External calls fails when publisher is down - UCM 7.1.3

[solomonthuo]

Hi pros

I have a UCM 7.1.3 cluster with a publisher and a subscriber. All external calls are routed to some 2 E1s connected to an MGCP controlled 3845 VGW. when i disconnect the publisher by shutting down the connected ports on the switch, i experience the following:

1. I cannot make outbound calls via the GW but inbound calls works ok. I have counter-checked the UCM groups, route lists and route groups. Also the gateway shows that subscriber is registered when i do "show ccm-manager".

2. I have configured extension mobility service URL using a DNS name that returns the IP address of the subscriber and the publisher. when the publisher is down and i try to login or logoff the phone just displays requesting and gets stack there.

any help is appreciated

solomon

[Marwan ALshawi]

make sure the database replacation between the pub and the sub is working by login to the sub and check all

the settings there

then make sure all the required services incuding ccm are activated in the sub

good luck

[solomonthuo]

Hi

thanks for your concern. I checked all the services and they are running. Just some background info, when i checked the cluster summary under unified reporting console, i noticed that some components shows they are not installed,  (check file attached). I built this cluster from  a standalone publisher on UCM 6.1.3,  i backed up then installed 6.1.3 on a new MCS7835, restored the backup then upgraded to 7.1.3. after this i did a fresh install of subscriber and clustered the two boxes. Initially failover never used to work at all and i opened a TAC case, the TAC engineer dropped the DB and recreated again and replication worked. any ideas?

regards

solomon

[karlfisher]

Hi,

Just a thought, have you checked that CUCM group used by the routelist(s) ? If the CUCM group only has the publisher listed, then calls to the routelist will fail when the publisher is off-line.

Regards

Karl.

[solomonthuo]

Hi karl

thanks for your response, this is the first thing i checked when i started testing and its configured properly

[gkelly]

Hi solomon,

I hate to say it but IMO all those components missing and the fact that the database wasnt replicating at the beginning suggests a corrupted install. Did you run a media check before installing? A dodgy network connection during installation can cause problems too (try disabling network failover if you have dual nics).

Might be worth blowing it away again and re-adding it to the cluster. TAC might be able to dig around with remote access and eventually recreate everything, but I would think it will be faster just to rebuild.

Cheers
GK