We are implementing a new architecture for one of our portals in order to set up a cluster in the backend (two servers) with a load
balancer in front of it (we are balancing an LDAP service). The thing is that, according to the configuration guide for the backend
installation, we must configure the load balancer to return immediately to the calling client when the service it balances is
unavailable. The aim is to tell the balancer that if none of the balanced backend servers are available, it must return the error to
the client immediately, instead of waiting the client timeout settings to do the job.
However, our load balancer administrator told us that he doesn't know what setting is needed in the load balancer to establish that
behavior. At this time, if none of the balanced backend machines are available the client waits indefinitely for the connection to get
a timeout. Does someone know how to achieve this immediate error response?