CUCM Subscriber loses connection to Publisher how to route

[clinton coleman]

We are running cucm 12.5 and I have 1 publisher and 5 subscribers.   One of these subscribers is at a remote site known for WAN issues.  When that subscriber loses its connection to the publisher,  all phones are still registered to the subscriber but my outbound/inbound calling is down.   Station to Station calling works fine.   At this remote site  we have 2 cisco cubes fed by PRIs.  Its my understanding if the subscriber cant see the publisher then it cant see that trunk either.  What are my options for failover for inbound/outbound?  My unstanding is SRST will not kick in unless the phones unregister.   Any assistanceIP Telephony and Phones, CUCM, Unified Communications, Voice Gateways would be appreciated. 

[John Steele]

some quick items.

make sure the CUBE dial-peers point to all subscribers and you check "run on all active ccm nodes" on the route list and SIP Trunk.  Run on all active ccm nodes will help with outbound calls.

You can also change your design a bit with subscribers.  Maybe do 1:1 redundancy model having a primary subscriber and then a failover subcriber that never has anything registered to it; unless its primary is down.  Of course, this may not work for you considering the CUBES are in same location.  Maybe move your CUBES to better location?

[clinton coleman]

The trunk is set to run on all active.  Maybe I'm incorrect and I probably am, but I thought if the subscriber lose connection to the publisher than it has no access to any of the trunks.   How can I direct the outbound traffic from the subscriber that has lost connection to the publisher to the local cube router?

[Roger Kallberg]

As long as the SBC knows about all the CPE nodes in the CM cluster it should work when the publisher is down. How are you matching the call inbound to the router from CM and how do you address the CM(s) in the router for outbound calls to CM?

[clinton coleman]

Not sure I understand your question.  We have route groups to the trunk going to the cubes that are located at the same site as the questionable subscriber.

[Roger Kallberg]

In the Cube you have dial peers, how are you matching the dial peer for calls from CM and how are you sending the calls to CM on the dial peer?

There are multiple ways to do this. What I'm saying is that if the Cube (SBC) knows about the IPs for ALL CM CPEs it should accept calls from any of the cluster nodes that acts as an Call Processing Engine. Look at this document for details on this, Explain Cisco IOS and IOS XE Call Routing 

[clinton coleman]

Below are my configured dial peers on this Cube.   This Cube is using PRI's from the carrier and I also noticed that the voice class server group does not have the subscriber ip address in it.

dial-peer voice 1 pots
description Incoming from PSTN
incoming called-number .%
direct-inward-dial
!
dial-peer voice 10 voip
description Inbound from CUCM
session protocol sipv2
incoming uri via 11
dtmf-relay rtp-nte
codec g711ulaw
fax-relay ecm disable
fax nsf 000000
no vad
!
dial-peer voice 100 voip
description Outbound to CUCM for DIDs
translation-profile outgoing 10D-to-5D
session protocol sipv2
session server-group 11
destination e164-pattern-map 10
voice-class sip profiles 10
voice-class sip options-keepalive profile 100
dtmf-relay rtp-nte
codec g711ulaw
fax-relay ecm disable
fax nsf 000000
no vad
!
dial-peer voice 20 pots
trunkgroup PRI-PSTN
description PSTN Outbound 911
destination-pattern ^911$
no digit-strip
!
dial-peer voice 21 pots
trunkgroup PRI-PSTN
description PSTN Outbound 911
destination-pattern ^9911$
no digit-strip
!
dial-peer voice 22 pots
trunkgroup PRI-PSTN
description PSTN Outbound LD
destination-pattern ^91[2-9]..[2-9]......$
no digit-strip
!
dial-peer voice 23 pots
trunkgroup PRI-PSTN
description PSTN Outbound Local 10D
destination-pattern ^9[2-9]..[2-9]......$
no digit-strip
!
dial-peer voice 24 pots
trunkgroup PRI-PSTN
description PSTN International
destination-pattern ^9011T
no digit-strip
!
dial-peer voice 401 pots
description Incoming from ASCOM
translation-profile incoming ASCOM-E164-IN
shutdown
incoming called-number ^[1204]...$
direct-inward-dial
port 0/1/1:23
!
dial-peer voice 402 pots
description Ascom Outbound Calls
shutdown
destination-pattern ^1[23]..$
no digit-strip
port 0/1/1:23
forward-digits all
!
dial-peer voice 403 pots
description Incoming from ASCOM
translation-profile incoming ASCOM-E164-IN
shutdown
incoming called-number ^9[45]..$
direct-inward-dial
port 0/1/1:23
!
dial-peer voice 404 pots
description Incoming from ASCOM
translation-profile incoming ASCOM-TO-LOCAL
shutdown
incoming called-number ^9270.......$
direct-inward-dial
port 0/1/1:23
!
dial-peer voice 405 pots
description Incoming from ASCOM
translation-profile incoming ASCOM-TO-LOCAL
shutdown
incoming called-number .
direct-inward-dial
port 0/1/1:23
!
dial-peer voice 406 voip
description Outbound ASCOM to PRI
preference 2
shutdown
destination-pattern ^9270.......
session protocol sipv2
session server-group 11
dtmf-relay rtp-nte
codec g711ulaw
fax-relay ecm disable
fax nsf 000000
no vad
!
dial-peer voice 407 voip
description Outbound ASCOM to TollFree
preference 2
shutdown
destination-pattern ^91[2-9]..[2-9]......$
session protocol sipv2
session server-group 11
dtmf-relay rtp-nte
codec g711ulaw
fax-relay ecm disable
fax nsf 000000
no vad
!
dial-peer voice 408 voip
description Outbound to CUCM for Ascom
preference 2
shutdown
destination-pattern [1-5]....
session protocol sipv2
session server-group 11
voice-class sip options-keepalive profile 100
voice-class sip bind media source-interface GigabitEthernet0/0/1
dtmf-relay rtp-nte
codec g711ulaw
fax-relay ecm disable
fax nsf 000000
no vad

[Roger Kallberg]

The server group should contain all three CPE CMs that are in the CPG used by the device pool set on the SIP trunk for the gateway. If there are other CPE CMs you should add them to the trusted list of IPs in voice service voip. The SIP URI list should contain all CPE CMs. If you have all these you should be able to send calls via any of the CMs and not be affected if the Publisher is down.

[clinton coleman]

Looks like the server group has one of the ip addresses wrong for the three CPE CMs  would that cause this issue?   So changing it should fix it?   As for the URI they are all correct.

[Roger Kallberg · Accepted solution]

The IPs in the server group used would be automatically allowed to communicate with the router. If one IPs is incorrect it means that this CM node isn’t allowed to communicate with the router. As mentioned if you have other CPE CM nodes than the once in the server group those should be allowed to communicate with the router by adding these to the trusted list in voice service voip.

[Roger Kallberg · Accepted solution]

Full example of what I mean.

voice service voip
 ip address trusted list
  ipv4 10.64.160.32
  ipv4 10.64.160.33
  ipv4 10.64.160.34
  ipv4 10.64.160.36
  ipv4 10.80.128.37
  ipv4 10.192.21.31
  ipv4 10.192.21.32
!
voice class uri CUCM sip
 host ipv4:10.64.160.32
 host ipv4:10.64.160.33
 host ipv4:10.64.160.34
 host ipv4:10.64.160.36
 host ipv4:10.80.128.37
 host ipv4:10.192.21.31
 host ipv4:10.192.21.32
!
voice class server-group 1
 ipv4 10.64.160.32 preference 1
 ipv4 10.64.160.33 preference 2
 ipv4 10.64.160.34 preference 3
 description Inbound calls from PSTN to SELU CUCM
 huntstop 1 resp-code 404 to 404
!
dial-peer voice 1000 voip
 description Outbound calls from CUCM
 translation-profile incoming NOPLUS-IN
 modem passthrough nse codec g711ulaw
 session protocol sipv2
 incoming uri via CUCM
 voice-class codec 1  
 voice-class sip asserted-id pai
 voice-class sip tenant 1
 dtmf-relay rtp-nte sip-kpml
 fax-relay ecm disable
 fax rate 14400
 no vad
!
dial-peer voice 1010 voip
 description Inbound calls to CUCM
 modem passthrough nse codec g711ulaw
 session protocol sipv2
 session server-group 1
 destination e164-pattern-map 1
 voice-class codec 1  
 voice-class sip tenant 1
 voice-class sip options-keepalive profile 1
 dtmf-relay rtp-nte sip-kpml
 fax-relay ecm disable
 fax rate 14400

As you can tell we have three CM CPEs in the server group, but have more allowed and matched by the URI and the trusted list so that the router will allow communication from all the CPEs.