In poking around the Server Administration API I found that for a map service there are (allegedly) two values for the loadBalancing property: ROUND_ROBIN and FAIL_OVER. Round Robin is the default and I understand what that does but I haven't found an explanation of what the Fail Over value actually does.
So I figured I'd just try it out and watch the service statistics to see what happens. When I try to change that value to FAIL_OVER (using the Server Admin Directory page) I get this message: Load Balancing should be one of the following values : 'ROUND_ROBIN,STICKY,RANDOM'.
Our site has all its map services in one cluster of two machines. We're using the Web Adaptor and the servers are running Windows Server 2008 R2.
The Server Admin API page makes no mention of STICKY or RANDOM. Are the values different because we're using the Web Adaptor? Does anyone know what these values actually do in terms of calls to a map service? I searched around but only found information for load balancing hardware/software, nothing related to Server 10.1.
I think that what you have found is a documentation error in the admin API. IMHO, you should file this as an ESRI incident . They should clearly document the available options and their meaning without having to understand the underlying EJB mechanics.
I think that what you have found is a documentation error in the admin API. IMHO, you should file this as an ESRI incident . They should clearly document the available options and their meaning without having to understand the underlying EJB mechanics.