For a virtual service to be placed on an Avi Load Balancer Service Engine, the Avi Load Balancer Controller first looks for SEs with available capacity and network reachability to make a virtual service placement decision. Reachability means the Service Engine has access to the networks required per the configuration of the virtual service and its pool members.
In the following example, a Service Engine is required to have access to the following networks:
20.1.1.x: VIP network
10.1.1.x: Pool-1 and Pool-3 network
192.168.1.x: Pool-2 network
For the first network, 20.1.1.x, the Service Engine must have an interface in this layer 2 network. It will need to be able to send gratuitous ARPs for the VIP. The Service Engine will also need an IP address of its own within the network.
For the pool/ server networks, the Service Engine is preferred to have an interface in this layer 2 network. However, if proper routes exist, the SE can access the servers through layer 3 routing. The Controller will prefer SEs with local L2 adjacency to servers versus SEs with L3 routed access.
If no SEs exist that meet the available capacity and network reachability requirements, then:
An existing SE must have its network properties updated (for reachability)
Virtual services must be scaled or migrated (to create capacity), or
A new Service Engine must be created
If the Controller has write-access mode activated, it will determine the proper action to ensure the virtual service is deployed. If Avi Load Balancer is deployed in read or no-access modes, an administrator may be required to make the necessary changes.
Issues such as network reachability are most commonly encountered when creating a new virtual service, but they might also pop up when adding a new pool to a virtual service or a new server to a pool.
