Starting from the onload-201710 release, Onload deploys a user-space control plane daemon.
A single onload_cp_server process is created per network namespace in which there is an active onload_stack.
The onload_cp_server process is spawned when the
first Onload stack is created in a namespace, and
stack creation will wait until the process becomes ready - this might result in a
noticeable delay. Onload also spawns a control plane
server for the default (main) network namespace at load time, thus avoiding the delay
for the majority of use-cases.
The onload_cp_server exits after a “grace period” when the last stack in the namespace has been destroyed. A new stack, created in the same namespace, before the grace period expires, can use the existing onload_cp_server avoiding the stack creation delay. The grace period, in seconds, can be managed - see options below.
The compiler architecture of the onload_cp_server must match that of the host kernel.
See also the Management Information Base in Management Information Base.