Question-: You are having node Cassandra cluster spanning two data centers. One of the seed nodes from this cluster is down and you have to replace that node. What all you would be doing? A. You would update the Cassandra.yaml file for each node and remove the IP of dead node as seed node. B. You would update the Cassandra.yaml file for each node and add the IP of new node as seed node. C. You would be performing rolling restart on all nodes so that nodes are aware of the changes in the seed list D. You would be updating the jvm.options file of new node and add the IP address of the dead node in replace_address property.
Answer: A,C,D
Explanation: You should not add new node as a seed node, while replacing the node. All other options are correct.
Question-: What all are the functionality of the seed nodes? A. They would be contacted while bootstrapping to get the gossip info. B. Any time any node can contact to seed node to get the gossip info from the seed node. C. Seed node are always used when data read from the cluster. D. They are also known as coordinator node
Answer: A,B
Explanation: Seed nodes are useful for getting the gossip (cluster info) when any node needs it. Even while bootstrapping they can be contacted to get the gossip info. Any node in the Cassandra cluster can be a Coordinator node, when you read the data from the Cassandra cluster and you contact one of the nodes in the cluster. Whether this node has the data or not. It would sever the read request by fetching the data from the node which is having it and it become the coordinator node. Seed nodes are not master nodes or coordinator nodes. There is no concept of master node in the world of Cassandra.
Question-: You want to replace the currently running node in the cluster for applying software patch on particular node, which of the following is correct for that? A. You would first add a new node and then remove the old node on which the patch should be applied. B. You would be replacing node by using replace_address property in the jvm.option file. C. You must make sure that the consistency level One is used on the old node D. You must make sure that the consistency level One is not used on the old node
Answer: A,D Replacing the running node may be required for applying hardware upgrade or for applying the software patch etc. There are mainly two approaches for this as below. A. Adding a new node to the cluster and then decommission the old node : Once this new node is up and running. You do the following steps a. Note the Host ID of the original node. b. Using that Host ID of the original node, decommission the original node from the cluster using the nodetool decommission command. c. Now run the “nodetool cleanup� command on all the other nodes in the same datacenter (no need to run this cleanup command on other datacenter) B. We can replace the currently running node and avoid the stream the data twice . a. We have to make sure that data written using the consistency level ONE, you risk losing data because the node might contain the only copy of a record. Hence, we have to make sure no application uses the consistency level ONE. b. Stop the node, which needs to be replaced. c. Add the replace_address property on the new node with the old node IP address.