Exchange CASARRAY and mailbox servers client timeout

Hi, i pretty new to F5 and liking what i see but have a slight problem on realtion to users connections on when we fail over the backend mailbox servers. We have 2 CAS servers in a casarray which all traffic come throught the Ext and internal F5 Virtual load balancers, however the issues are premerly which internal connections on the internal Load balancer.

 

We have a one rule setup for the casarray :-

 

iRule to select pool and persistence method when all Exchange

 

Client Access services are accessed through the same BIG-IP

 

virtual server.

 

when HTTP_REQUEST {

 

switch -glob [HTTP::path] {

 

 

"/Microsoft-Server-ActiveSync*" {

 

Direct all ActiveSync clients to a common pool; use

 

HTTP cookie persistence

 

persist cookie

 

pool Exchange_2010_CAS_single_as_pool

 

HTTP::class disable

 

}

 

 

"/rpc/rpcproxy.dll" {

 

Grab all requests for Outlook Anywhere; the following

 

checks assign correct persistence methods.

 

switch -glob [HTTP::header "User-Agent"] {

 

"MSRPC" {

 

This User-Agent section matches most versions of

 

Outlook and Windows using Outlook Anywhere.

 

The OutlookSession cookie is new to Outlook 2010.

 

if { [HTTP::cookie exists "OutlookSession"] } {

 

persist uie [HTTP::header "OutlookSession"] 3600

 

}

 

else {

 

persist uie [HTTP::header "Authorization"] 3600

 

}

 

}

 

 

"*Microsoft Office*" {

 

This section matches some versions of

 

Outlook 2007 on Windows XP

 

persist uie [HTTP::header "Authorization"] 3600

 

}

 

 

default {

 

This section catches all other requests for

 

Outlook Anywhere, and sets a persistence method

 

that does not require the client to support

 

HTTP cookies

 

persist source_addr

 

}

 

}

 

 

Finally, this assigns the Outlook Anywhere pool and turns

 

off full HTTP parsing and compression. If the preceding

 

clients should be sent to separate pools, the pool statement

 

should be removed here, and a separate pool statement

 

placed in each of the preceding logic branches.

 

Other modules (APM, ASM, etc.) should be disabled here

 

as well, if active for other traffic though this virtual

 

server.

 

pool Exchange_2010_CAS_single_oa_pool

 

HTTP::disable

 

COMPRESS::disable

 

HTTP::class disable

 

}

 

 

"/xml/autodiscover.aspx" {

 

Requests for Autodiscovery information. The selected pool

 

might be unique, or might be the same as e.g. your pool

 

for OWA or ActiveSync. In this example, we use the same

 

pool that receives ActiveSync traffic.

 

persist cookie

 

pool Exchange_2010_CAS_single_ad_pool

 

HTTP::class disable

 

}

 

default {

 

This final section takes all traffic that has not

 

otherwise been accounted for and sends it to the

 

pool for Outlook Web App

 

persist cookie

 

pool Exchange_2010_CAS_single_owa_pool

 

 

If using the Web Accelerator module, uncomment the

 

following line and change the name to that of

 

your WA class.

 

HTTP::class select Exchange_2010_CAS_single_wa_http_class

 

}

 

}

 

}

 

But when we fail over to another mailbox server there is a delay for reconnects, we are using onconnect profiles but not SNAT??? and also we have aBES server and 6 months ago i had to change the time out on each persistant uie rule from 7200 to 3600???? would the time have anything to do with client re-connects taking so long to reconnect to another mailbox server???

 

 

I ran a few test on realtion to the outlook client and i added a in my local host file the ip address to one cas server... and it worked fine.. but whe i changed back to the casarray VIP address on the internal LB - it toom arounf 10 minutes to reconnect???

 

ITs not a major issue, but if anyone can spot a mistake on the irule of know of this issue i would be so gratefull :-)

 

 

Huwster :-)