Analyse test results and build reports
Mode: the Analyse section is shown in Expert mode (sidebar footer toggle Mode: Basic / Expert).
Overview
Analyse is where a finished test run stops being raw numbers and becomes an answer: did this build get slower, which transaction regressed, and what do I send to the people who asked. It is the platform's observability layer over data you own, so everything here reads from the datasources you configured, not from a store the platform hides.
Five pages, one path:
| Page | Use it to |
|---|---|
| Analyse: Overview | Reach the four sub-features: Dashboards, Reports, Analysis, Datasources |
| Analyse: Custom chart | Overlay one run's transaction timings and performance counters on one time axis |
| Analyse: Analysis | Compare runs against a baseline and pin the result to a dashboard |
| Analyse: Saved charts | Reopen and re-render a chart you kept |
| Analyse: Reports | Produce a PDF or XLSX a stakeholder can read without a login |
Before you begin
- At least one datasource is configured and returning data. Analyse reads InfluxDB, SQL and JSON connections. See Configuration: Datasources.
- At least one load test has completed and appears in Testing: Test Runs. Every page in this section needs a run to point at.
- To plot performance counters, PerfCounter collection was enabled for the run's agents before the test ran. It cannot be added afterwards. See Configuration: Systems Under Test: PerfCounter Settings.
- Your role grants access to Analyse. If the Analyse group is missing from the left rail, that is a permissions question, not a configuration one. See Operations: Permissions.
Steps
Step 1
Start on Analyse: Overview. The four navigation cards - Dashboards, Reports, Analysis, Datasources - are the whole section. If a card leads to an empty page, the cause is almost always upstream: no datasource, or no completed run.
Step 2
Look at a single run first on Analyse: Custom chart. Choose the datasource, then the run - only runs belonging to that datasource are offered - then put a measurement timing or a performance counter on the Y-axis. All transactions aggregates the whole workload; a named timing narrows the chart to one transaction. The canvas stays empty until both Datasource and Run are set, which is the usual reason a first attempt looks broken.
Overlaying a transaction timing with a system counter from the same run on one time axis is the fastest way to tell a slow application from a starved machine.
Step 3
Compare builds on Analyse: Analysis. In the OPERATION panel choose the operation, pick the metric family as Category > Subcategory > Measurement, then set one baseline test key and one or more compare test keys. Every regression and improvement in the preview is calculated relative to that baseline, so choose a run you trust as the reference rather than the most recent one. The PREVIEW panel renders the chart inline before you commit to anything.
Step 4
Keep what is worth keeping. Use Save to folder in the builder, and the chart appears on Analyse: Saved charts grouped by folder. A saved chart stores the definition, not a picture: reopening it re-runs the analysis against current data. Each chart belongs to exactly one folder and can be moved later.
Step 5
Assemble a view on Analyse: Dashboards. Create a blank dashboard and add saved charts to it, use New from run to auto-generate one pre-populated from a specific test run, or start from a template on the Templates tab. Dashboards persist between sessions and can be shared; dashboards other people shared with you appear on the Shared with me tab.
Step 6
Produce the deliverable on Analyse: Reports. The page has three tabs. Create the layout on Templates first - a report cannot be generated without one. Then on Generate, pick the template, category, datasource and one or more test runs, optionally adding a baseline run for a side-by-side comparison, and set a format override only if you want to force PDF or XLSX against what the template defines. Finished files are listed on History and can be downloaded again at any time.
Verify the result
A chart that re-renders is the real test. Open a saved chart and confirm it redraws against current data rather than showing a stored image. On a dashboard, confirm the panels populate for a run you know has data.
For a report, generate once and open the file from History. If the output is missing a section, the template defines it, not the Generate tab: fix the template and regenerate.
Troubleshooting
- Baseline or Compare test keys lists are empty. No completed run exists for that datasource yet. Confirm the run finished on Testing: Test Runs.
- The custom chart canvas stays blank. Both Datasource and Run must be selected before anything renders.
- Performance counters are not offered for a run. PerfCounter collection was not enabled for that run's agents before the test ran. The data does not exist and cannot be back-filled; enable it and run again.
- Saved charts shows an empty-state panel. Nothing has been saved with Save to folder yet. Charts are saved from the Analysis builder or Custom chart, not from this page.
- The Generate tab will not produce a report. At least one template must exist on the Templates tab first.
- The Analyse group is missing from the sidebar. This is a role permission, not a datasource problem.