DevTest tracers generate metrics at one or more of the four levels:
Metrics that are created at the simulator and test step levels are also aggregated to a user-configurable maximum level (test case, by default). If a test case executes on two simulators and it contains two test steps, the "Responses Per Interval" metric is generated for each test step under each simulator. Then (depending on the configuration) the values are aggregated to generate the "Responses Per Interval" metrics for each simulator. Finally (depending on the configuration) the values are aggregated again to generate the "Responses Per Interval" metric for the test case. The highest level to which any metric can be rolled up is the test case level. Thus, the only metrics that are generated at the DevTest level are listed in the following table with a level of DevTest.
Data from all events of relevant type in each agent reporting interval of 15 seconds are combined as shown.
Metric |
Level |
Source |
Average Response Time (ms) |
Test Step |
Average of "Long Description" value (converted to integer) from all "Step response time" events (id 18) |
Responses Per Interval |
Test Step |
Count of all "Step response time" events (id 18) |
Errors Per Interval |
Test Step |
Count of all "Step error" events (id 20). |
Failures Per Interval |
Test Step |
Count of all "Abort" (id 50) and "Cycle failed" (id 13) events. |
Tests Running |
Test Case |
Count of test cases running at the end of agent reporting interval. Count is incremented for each "Test started" event (id 4) and decremented for each "Test ended" event (id 5). |
Virtual Users Running |
Simulator |
Count of virtual users running at the end of agent reporting interval. Count is incremented for each "Cycle started" event (id 11) and decremented for each "Cycle ending" event (id 24). |
Test Runner Errors Per Interval |
DevTest |
Count of all "Cycle runtime error" events (id 25) |
Staging Errors Per Interval |
DevTest |
Count of all "Model definition error" events (id 23) |
The Errors Per Interval metrics for any test step, simulator, or test case are reported only when the first error event is detected. The same is true for Failures Per Interval.
If only a few, short test cases run, the Tests Running and Virtual Users Running metrics can continue to report 0 for each agent reporting cycle. This happens when both start and stop events occur in the same 15-second period, and the count at the end of the period is 0.
When the metrics are being aggregated, the lower-level metric can be created in one agent time reporting period. The aggregated metrics can be created in the following period. Occasionally, there are points where lower level and rolled up metrics within a time period do not agree in value. This situation is unavoidable.
Copyright © 2014 CA Technologies.
All rights reserved.
|
|