Skip to main content

FAQ

What is the license?​

Apache 2.0. The source is at github.com/vegaload/vegaload.

Do I need an AI agent to use VegaLoad?​

No. VegaLoad is one binary with a normal command line. The agent path is optional. See Your first test.

Which systems does it run on?​

Linux, macOS and Windows. See Install for the archives, Homebrew and Docker.

Which languages can scenarios use?​

JavaScript, TypeScript and Python. Python scenarios need Python 3 on the machine. Most protocol tests need no script at all: you give -target and -protocol. See Scenarios.

What leaves my machine?​

By default, nothing. Reports and the audit log (~/.vegaload/audit.log) are written locally. Two things can send data out:

  • VEGALOAD_LLM_PROVIDER is optional. If you set it, diagnose sends the report data to the provider you chose.
  • The GitHub Action downloads a VegaLoad release.

The load itself goes to the target you name. See Safety.

Can an agent run a test against any host?​

Not without a check. A run against a host that is not localhost or allow-listed is refused unless it is allowed with -allow-target or -yes. Read Safety for what an agent can and cannot change.

Can I run it in Docker or Kubernetes?​

Docker, yes. A small image is published at ghcr.io/vegaload/vegaload. The image holds only the binary, so Python scenarios need a different base image. See Install. For Kubernetes, the repository has a Helm chart that runs a test as a Job.

How does it compare with k6, JMeter, Locust and Gatling?​

See Compare. We say where each tool is the better choice.

How do I scale a test across many machines?​

See Scale with Harness.

A run failed and I do not know why​

Start with Troubleshooting. Then try vegaload diagnose, or see Explain a run.