Dead upstream - do I really need to restart unbound when it comes up again?

[rainer at ultra-secure.de]

Hi,

because unbound does not support "classic" BIND-style RPZ feeds, I have 
to front my unbound instances with BIND resolvers - they forward every 
outbound query to them.

Yesterday, I moved one of those BIND RPZ-servers to physical hardware 
and because it needed to use the same IP, I had to shut the down the 
original host (a VM) for some time.

After it was up again, my unbound instances did not send traffic to it, 
though. That only started after I did a restart (a reload was not 
enough).

Is this normal? To be expected?

Can this behavior be influenced?