Live results#
The Live results panel appears on the Run page once a test starts. It has two halves, kept apart on purpose.
Client-side metrics#
What the load generator saw, from outside the application. Read from the results file as JMeter writes it, so these are the same figures the JTL Analyzer will produce from that file afterwards — not a second opinion.
| Figure | What it is |
|---|---|
| Samples | Requests completed so far |
| Configured / Active / Peak users | Threads asked for, running now, and the most so far |
| Completed users | Threads that have finished their work |
| Failed | Samples that failed, by JMeter's own judgement |
| Error rate | Failed as a percentage of samples |
| Throughput | Samples per second |
| Average / Median | Response time, in milliseconds |
| 90th / 95th / 99th | Percentile response times |
| Slowest | The worst single sample |
These refresh on their own while the test runs. The refresh interval is yours to choose, and can be switched off.
Server-side metrics#
What the application said about itself, over the same window, read from your APM provider. This half only appears if monitoring was switched on before the run started; see APM integrations.
Waiting for data#
For the first minute or two of a run, the server-side area shows Waiting for APM data rather than empty space or a flat line.
That is not a fault. Providers publish metrics in buckets and ingest them with a delay — Dynatrace, for example, stores one-minute resolution and ingests with a short lag — so there is genuinely nothing to draw for a window that only just began. Each graph appears as its metric starts reporting.
Note A gap stays a gap. JMXPress never draws a zero for a period a provider did not report, because a flat line along the bottom is a measurement — it says the application served nothing — and that is not what an unanswered window means.
Why the two halves will not match#
They are measuring different things and are not meant to agree:
- Client-side response time includes the network, any proxy, and whatever queues in front of the application.
- Server-side response time is what the application itself recorded.
The difference between them is useful. A client-side figure far above the server-side one points at the path to the application rather than at the application.
Rate limits#
JMXPress respects each provider's own floor on how often it may be asked. If a refresh comes round sooner than that, it says so — "Dynatrace is not worth asking again for another 46 seconds" — and keeps the readings already on screen rather than spending an API call on the same answer.
After the run#
The client-side figures are in the results file, which the reports are generated from. The server-side figures collected for the run's window are stored with it and can be opened from the run's APM report.