There may certainly be valid reasons for the commitlog_directory option not being homogenous across all nodes. The message you see in the logs is overly ominous, and in valid cases can be ignored.
Typically that occurs when clicking "Configure Cluster" from the OpsCenter UI, which prefills all cassandra.yaml options based on the values from an existing nodes. The message displayed means that because nodes do not have identical config values, OpsCenter can't pick a node to use as a template, so it puts that choice in the user's hands.