[omniORB] Choosing specific end-points

[Vladislav Vrtunski]

Duncan Grisby wrote:

>On Tuesday 5 April, Vladislav Vrtunski wrote:
>
>  
>
>>We thought we could overcome the problem by handling the end-points,
>>and implementing our own network failure detection mechanism. Our idea
>>was to switch endpoints when we detect (hopefully faster than omniORB
>>does) that active network has failed.
>>    
>>
>
>Why not just set a shorter call timeout in omniORB?
>
>  
>
Because, in that case, it seams that connection is not being closed 
(correct me if I am wrong). When the remote call is repeated, omniORB 
tries to use the existing connection and fails again without trying to 
use the other endpoint.
The problem occurs during lengthy remote calls, when network cable is 
unplugged in the middle of the call.

-- 
Vladislav Vrtunski
DMS Group
Serbia & Montenegro