[CASSANDRA-13772] Add a skip read validation flag to cassandra-stress

• Type: Improvement
• Status: Resolved
• Priority: Low
• Resolution: Fixed
• Fix Version/s: 3.11.1, 4.0-alpha1, 4.0
• Component/s: Tool/stress
• Assignee: Noam Hasson
• Reporter: Noam Hasson

Authors

Noam Hasson

Reviewers

Jason Brown

• Created: 17/Aug/17 08:49
• Updated: 15/May/20 08:03
• Resolved: 24/Aug/17 21:43

Description

When running cassandra-stress with read operations, you must make sure all the data was populated beforehand or else you will get the following errors:
java.io.IOException: Operation x0 on key(s) [4d31314e32314b395030]: Data returned was not validated

at org.apache.cassandra.stress.Operation.error(Operation.java:127)
at org.apache.cassandra.stress.Operation.timeWithRetry(Operation.java:105)
at org.apache.cassandra.stress.operations.predefined.CqlOperation.run(CqlOperation.java:91)
at org.apache.cassandra.stress.operations.predefined.CqlOperation.run(CqlOperation.java:99)
at org.apache.cassandra.stress.operations.predefined.CqlOperation.run(CqlOperation.java:245)
at org.apache.cassandra.stress.StressAction$Consumer.run(StressAction.java:453)
java.lang.RuntimeException: Failed to execute warmup
at org.apache.cassandra.stress.StressAction.warmup(StressAction.java:117)
at org.apache.cassandra.stress.StressAction.run(StressAction.java:62)
at org.apache.cassandra.stress.Stress.run(Stress.java:143)
at org.apache.cassandra.stress.Stress.main(Stress.java:62)

Even if you use the "-errors ignore" flag you'll get a lot of the following messages which will both slow the stress and prevent it from displaying the metrics:
Operation x0 on key(s) [4b3539393831374e3431]: Data returned was not validated
Operation x0 on key(s) [4f4b3936363233375030]: Data returned was not validated
Operation x0 on key(s) [4d304c4b32384f4b3031]: Data returned was not validated

What I propose is to add a flag to skip the read validation, such as:
-errors skip-read-validations

This way when needed you can run a mixed workload and ignore validation of unpopulated data.