Library and pytest

k9.run() runs a script, a plan file or a k9.plan.Plan (in a child process) and returns a RunResult:

import k9

result = k9.run("load/smoke.py", vus=20, duration="30s", env={"BASE": base_url})
print(result.iterations, result.metric("http_req_duration")["p(95)"])

Trends in a RunResult always have count, avg, min, med, max, p(90), p(95), p(99) and p(99.9), plus any summary_trend_stats asks for.

Crossed thresholds raise k9.ThresholdsFailed (an AssertionError); pass check_thresholds=False to get the result anyway. A test that can't start raises k9.ConfigError or k9.ScriptError. All three are k9.RunErrors, with the exit code, stderr and, if the run finished, its result. summary_trend_stats=["p(99)"] adds percentiles beyond the default p(90) and p(95).

Installing k9-load also installs a pytest plugin. The marker runs the load test, and crossed thresholds fail the test:

import pytest

@pytest.mark.k9("load/smoke.py", vus=10, duration="30s")  # relative to this file
def test_checkout_holds_up(k9_result):
    assert k9_result.metric("http_req_duration")["p(95)"] < 300

def test_two_runs(k9_run):
    assert k9_run("load/a.yaml").ok and k9_run("load/b.yaml").ok