Try:

./nodetool drain

The drain command stops writes from occurring on the node and flushes all data to disk. Typically, this command may be run before stopping a Cassandra node.

We will now stress the node using a simple tool called Apache Cassandra(TM) Stress. Once your node has restarted, navigate to the /home/ubuntu/node/resources/cassandra/tools/bin directory in the terminal. Run cassandra-stress to populate the cluster with 50,000 partitions using 1 client thread and without any warmup using:

./cassandra-stress write n=50000 no-warmup -rate threads=1

Initially, we will see a long list of setting for the stress run. As cassandra-stress executes, it logs several statistics to the terminal. Each line displays the statistics for the operations that occurred each second and shows number of partitions written, operations per second, latency information, and more.

Navigate back to /home/ubuntu/node/resources/cassandra/bin and run:

./nodetool flush

The flush command commits all written (memtable, discussed later) data to disk. Unlike drain, flush allows further writes to occur.

Check the new load on the node. Run:

./nodetool status

We will now examine the data cassandra-stress wrote to our node. Start ./cqlsh. Execute the following CQLSH command to view the current keyspaces:

DESCRIBE KEYSPACES;

Notice the presence of keyspace1 which cassandra-stress created.

Switch to that keyspace by executing the following:

USE keyspace1;

View the tables in keyspace1 by executing the following:

DESCRIBE TABLES;

Query the first five rows from standard1 by executing the following query:

SELECT * FROM standard1 LIMIT 5;

The data that was written is not very meaningful, since they are all arbitrary BLOB values.