Details
-
Bug
-
Status: Resolved
-
Low
-
Resolution: Fixed
-
None
-
Low
Description
There were a couple of minor issues with the new stress:
1) The warmup period did not scale up as the cluster size increased
2) The mixed workload did not work with CQL
At the same time, I have introduced a change in behaviour in the way the default column values are generated so that they are deterministically based on the key. I have then modified read operations to verify that the data they fetch is the same as should have been inserted, so that stress does some degree of data quality checking at the same time. For the moment the values generated never vary for a given key, so this does nothing to test consistency, it only tests for corruption.