First recommendation is to provide HA/resiliency to RADIUS. Not sure this is something you'd need to enable by default. Try to use it as fail-safe. If you still need it, see below:
You would need to enable dot1x critical.
Add this to your port:
dot1x critical
dot1x critical vlan
Optionally, if you want to initialize the port once the switch discovers RADIUS is back, add this to the port:
dot1x critical recovery action reinitialize
Also, add a test username at the end of your RADIUS server definition (this allows the switch to actively seek the server while it's in a down state):
radius-server host 10.123.100.6 test username
Also add this for deterministic deployment:
radius-server dead-criteria time 15 tries 3
Hope this helps,