CUCM Publisher Shutdown

[gayathridissanayaka]

Hi,

We have a testbed CUCM network. In the network we have one publisher, and two subscribers. We are intended to use CUCM publisher for administration purposes only and the devices are registered to other subscribers accordingly via two device pools.

My question is that whilst there is an incoming call coming in to the devices in this network, what happens if we shutdown the CUCM Publisher only.

I was expecting the calls to still ring in to the CUCM even when the CUCM Publisher was shutting down. But what we experienced was that the incoming calls stopped ringing when the publisher shutdown command was given. 

Is this the normal behavior or should we be seeing calls coming in to the IP Phones when the publisher is shutting down.

It is much appreciated. Thank you

Kind regards,

Gayathri

- See more at: https://supportforums.cisco.com/discussion/12541941/cucm-publisher-shutdown#comment-10598741

[Vivek Batra]

With appropriate configuration in place, there should not be any impact on call processing when PUB is down as SUB is independent call processing entity in a cluster.

What is exact call flow which failed when PUB is down?

Thanks

Vivek

[gayathridissanayaka]

Thanks for the confirmation. What it is is that we have analog signal phone telephones connected to a Meridian PABX and ApplianX gateways to interconnect this to the IP Network. ApplianX gateway does the DPNSS to SIP conversion. So in this scenario the type pf the trunk registered to the CUCM is a SIP trunk. Once the calls enter in to the CUCM they will be  pointed to a hunt pilot, then hunt list and line groups then the devices added to the line group will have the incoming calls, as in a normal scenario.

We wanted to test the impact on the publisher to the network.ie we made some incoming calls to these devices and whilst the call is presented on the front end device, did shutdown the publisher server, and the results was whilst the CUCM's been shutting down, the incoming calls stopped ringing and wasnt presented on the front end IP devices.

As we want only the subscribers to call handle we assigned the devices and the trunks to device pools which dont have publisher IP in it, and still getting the above issue.Thanks

Gayathri

[Vivek Batra]

As well described by Terry before, this should definitely work even in absence of PUB. Please analyse the complete flow that there is no dependency of PUB during a call. As you said, you already taken care of it in device pool. I'm sure you meant to say CM group under DP only has SUP, no PUB there. Also check not only on phones, you have similar DP configuration on SIP trunk, Hunt list etc as well.

Thanks

Vivek

[gayathridissanayaka]

Hi all .

I found that the Hunt List and the Route List have been assigned with the default CM group which had the publisher on. Somehow, when assigning the CM group with two subscribers that we need to do call processing to the hunt list, incoming calls weren't disrupted. I have also assigned the same CM group to the Route List, and now when the publisher is shutdown, I am unable to make any outgoing calls at all. It doesn't even reach the Applianx gateways. I also want the two subscribers to load balance the outgoing calls. What configuration actually decide , what subscriber to use when making an outgoing call please? Much appreciated.

[Terry Cheema]

Ok - Check confirm below things again:

1) Route List is registered to the Sub server

2) From your SIP trunks as I had mentioned in my previous post - make sure the checkbox against "Run On All Active Unified CM Nodes" is checked

3) Lastly on your ApplianX make sure you have entries for Sub servers as well and its configured to accept the calls from sub servers.

Please rate all helpful posts.

-Terry

[gayathridissanayaka]

Hi,

I wrote a reply before and It doesn't show it in here so I'm writing again. 

I have configured as you have stated above 1,2, and 3. Still when the publisher is shutdown I cannot make any calls .Some how when I enabled the "Run On All Active Unified CM Nodes" on the Route list(you have only mentioned the trunks) it worked, it let me make outgoing calls. Can anybody explain the difference in enabling this on both routelist and trunks and only on trunks. I also would like to know if it is possible to identify or predict which CUCM node will be utilised for call processing? Comments are appreciated. I have attached a .jpeg of the outbound cm group and device pool sections. Thank you

[Wilson Samuel]

Hi Gayathri,

The general rule of thumb is that, all attributes of the database (be it phone or trunk or anything else) has the R/W copy on the Publisher. Hence, if you do not select the "Run on All Active Unified CM Nodes" any time any communication happens that utilizes RL/RG/Trunk, the Subscriber checks on its own database and then forwards the request to process (for RL/RG/Trunks) to the Publisher

Since in your case, you are removing the Publisher from the equation (CCM Group) I think that is what is causing the issue.

http://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/session_mgmt/deploy/8_5_1/trunks.html#wpxref55318

HTH

[Terry Cheema]

Make sure ApplianX is configured to send calls to sub or if multiple destinations are speicified sub has higher priority.

In CUCM SIP trunk - make sure run on all active nodes is checked.

-Terry

Please rate all helpful posts

[Terry Cheema]

Hi Gayathri,

There should be no impact to the call processing at all. As long as there is no device registered to the Publisher, you should see no impact on any inbound/outbound calls.

If you are having any issues, check everything starting from devices, route lists/trunks/gateways etc. and make sure nothing is registered to the publisher. Also check from your gateway/dial-peers are sending calls to the Sub servers instead of Publishers.

-Terry

Please rate all helpful posts

[gayathridissanayaka]

Hi Terry , 

Thank you for your answer. I shall check and getback with the results. 

Gayathri