What are the implications of an in-place restart?

Hello,

I’m learning Nomad with a Raspberry Pi cluster, and would like to understand allocation restarts more deeply - specifically with a view to updating the underlying docker container that an allocation uses.

My question is this: What are the implications of an in-place restart with respect to a reallocation?

Does an in-place-restart re-pull the docker image, or does one have to reallocate in order to get Nomad to re-pull the Docker container to fetch an updated version thereof?