Performance optimizations

For load test executions with extreme high load, depending on your system capacity, creating and persisting a report for each and every node execution will impact the maximum load your Step scenario can generate. In such cases, it is possible to disable the persistence of the nodes by setting one or more of the listed predefined variables, in the case of using a set control):

• tec.execution.reportnodes.persistbefore = false: This only persist the node state after its execution; thus you won’t see the nodes in RUNNING state but will have all information needed for your analysis and reporting. Note: This is the default outcome when nodes are executed within the same threadgroup.

• tec.execution.reportnodes.persistonlynonpassed = true: By setting this option, only nodes which fails or those with technical errors gets persisted, drastically reducing the pressure but keeping the information needed to analyze error cases.

• tec.execution.reportnodes.persistafter = false: Finally, this last option combined with the “persistbefore” will disable the persistence of all report nodes which is usually not required or recommended.

The changes on the persistence of report nodes doesn’t impact the creation of the performance measurements. You will always be able to carry out the performance analysis and reporting of your execution.

Control the number of Multiworkers

In some contexts, mainly for functional testing, you may want to control the overall number of workers or threads to be used when starting a test execution, disregarding the values set in your Plans.

This functionality can be enabled by adding the following execution parameters in the admin screens settings page.

• id: “execution_threads_auto”
• label: “execution_threads”
• type: TEXT

Once done, you will see an additional field when starting a test execution.

How it works: When starting a test, by providing a value to this new field, you set a number of threads/workers that are used by the first node implementing the thread pool. The new value overrides the number of threads/workers used by the first node. This includes, among others: nodes of type Testset, For, and ThreadGroup. Each child node implementing a thread pool will see its number of threads reduced to 1.