Unity Failover Questions/Recommendations

[dtran]

I have an IPT DR (Disaster Recovery) environment with a single CallManager cluster and Unity at HQ office and DR office. I have two CallManager servers (one Pub and one Sub), a Unity primary server and Exchange 2k server located at HQ office and at DR office I have 2 CallManager Subcribers and a secondary Unity server. And between two sites is a full DS3 connecting the two sites. My questions and issues are.

- I just recently configured failover on primary and secondary Unity server but for reasons secondary Unity server went active on its own twice since. Can someone please tell me why this happens ? I am confident CallManager is configured properly.

- And when secondary server is active users were experiencing delay when retreiving voice messages. Is this a normal behavior since the message store is located at HQ office ? for example: when users press the messages button they can hear system prompt "please enter your password" properly but when they enter the password and press pound there is a delay before they can their name annoucement and the users are located at HQ office.

- And which CallManager should I configure the voice ports on the secondary Unity server to register to ? CallManager in HQ office or CallManager in DR office ?

- Could replication issues between primary and secondary Unity trigger the failover ?

I am running Unity 4.0.3 sr1 with Exch2k and CallManager 4.0.2a sr1.

Any helps/recommendations is appreciated !!! Thank you in advance !!!

Danny

[gpulos]

failover would kick in due to the primary/secondary servers not getting their status from one another. (heartbeat) this could be due to many things including nodeManager failing to send keep-alives to the secondary, network failure/delay, manual/forced, nodeMgrSvc stops, and more....

see the following link for more info:

http://www.cisco.com/en/US/products/sw/voicesw/ps2237/products_feature_guide_chapter09186a00801b9253.html#wp1159389

[Tommer Catlin]

You can adjust the trigger to manual if you like. Automatic can freak out sometimes and flip by iteself. You will want to upgrade to a newer Unity version to help this out also. 4.0(5) with patches seems to work the best. (and later versions) But if you have Unity failover across the WAN, you may want to create a trigger that really signals that the server is offline because of xyz... if your heartbeats drop, it will shutdown primary always.... and what you do not want to get is the flipping back and forth of the Unity servers... it confuses the heck out of CCM when it does it so fast so many times. The TSP's seem to lock up or start acting funny.

Callmanager is probably configured fine. IT's the Unity triggers. Check also your network cards are configured in Fault Redudant and not load balance... this will trigger failovers automatically also when its not needed.

If your DNS is slow repsonding, this will trigger failovers. SQL is trying to "talk" to the other SQL server and if it cant resolove the name.. bam... failover.

hope this helps!