Question-41: Why defining the seed node in the Cassandra cluster is critical?

Answer: Whenever a new node is launched in the cluster, it has to know which node should be contacted to get the information about other nodes in the cluster. Seed nodes are the nodes which helps the new node to get the latest information about the cluster. It is a critical when only node first time starts up.

Question-42: Is it true that a node always contacts the seed nodes to get the latest information about the cluster?

Answer: No, only when node comes up first time then only it contacts the seed node to get the information about the cluster. Once node is up and already participating in the cluster then it remembers other nodes it has gossiped with, for subsequent restart.

Question-43: Is seed node single point of failure in Cassandra cluster?

Answer: No, deciding seed node has no purpose other than bootstrapping the gossip process for new nodes and they are not a single point of failure and do not have any other purpose in the cluster operation other than helping the bootstrapping of new nodes. Even, you don't define a single node as a seed node, but rather in a big cluster there are more than one node defined as seed nodes by default at least 3 nodes. 

Question-44: If cluster is having less than 10 nodes then it is always advisable that you define every node as a seed node?

Answer: Making every node as a seed node is not at all recommended. Because it affects the gossip data transfer as well as overall maintenance increases. It is always recommended that you keep 3 nodes as a seed node per datacenter.

Question-45: Suppose a particular node is having network outage for 5 minutes then this node would be dropped from the cluster?

Answer: No, In Cassandra cluster a node can never be dropped automatically. If you want to remove a node from the cluster then you have to explicitly remove the nodes from the cluster.