Failover Problem with CallManager Servers

[ciscofreak1241]

Hello all,

I setup a Publisher and a Subscriber. The Subscriber is the the primary CCM for Call Processing and the Publisher is the backup. My DHCP on the network is configured with two OPTION 150 IP's with both IP's of the CCM (The subscriber ip is first then the publisher is second). So, here is the situation:

When I manually fail the Subscriber, to force the IP Phone's to re-register with the backup CCM (publisher) it states that it is trying to OPEN the First IP Address listed in OPTION 150 and SRST (which is the default gateway). So, it is not going to the second IP in the OPTION 150 configuration. So, the IP Phone is not registering because of that.

My workaround was to create a SRST reference to the backup CCM (publisher). Doing that it does fail over to the other CCM. So, based on examples there are only one IP listed for OPTION 150 and not two. I'm thinking that two IP's listed do not work.

So, my question is if that is what I needed to do, which was add an SRST reference or is that something I should not be doing? If so, please let me know what should be done correctly for setting up proper failover between a backup and a primary CCM. Any information, websites, resources, etc would be great!

Thank you

ciscofreak

 

[mtashiro]

Option 150 should just have the Publisher IP. Once the phone "talks" to the publisher it will automatically get the IP of the Subscriber and the SRST device.

 

[ciscofreak1241]

Thanks for the response.

Adding only the Publisher IP Address into Option 150, does not register with the Subscriber and registers with the Publisher. However on the IP Phone under network configuration does see the Subscriber as the primary CCM.

I configured the CCM Group and ordered it with subscriber first then the publisher.

That is why I configured the subscriber to be the IP in option 150 and configured a SRST reference to point to the Publisher. That's the only way that seems to work well with failover.

If there is another way that would work please let me know?

Thanks

ciscofreak

 

[BuckWeet]

ciscofreak that is not correct.. mtashiro has it right..

the option 150 just defines a TFTP server address, which the publisher has a tftp server on it.. all the phones do is go get this config file and based within this config file is the parameters defined to the phone..

so if you have the ccm group order right and its not working, then there are issues elsewheres.. you need to verify that the subscriber is up and working properly..


BuckWeet

 

[ciscofreak1241]

Ok, no problem that is what I needed to know. Thanks.

The Subscriber is up and working with all services running (CallManager, TFTP, so forth). Call Processing from the Subscriber is working good. The Publisher see's the Subscriber even when IP Phones register with the Subscriber, I see them listed under devices from the Publisher. My IP Phone network configuration see's Subscriber first then Publisher second. So, I believe there is more that is needed to be done than just ordering the CCM by priority in the CCM Group configuration.

So, what URL reference on the Internet can I use to know what I need for setting up failover. There are no good references on the Internet that talks about this. I have reseached a few IP Telephony forums about this and a few people have done exactly what I have done for failover, which is the OPTION 150 with the Publisher, and it doesn't work. So, at least there are other people who are experience the same thing. We understand the theory and what suppose to happen including the configuration and TFTP process. So the theory we understand, but it's happening or working that way for us.

Has anyone successfully configured failover for at least two CallManager Servers?

If so, please let me know your suggestions and what you did to get it to work whether it is a website, even a book, or your personal experience with getting this work with a victory party afterwards. I know that Callmanager literature is extremely limited on the Internet and in press, so any help would be fanatasic.

Any information and help with this would be great!

Thank you

ciscofreak

 

[BuckWeet]

We have an 8 CCM cluster setup, and it works properly..

I would try to delete the CCM server from the cluster, then recreate it within the ccm admin page..

also make sure the MAC address and such of the CCM server is correct..


BuckWeet

 

[uddinabu]

Hey,
Could anybody please tell me what will be the backup database file size(Gigabyte) for CallManager and also file file for Unity voicemail size?

Thanks in advance

 

[pndscm]

ciscofreak, turn off the TFTP server from the Subscriber. You only need one TFTP server per cluster.

There is no other configuration for failover than to order the servers. If it is not working, there are more problems as was stated above.

Be sure your device pool settings are correct for the phones, also. Make sure you are up-to-date with patches, etc. If all else fails re-boot both servers, Pub then Sub and then see what happens.

 

[ADB100]

What happens when the CCM running the TFTP server fails then? As I see it from what you are saying there is a single point of failure here - the TFTP server?

I thought option 150 was supposed to be an array and have multiple IP addresses listed so that in the event of a failure any new phones or phones rebooting can continue to work using the 2nd TFTP server?

Have I missed something here?

Thanks

Andy

 

[Wavesg]

You can configure another server to run as an TFTP backup, but really if your publisher fails you would want to stop any changes from taking place since your main database is down. The TFTP service is needed only when adding new phones or changes are made that affect the global cluster.

Stay cool; it’s not over yet!

 

[ADB100]

Wavesg - you say that the TFTP server is only used when adding new phones - I thought just booting an IP Phone requires the TFTP Server to be available since this is where its configuration is held?

I am thinkinf from a design perspective you would normally build redundant services such as DHCP Servers, DNS Servers etc. I would have thought the TFTP Service fell into this catagory? Certainly in deployments I have seen there are maybe three CallManagers - a Publisher and two Subscribers. These are generally located around the campus along with redundant DHCP & DNS Servers. IP Phones in one area would be 'homed' to one of the Subscribers with the other Subscriber as the 2nd choice CCM Server, the Publisher being the 3rd. The DHCP Servers would hand out the TFTP server addresses: the 'local' CCM Subscriber being the primary and the 'remote' CCM Subscriber being the secondary (Publisher being the 3rd).
This way you can distribute your call processing between the two Subscribers, plus have failover in the event that a Subscriber fails.
I am sure in the Cisco AVVID Guidelines it describes redundant TFTP services as well as DHCP & DNS?

Thanks

Andy

 

[Wavesg]

I guess what I'm trying to say is that if you lose your Publisher it would be safe to stop any changes on your cluster until you can restore the main dbase in this case the Publisher. Remember you can only have one publisher in on cluster. If you want to have a backup TFTP, DHCP or DNS server you can do so, but anyone would be more concern in getting the publisher back up on line then worrying about adding a phone. I would be more worry in getting the right back up to restore the publisher than where is the new phone getting the next IP address of the TFTP server. It's just a thought

Stay cool; it’s not over yet!

 

[HmmG3]

Andy,
The DHCP Option 150 is only for telling the phone where to download its configuration (TFTP server), not which CCM to use. When the phone boots up, it talks to the TFTP server and downloads a list of IPs. This list tells the phone which CCM to log into as the primary, along with a list of backup CCMs. When you configure the phone's Device Pool, there is a specification for Cisco CallManager Group. This is where you specify the CCM list.

So if your Publisher/TFTP server goes down, any phone powering up or rebooting won't register. However, existing phones have already downloaded the list and will continue to operate. Of course if your Publisher goes down, you have more to worry about then a few end user phones that won't register!

 

[ciscofreak1241]

I solved this issue a couple of weeks ago on my own by going down to the basics. It was a DNS issue.

When I looked at the Network Configuration on our IP Phone it said that Call Manager TFTP was "Active", but there was nothing listed next to the Publisher nor the Subscriber. So, I thought about it and said that it probably can't resolve the name because the DNS I was using was external. Anyhow I created DNS entries and now the phones shows that the Subscriber is "Active" and the Publisheris "Standby". The failover works!

It's a little disconerning that no book or reference talk about the importance of this with DNS espeically with failover and other services. I worked with the Cisco TAC and the results were the same with no luck. Luckily this is resolved.

So, DNS is extremely important if you want your IP Phones to resoslve the names of your CCM Servers for communication.

ciscofreak