How to make sure that data is replicated n no of times in Cassandra

Question

If say RF = 3, in a 5 nodes single data center, then how can I see where exactly my data lands in which node for any keyspace.tablename, and how to prove that data is exactly replicated on 3 nodes.

I have tried various nodetool commands but none give me the exact result or what I am trying to see.

https://stackoverflow.com/questions/31669991/how-does-cassandra-find-the-node-that-contains-the-data — undefined_variable, Jun 28 '18 at 13:18
This link does not explain or confirm the RF and on which nodes the data is exactly saved. — Piyush_Rana, Jun 28 '18 at 14:04
bring the node down on which it says data is there... and then query for same data... you will still get data — undefined_variable, Jun 29 '18 at 05:42

score 3 · Answer 1 · answered Jul 09 '18 at 13:14

You can prove this using Inserts and queries with various Consistency Levels (CL) from the client side (cqlsh for instance). I'll explain:

Create the Keyspace with RF=3 -> Create the table
INSERT data with CL=ALL (this means all 3 replicas must ACK that the write was successful)
While all nodes are up query (SELECT) for the data using CL=ALL (all 3 replicas must send response for the query to be considered successful)
Stop C* on 1 of the nodes
While 1 node is down (DN state in nodetool status) query (SELECT) for the data using CL=ALL (all 3 replicas must send response for the query to be considered successful).

a. If the query succeeds, than all 3 replicas are still up and you need to stop another C* node.

b. If the query fails, set the CL to CL=QUORUM (only 2 out of 3 replicas must send response for the query to be considered successful), and execute the query again.

you can continue this sequence with CL=ONE and even CL=ANY.

You can read more about cqlsh and how to modify the CL here

How to make sure that data is replicated n no of times in Cassandra

1 Answers1