we have 3 mailbox servers which also include CAS servers and HLB which points to those CAS servers other than that we have separate two CAS servers in different site for external client connections. 

in implementation phase all DNS entries are mailflow pointing to exchange 2010.

while testing server failover database automatically gets mount on other server but outlook clients goes haywire.

connection goes to exchange 2010 cas array which is quite unusual.

we have separate internal url for intenal.domain.com which is published internally and external use on internal CAS servers for

autodiscover EWS, OWA, OAB, ECP.

and there is DNS entry for internal.domain.com which points to HLB points to 2013 CAS.

still not able to understand what happens when one f the server goes down and why users are unable to connect?  

Hi

Are all your DNS entries pointing to Exchange 2013 as it will handle traffic and proxy to the older versions in your org?

Exchange 2013 doesnt work with cas array's anymore. How are your outlook clients connected? is it to the cas array name which can be mail.domain.com for example or to a single server?

When you do the failover and ping the dns entry what happens? Does OWA work at all?