Dropped Messages Indicate Overload Before Timeout
criticalcassandra_dropped_messages represents Cassandra's load-shedding mechanism when requests can't be processed within timeout. This is a leading indicator of cluster overload — messages drop before clients see timeouts, revealing capacity problems earlier than error rates.
Alert immediately when cassandra_dropped_messages > 0, especially if persistent. The specific message type (read, write, mutation, etc.) indicates which operation type is overwhelmed. Cross-reference with cassandra_client_request_error rates to understand client-visible impact.
Dropped messages during known node outages are expected, but persistent drops during full cluster availability signal undersized cluster or slow nodes. Increase capacity by adding nodes, or reduce incoming traffic. Investigate slow queries or hot partitions causing timeout exhaustion. Review timeout configurations to ensure reasonable SLAs.