Skip to main content

Analyze Load Test Results

Last updated on

After you run a load test, Harness displays real-time and post-execution results on the Run detail page. This page explains every section of the results view and how to interpret the data.

Accessing Results​

  1. Navigate to Resilience Testing > Load Testing
  2. Click on a test name to view its executions
  3. Select a specific run (e.g., Run #1) to open the results view

Run Overview​

The left panel shows metadata about the test and its execution:

FieldDescription
Test NameName of the load test
Load Test InfrastructureThe Linux infrastructure that executed the test, along with its connection status
TypeLoad testing framework used (e.g., Locust)
UsersPeak concurrent virtual users configured
DurationTotal configured test duration
Ramp-Up DurationTime to linearly ramp from 0 to peak users
Run StartedTimestamp when the execution began
Time ElapsedHow long the test has been running (or total runtime if complete)
StatusCurrent execution status: Running, Finished, or Failed

Summary Cards​

Four key metrics are displayed as summary cards at the top of the results panel:

MetricWhat it measures
Total RequestsTotal number of HTTP requests sent during the test
Request Per SecondAverage throughput (requests/second) across the entire test duration
Error RatePercentage of requests that returned errors (non-2xx responses or assertion failures)
Avg Response TimeMean response time across all requests, in milliseconds

Charts​

Active Users​

A time-series line chart showing the number of concurrent virtual users over time. This visualizes:

  • The ramp-up phase as users increase linearly from 0 to the configured peak
  • The steady-state phase where user count plateaus at the configured maximum
  • Any early termination if the test is stopped or fails before completing

Total Requests Per Second​

A time-series chart with two lines:

  • Request Per Second (green) — throughput over time
  • Errors/sec (red) — error rate over time

Use this chart to identify:

  • Whether throughput remains stable under sustained load
  • Spikes in error rate that may indicate the system is reaching capacity
  • Correlation between increasing users (from the Active Users chart) and throughput changes

Response Time Distribution​

A scatter plot showing individual request response times over the test duration. Each dot represents a request, color-coded by request name and outcome:

  • Success — requests that returned a successful response and passed all assertions
  • Failure — requests that failed or violated an assertion

Use this chart to spot:

  • Response time degradation as load increases
  • Outlier requests with abnormally high latency
  • Whether response times remain within acceptable thresholds

Interpreting Results​

Healthy test indicators​

  • Error Rate stays at or near 0.00%
  • Avg Response Time remains within your SLA or performance budget
  • Requests Per Second scales proportionally with active users during ramp-up
  • Response time distribution shows consistent clustering without upward drift

Warning signs​

  • Error rate climbing during ramp-up — your system may be hitting capacity before reaching the target user count
  • Response times increasing over time — potential memory leak, connection pool exhaustion, or resource saturation
  • Throughput plateauing while users increase — a bottleneck is capping request processing (CPU, database connections, rate limiting, etc.)
  • Scattered response time outliers — intermittent issues like garbage collection pauses, DNS resolution delays, or cold starts

Failed status​

A test run shows Failed status when:

  • The test infrastructure lost connectivity during execution
  • The Locust process exited with an error
  • The test was manually stopped before completion
note

A high error rate does not automatically mark the test as Failed. The status reflects execution health, not application health. A test that completes with 100% error rate still shows as a completed run — review the Error Rate metric to assess application behavior.

Next Steps​