Launching a test¶
To launch a test, click on the Review & Launch button from the execution scenario page. The launch test wizard appears. It has three steps: Check Design, Check Runtime and Launch test. The first two check the scenario before it runs, so that you do not consume credits on a test that would fail.
Check Design¶
The first step runs the sanity check of each Virtual User of the scenario, including the Set Up and Tear Down Virtual Users. This step validates that none of your Virtual Users contains critical errors. Such errors, like a missing CSV variable file, would prevent the test from running. You need to fix them before continuing to the next step.
The Used By column sums up the user profiles that run each Virtual User, by role: for example User Journey, when a single user profile runs it, or Set Up · 6 User Profiles. Click its header to sort the Virtual Users by it.
The More menu of a row opens the user profiles of the Virtual User:
- Open User Profile, when a single user profile runs it,
- Show User Profiles, when several do: the Used By dialog lists them, with the role the Virtual User plays in each (Virtual User, Set Up or Tear Down). Click one to open it.
Tip
You can click on Open Virtual User to open this Virtual User and proceed to a validation in order to see the list of sanity check messages.
Check Runtime¶
The second step checks what the test needs to run. It has two blocks.
Scenario¶
The Scenario block lists the problems found by the scenario check: a missing provider, a location with no agent, a load that never ends, a server the load generators cannot reach... Each row shows:
- The Severity: Error, Warning or Info,
- The User Profile to fix, empty for a problem that concerns the whole scenario,
- The Message, as a short title.
Click a row, or its Inspect button, to open a dialog that explains the problem and lists its fixes. Previous and Next go through the other problems. The button right of Close opens the page where the problem is fixed, or a Fix it menu when there are several.
Each problem is explained step by step on the Scenario Check page.
Info
For on-premise providers, the check places the load generators on the agents the way a test does. It reports a location with no agent up, agents too small or too few for the load, and agents busy with other tests. See Agents and capacity.
Monitors¶
If you configured monitoring connections, OctoPerf automatically checks every enabled connection before launching the test. The Type column shows the kind of each connection, and the Status column the result of its check. When a connection fails, the Reason column says why. A connection that does not answer fails after 30 seconds.
Click a row, or its Refresh button, to check that connection again.
Before starting the test, you need to either:
- Fix any connectivity issue,
- Disable the invalid monitoring connections.
Tip
Use the More menu of a row to open the Monitoring Connection, or to enable or disable it.
What blocks the launch¶
- An Error of the scenario check blocks the launch, including No subscription can run this scenario,
- An enabled monitoring connection that fails blocks the launch,
- Warnings and Infos never block the launch.
Start test¶
Launch test¶
The final step allows to select:
- The report name,
- The report template (only visible if at least one template is available),
- The subscription to use, among the ones that can run the scenario.
Tip
Leave the Template field blank to use the default one. You can also switch to an other template while the test is running or completed.
Click on the Launch Test button, the test startup procedure is shown within seconds. Please allow up to a few minutes for the test to start.
Summary of the previous steps¶
When a previous step blocks, is still running or reported warnings, a banner at the top of the Launch test step sums it up. Its title gives the verdict:
| Title | Meaning |
|---|---|
| Launch blocked | A step has errors: the test cannot be launched until they are fixed. |
| Checks in progress | A check is still running: the test can be launched once it finishes. |
| Launch at your own risk | The steps reported warnings: you can launch the test, but launching it means accepting them. |
The banner has one line per step concerned, for example Check Runtime has blocking errors., with a link to open that step.
Key points to remember:
- Check Design: sanity check errors (Error level) are blocking.
- Check Runtime: scenario check errors are blocking, and all enabled monitors must be functional.
Info
A scenario with an Error never starts, wherever it is launched from: the REST API, the Maven plugin, the CI integrations, JIRA, the scheduler and the MCP server refuse it as well. See Scenario Check.
Test starting¶
Once all your load generators are green, the tests will start shortly. You can see more details through the Show logs button on top.
No subscription found¶
When none of your subscriptions can run the scenario, the Launch test step shows no Subscription block, and its banner says that Check Runtime has blocking errors. The Scenario block of the Check Runtime step reports No subscription can run this scenario, with what stops each subscription: too many Virtual Users, a test too long, too many user profiles...
You can either lower the setting that is too high (load, duration or number of user profiles), or get a subscription that fits, from Subscriptions in the user menu.