Stress Test for Cassandra

5
votes

i'm using the stress test provided within cassandra in DataStax Enterprise. I'd like to have some information about it and cassandra too, if someone knows it. - First of all, which nodes the stress test use? I mean, when i use the comand (on a ring composed by 2 nodes):

./cassandra-stress -d ip1 --replication-factor 1

this means the test only use the first nodes excluding completely the second node?

  • Secondly, when i save a keyspace and create some data into it (for example column family 'cf1' and kespace 'ks1'), the folder /var/lib/cassandra/data/ks1/cf1 is empty...it is normal? When i connect to cassandra-cli, it seems to read from /.cassandra/assumpiont file.

  • Third, is it possible to set a single entry point from cassandra? I mean to force only one node to manage the calculus of the hash mapping to all the other nodes.

1
cassandracassandra-stress
Is this c* 2.1? Stress changed significantly between 2.0 and 2.1 – phact

1 Answers

7
votes

which nodes the stress test use?

That's up to you. You can supply the --nodes parameter (a comma separated list without spaces, eg: 127.0.0.1,127.0.0.2,127.0.0.6 etc.) More documentation.

the folder /var/lib/cassandra/data/ks1/cf1 is empty...it is normal?

Yes. It will be empty until cassandra flushes it's data to disc. If you want to see data being populated in the folder you can trigger a manual flush via nodetool (./nodetool flush)

Third, is it possible to set a single entry point from cassandra?

No. This would create a SPOF. We want to avoid those.