Skip to main content

Argus stress tests

You’ll load a verified HTTP target from one or more Argus Sensors at once.

Use stress tests to measure how an API behaves under concurrent virtual users. Uptime checks stay separate — they are not replaced by stress jobs. Core orchestrates; only Sensors send traffic.

Before you start​

  • One or more online Sensors
  • Permission to stress your own API (do not aim at third-party hosts you do not control)
  • A clear success signal (expected status codes, acceptable latency)

Concepts​

TermMeaning
Verified targetHost you allow for high-volume tests (required)
FlowOrdered steps (HTTP and Wait). One iteration runs the whole flow once
Concurrent usersHow many virtual users run in parallel across selected Sensors
DistributionHow load is split across Sensors (equal, weighted, or custom)

Estimated HTTP requests ≈ Total flow iterations × number of HTTP steps in the flow.

Add a verified target​

  1. Open Argus → Stress Tests → Verified Targets.
  2. Click Add target and enter the Host.
  3. Save. High-volume runs may only hit verified hosts for your organization.

Create a flow​

  1. Open Flows → Create flow.
  2. Name the flow and add steps with Add HTTP step and Add wait.
  3. For each HTTP step set method, URL (on a verified host), optional headers/body, and expected status codes.
  4. Save.

Create and start a run​

  1. Open Runs → Create & start run (or create, then Validate / Start).
  2. Pick the Flow, Sensors, Concurrent users, and Total flow iterations.
  3. Choose Distribution (use Preview distribution if you need to check the split).
  4. Validate, then Start.
  5. Watch live progress; Cancel if needed.
  6. When finished, open Run insights for success rate, latency, status codes, per-step breakdown, and per-sensor results.

You’re done when​

  • The run completes (or you intentionally cancel).
  • Insights show overall latency and error rate you can compare to a quieter baseline.

Mistakes to avoid​

  • Do not point stress tests at unverified arbitrary hosts.
  • Do not confuse Total flow iterations with total HTTP requests.
  • Do not expect Core to send traffic — only Sensors do.
  • Do not use production load you have not sized the target for.

If it fails​

SymptomFix
Validation fails on hostAdd the host under Verified Targets
Run stuck / no samplesConfirm Sensors are online and can reach the target
High error rate immediatelyLower concurrent users; check expected status codes and auth headers
Empty insightsWait for Sensors to report, or re-check that the run finished