When conducting a risk assessment for Docker daemon exposure, which configuration change MOST reduces the network attack surface of the daemon?
-
A
Enabling the Docker daemon to listen on `tcp://0.0.0.0:2375` for remote management
-
B
Configuring the daemon to listen only on a Unix socket and requiring mutual TLS for any TCP exposure
-
C
Disabling all container logging drivers to reduce daemon overhead
-
D
Setting the daemon's default ulimit to unlimited for maximum compatibility