Resources limits not being adhered

[Rapol]

Reflector is crashing when reaching 100mib of memory even though the limit is higher (300mib).

Version: 7.0.151
Pods: 1/1
resources:
  limits:
    cpu: "150m"
    memory: "300Mi"

This is deployed using reckoner which uses helm internally. This might be more of a user error on the k8s side but would like to know if anyone can point of to the right direction.