on the dial peer to have one used first.

2 - This CME will have dial peers pointing towards CME1. When the phones register to CME2, the calls will come directly to the phones.

3 - The whole idea of redundancy revolves around if the ethernet or router goes down. In this case, CUE is completely inaccessible. You would not have voicemail in this failover scenario, unless you still had IP reachability to CUE. This is unlikely if you do not have reachability to CME. Even if you had two CUE's, you would not have centralized voicemail. Users would not be able to check the voicemail from the other CUE while in failover, etc. This is a product for this, the UMG. It's unified messaging - but it's more for widely distributed CUE modules and isn't really designed for just 2 CUE modules.

4. Calls that come in to CME2 if the ethernet is down will fail. This is of course, unless you place some type of TDM link (FXO/FXS, T1 PRI) between the two CME in case you would like additional redundancy if the ethernet goes down. This would also solve the CUE problem.

Hope this helps.


ra_jeshkalra_2 Fri, 08/28/2009 - 15:33

Hi Cick,

My setup consist of CM1(1E1+CUE), CM2(1E1). Telco sends incoming call in round robin

For 2. If phones registered to CME1, i need help to clarify, what happens to incoming calls on CME2(via E1), since Telco always send calls to round Robin.

What happens to voicemail , in same case when phones reg to CME1, and few incoming calls come to CME2.

For 4., won't the calls hairpin back to Telco & come as incoming to CME1, >> this scenario I tested with CCM server cluster ans two voice gateways, only thing we will lose the original CLID.

>>So similarly will it work for CMEs setup also??


Nicholas Matthews Fri, 08/28/2009 - 16:30

2 - You create dial peers on CME2. Half the calls will come directly, the other half will come either H323/SIP from the other CME. This is not a problem.

As long as voicemail is reachable, this scenario will have no impact on voicemail.

4 - Generally telco will not allow you forward your own number back out to the PSTN. It may be possible for your telco to re-route the call if we send 'temporary failure' or 'no route to destination' when the ethernet is down. This is telco-dependent. You would not lose CLID in this scenario.

Please rate the helpful posts so that others can identify helpful information.


ra_jeshkalra_2 Fri, 08/28/2009 - 16:44

Hi Nick,

Sorry to bother you again.

For 2. I think you mean, IF phones register to CME1, then calls from E1 on CME2 will go straight to phones since here CMe is acting as a pur H323/SIP Gw.

Is that correct?

For 4. Do I need to give any specific CLI command on CME, to send 'temporary failure' or 'no route to destination'


Nicholas Matthews Fri, 08/28/2009 - 17:14

2 - phone register to CME1. If calls come on E1 on CME1 they go straight to phones. If calls come to CME2 it will travel H323/SIP to CME1, and then to phones.

4 - You will automatically disconnect with 'unallocated / unassigned number' I believe. This should be enough for your telco to reroute.


ra_jeshkalra_2 Fri, 08/28/2009 - 17:17

Gor 2. If calls come to CME2 it will travel H323/SIP to CME1, and then to phones. >> Is this the default behaviour, or I need to do some special config for this.>>> I think I need to isn't it? any commands?

Nicholas Matthews Fri, 08/28/2009 - 18:24

Yes, there are commands. If you do not know anything about SIP/H323, dial peers, or call routing, I would suggest some reading on

Suggested searches:

H323 gateway configuration

SIP gateway configuration

Understanding inbound and outbound dial peers

Communications manager express administrator's guide


ra_jeshkalra_2 Fri, 08/28/2009 - 18:36

Hi Nick,

I know about H323/SIP 7 what you are mentioning.

But how will you forward call from CME2 to CME1? Don't you require IP TO IP GW IOS for that?

I have worked on CCMs & VGs but first time on CMEs.


ra_jeshkalra_2 Sat, 08/29/2009 - 05:40

Hi Nick,

Sorry I think you are indication the pots/voip dial peer config to take the calls from CME1 to CME2.


Paolo Bevilacqua Sat, 08/29/2009 - 13:04

What a shame, seeing so much insisting questioning, many excellent answers from a certified professional, but not even a post rated.

Some people think they're are entitled to have everything for free, and even cheaper.

Wilson Samuel Fri, 09/04/2009 - 08:13

Originally not my question.. however thanks Nick and Paolo for excellent in depth answers.. my 5 Points to both of you guys..

Being a CCIE-V wanna be.. I gain a lot.. every day.. with expert answers like you guys!

God Bless

Sam Wilson

Nicholas Matthews Mon, 08/31/2009 - 06:06

PSTN--PRI--(pots dial peer)--CME2---(voip dial peer)--H323---(voip dial peer)--CME1--(IP phones)

No need for IPIPGW. It's just a simple voip dial peer.

(thanks for the ratings, Paulo)


ra_jeshkalra_2 Fri, 09/04/2009 - 08:06

Hi Nic,

thanks for your help so far.Just few question:

1.Is it a must for more than two party conf. call I should create OCtal Dn, I created dual-line DN, and the call gets dropped if I want to involve the third party.


Nicholas Matthews Tue, 09/08/2009 - 08:32

If you want to have more than 3 people in a conference you need to configure hardware conferencing. This will require DSPs. Take a look at 'Configuring Conferencing' in the CME Admin guide.


ra_jeshkalra_2 Tue, 09/08/2009 - 12:30

Hi Nick,

Thanks for the direction.

If I have to configure fax on FXS port on CME. Is it a must that I control the fax similar to IP Phone by using ephone-dn.

Can't I just configure like a traditional was fax with dial-peers and without SCCP in a normal cisco voice router? The idea is to keep the config simple. I am just looking for sending /receiving one fax at a time.

I just want to implement fax here, like with a normal voice gateway.


Correct Answer
Nicholas Matthews Tue, 09/08/2009 - 13:06

The only reason to run MGCP or SCCP on an FXS port is to have increased line-side features (hold, conference, park, etc) or ease of management (usually with CUCM).

In the case of a fax machine, you will never use the advanced features, and with CME SCCP FXS ports are harder to manage. For a CME system, you're correct and should just use normal Router controlled FXS ports instead of SCCP.

This is an example of a working fax configuration:

dial-peer voice 1 pots

destination-pattern 9.T

port 0/0/0

(outgoing pots dial peer)

dial-peer voice 2 pots

destination-pattern 1000

port 0/1/0

(fxs port with fax machine)


ra_jeshkalra_2 Thu, 09/10/2009 - 09:43

Hi Nick,

Thanks for all your brilliant answers.

1.Is it possible to use E1 of CME2 for outgoing calls if the E1 in CME1 is down(CME2 is failiover to CME1). I was trying to use a VOIP dial-peer to transfer calls to CME2 IP to achieve, with 9T as session target:

a)Surprisingly, if I bring E1 on CME1 down(phons still registered to CME1), on pressing 9, there is no dial tone on phones.

b)Also there seems to be clash with dest-pattern on VOIP dial peer. All callers with their CLID starting with 9 unable to call in to CME from PSTN.

2.How to resolve Music on Hold issue between phones internally, currently on hold music is non-continues and garbled( for a sec I hear garbled music then blank space & then garbled again) between phones for internal calls. But MOH works o.k. from/to calls to PSTN. The phones are enabled with moh-multicast command on CME.Also Lan side, I am told multicast is enabled.

How to resolve this issue of Garble MOH music for internal calls?

Do I have to check on IP phone something for registation to multicast group?

Thanks in advance.

Nicholas Matthews Thu, 09/10/2009 - 09:56

1) Yes this is possible.

You would simply create a pots and voip dial peer with the same destination pattern, and a lower preference value for the one you're wanting to use as the primary.

There should always be dial tone on phones. You may mean secondary dial tone? Secondary dial tone is with "secondary-dialtone 9"

b) I would 'debug voip ccapi inout' and figure out your dial peers. You probably have an incoming called-number 9.T or something similar on a dial peer which you don't need.

2) I would use unicast MoH. In this case, you're likely multicasting from both CMEs on the same LAN. If you disable MoH on one you will probably not see this problem.



ra_jeshkalra_2 Thu, 09/10/2009 - 16:02

1)I noted secondary dialtone is as a result of PRI signalling(when PRI up) from the CME1. The moment E1 on CM1 goes down, the secondary dial-tone also goes down, since phones still registered to CME1.

But with the pots, voip config as suggested, how can we get secondary dial-tone from E1 in CME2?

2)what is do be done for using unicast MOH & not multicast MOH. shouldn't it work by default, if I disable multicast everywhere.

>>Instead of disabling multicast in CME2, can't we give a new/different MOH source IP compared to that in CME1, since phones will register to only one CME at a time.

3)For extn mobility:

a)I have create the extn mobility profile and associated to the phone.But while loggin in extn mobility, I get login failed authentication error.

b)my customer has 150 physical ip phones, but CME3845 can support 250. So is it possible to create additional 100 profiles for people who don't have physical phone, but want to login to any of the above phone without affecting their profiles/association with IP Phones.

Basically the customer wants any user can sit/login on any phone.


Nicholas Matthews Thu, 09/10/2009 - 22:00

1) This is a common misconception. The secondary dial tone is sent to the phone before anything is ever sent to the PRI. The secondary dial tone is a SCCP message that is sent to the phone, not anything related to the PRI. As long as you have the secondary-dialtone command in, you will have this functionality.

2) This command:


no multicast moh

You may be able to get two different multicast streams from two different CMEs to work together. If you can do unicast I would, just for simplicity. I believe two streams will work for you also.

3) Extension mobility is a bit different. I had to file a bug for this specifically because of the problems with the URL when you fail over. Since both CMEs use the same primary address, the Mobility URL is built off of the same IP, which doesn't help for mobility. The bug ID is this : CSCsy21851. It isn't fixed yet, and won't make it into IOS for a while. The option right now is to manually edit the file so that each CME points to itself instead of pointing at a single IP.

I'm not quite sure about the EM part of it.



ra_jeshkalra_2 Wed, 09/30/2009 - 22:54

Hi Nick,

Thanks for your help so far.

As per the attached config, the ad-hoc and meetme conference was working, but now we get an error message "Cannot complete conference" when we try to do meetme or ad-hoc conference.

I suspect this might be due to introduction of some command under telephony-service section.

Any suggestions.

Any debug I can run?

Nicholas Matthews Thu, 10/01/2009 - 04:38

I would compare your adhoc dial peers against those in the CME conferencing section. Specifically, the use of 'huntstop' and 'preference' in terms of using the same DN for the ad-hoc conferences.


ra_jeshkalra_2 Thu, 10/01/2009 - 04:48

Hi Nick,

What is adhoc dial peer here?

I am using the ephone-dn as Octal for adhoc/meetme conference, I am not using preference and huntstop.

(I want only 8-party conference)

The meet-me and adhoc was working earlier, but suddently stopped working.

The change I did was to bring the pstn incoming calls from pots->pots(ip phones), instead of pots->VOIP dial-peer(session target self CME IP Add->pots(ip phones.

But this shouln't cause the problem.

Also can transcoding cause an issue?

(pls see under telephony-service)

I am getting error "Cannot complete conference".

>.I am using ephone-dn 198 & 199 for adhoc confernce(pls see config)

& 186,187,188 for meetme


Nicholas Matthews Thu, 10/01/2009 - 05:32

I mean ephone-dn and not dial peers, sorry. I believe you may need to do change the preference/huntstop issue.


ra_jeshkalra_2 Thu, 10/01/2009 - 05:37

I Nick,

As I mentioned, sicne I am using Octal ephone-dn, I am already getting 8 parties whom I can put in conference, hence I am not required to use preference/huntstop.

Are you able to look the config for something else?

Can transcoding be the culprit?


Nicholas Matthews Thu, 10/01/2009 - 07:29

If you're doing this with shared line DNs,(same DN), you may still need that configuration.

I would check 'show sccp' to make sure it's still registered, and then check your DN configuration against the configuration guide. Generally that's all there is to it.



This Discussion