Run tests on Antithesis

Antithesis runs software in a deterministic simulation and fuzzes it. Hegel tests can be used to write Antithesis workloads.

When running inside Antithesis, Hegel automatically uses the workload settings profile. The workload profile:

  • uses the urandom Backend, which takes every choice from the Antithesis fuzzer rather than from Hegel’s own random number generator

  • turns off the example database

  • suppresses every health check

Each test’s outcome is reported as an Antithesis assertion, so failures appear in the Antithesis triage report.

A stateful invariant declared with pause_faults=True (see Stateful testing) runs with fault injection paused: before each check Hegel runs the program named by ANTITHESIS_FAULT_CONTROL with --pause, and after it with --unpause, so no fault lands in the middle of the check.

Write a workload program

hegel.main() turns a function that takes a test case into a program. Each run of the program runs one test case.

import hegel
from hegel import generators as gs


@hegel.main
def workload(tc):
    n = tc.draw(gs.integers())
    assert n + 0 == n


if __name__ == "__main__":
    workload()

Each setting is a command-line option named after its hegel.toml key. To run the program with the workload profile outside Antithesis:

python workload.py --profile=workload

If the test case fails, the program exits with status 1 and prints a rerun with: --failure-blob BLOB line. Run the program with that option to replay the failure.