Active-standby for GTM level

We have 2 sites each with a BIG-IP VE F5 running 11.5.3. Both sites are running several services and each service has a Wide IP assigned to it. We are in need of a configuration that will support an active-standby environment. Below is the current configuration:

 

SITE 1 ----- SITE 2

 

1 Wide IP

 

1 Pool

 

Site 1 Virtual server 1.1.1.50

 

Site 2 Virtual server 2.2.2.50

 

Site 1 LTM pool ----- Site 2 LTM pool

 

1-to-many nodes per LTM pool

 

Site 1 Nodes 1.1.1.3

 

Site 2 Nodes 2.2.2.3

 

Use case for active-standby: All client request traffic must go to 1 site/node (SITE 1 - 1.1.1.3) until that node becomes unavailable. In which case all traffic should go to the next available node on the same site. If there aren’t any more on the same site a node on the other site (SITE 2 - 2.2.2.3) should be chosen. Once the first node comes back available, traffic should not failback to the first node (SITE 1 - 1.1.1.3). Traffic should stay on (SITE 2 - 2.2.2.3) until it becomes unavailable.

 

My question is, is it possible using the current configuration to accomplish the active-standby use case above? If so, what will it take to accomplish this? Our problem moSTLY lies with the way that GTM/DNS selects which site to use. Using manual resume would be a good solution for that except we can’t have user interaction. Manual resume might work if there is a way to automate re-enabling that SITE 1 when SITE 2 eventually goes down.