Getting Started
This walks through one complete test: uploading a plan, running it, and reading the result. It should take a few minutes.
Sign in
Open your Atlas instance and go to /app — for a typical installation that is atlas.cloudbeat.io/app. Sign in with your email address and password.
If you have access to more than one account, you'll be asked which to work in. You can switch later from the account selector in the sidebar.
Open a project
A project holds your test plan, its data files, and the history of every run. Pick one from Projects in the sidebar, or ask your administrator to create one if the list is empty.
The project page has three tabs:
| Tab | What it holds |
|---|---|
| Results | Every run of this project, newest first |
| Files | The test plan and data files |
| Run Test | Start a new run |
Upload your test plan
Go to Files and upload your .jmx along with every data file it reads.
Check your file paths first
Test plans written in JMeter on Windows usually reference data files by a full C:\...\data.csv path, which does not exist on a load generator. Open each CSV Data Set Config and reduce the filename to ./data.csv.
If a path can't be found at run time, every thread in that thread group stops before sending a single request — the test appears to run, but that group reports nothing at all. See Test Plan Requirements.
Run it
Go to Run Test, choose where to run:
- Servers — pick individual load generators. Machines that are switched off can still be selected; they are powered on for you.
- Cluster — run on a cluster and say how many units you want.
If the project holds more than one .jmx, pick which one to run. Then press Run.
You do not need to deploy or prepare anything first. Atlas reserves the machines, copies your files, and starts JMeter.
Watch it run
The page switches to a live view showing each stage — powering on, deploying, then running — with a Stop button.
Once results start arriving, open the dashboard for the run from the Results tab. You'll see:
- Active users, response time and error rate on one chart
- Headline numbers: duration, peak users, requests per second, average response time, error rate
- A per-request table you can sort and drill into
The dashboard updates itself while the test runs. See Reading Results for what each number means.
Stop it, or let it finish
A test ends on its own when the plan says so — a thread group's duration or loop count running out. Press Stop to end it early.
When a run finishes, the load generators are released and any that Atlas powered on are switched off again.
Next steps
- Running a Test — all the options, and how data files are divided between machines
- Reading Results — the dashboard in detail
- CLI — do all of this from a terminal or a CI pipeline
- Recorder — generate a test plan by clicking through your app