What’s new

JUnit report reworked — a campaign is now reported like a test suite: one test case per step, skipped when the step was not executed, failed or errored according to the severity of its messages, with the step status, executions, messages and measurements attached as standard output. The whole campaign is written to a single file.

Campaign configuration validated before submission — the configuration rules, such as the maximum number of zones, are now shown and enforced directly in the form instead of being discovered through a rejected submission. Zone configuration is explicitly optional.

Clearer zone configuration — zones are sorted by label and fully described, with their flag and description, and default to a 100% share.

Campaigns no longer hang — a campaign with no remaining factory to acknowledge its shutdown terminates immediately, and a scenario that was never closed takes the final status of its campaign instead of being reported as running forever.

New extension point — an additional campaign hook is triggered when the warm-up of the scenarios starts, the point from which a campaign effectively consumes resources.

TimescaleDB plugin - The values of the timer meters were read as microseconds instead of the correct milliseconds.

Generating the report of a campaign configured with a zone failed with a template error; zone resolution was also made significantly cheaper.

Requesting the abort of a campaign whose factory could not be created or could not start left the campaign stuck in a non-terminal state. The abort is now honoured whatever the state of the factories, and the delay before the cancellation is propagated is configurable.

The completion hook was not properly called when a campaign failed at start.

Clearer labels on the deletion buttons of the reports and series pages.

Refinements to the scenario report display, and web application dependencies upgraded.

Internal meters are now prefixed with an underscore.

HTML execution reports — a campaign can now produce a self-contained, user-friendly HTML report, generated automatically at the end of a run or downloaded from the campaign details page. It details the campaign summary and, for each scenario, its own summary, the zones it ran on with the configured load distribution, and every visible step in execution order with its measurements and messages.

Textual report for standalone runs — the same results are also available as a readable console/text report.

Reports without an external time-series database — campaign measurements are now collected and stored by the Head, so standalone runs produce complete reports out of the box.

Accurate meters — meters are no longer aggregated when read for charts and reports: step measurements keep their original granularity, and the campaign-level values are returned alongside for comparison. As a consequence, a data-series built on meters no longer takes an aggregation operation.

Scenario and step details in the web UI — the campaign details page now shows the same breakdown as the report, once the campaign has completed.

Slack and Email plugins - Graphical improvements to notifications and email notifications can now transport the textual, the HTML and the JUnit reports when enabled.

The timer step compared values using the wrong time unit.

The minion identifier was attached as a tag to the timeout failure meter, inflating its cardinality.

The campaign, data-series and report details screens could not be refreshed.

Steps that were not executed were not displayed as disabled.

Various report generation failures.