The same benchmark engine ships as a standalone configmax binary. It runs the identical test code
directly against a kubeconfig, with no operator installation required. Use it for quick smoke runs
and early iterations on a development cluster. Prefer the
operator for any long-running test, because the CLI runs from a
workstation and a laptop sleeping or a network interruption will cancel the run.
# 1. Point the config at your cluster: copy benchmarks/configs/default.yaml
# and set cluster.kubeconfigPath, then enable the test or tests you want, for example:
#
# cluster:
# kubeconfigPath: "/path/to/kubeconfig"
# networking:
# subnetsPerVPC:
# enabled: true
# mode: discovery
# discovery:
# initialBatch: 5
# growthFactor: 2.0
# hardTimeoutSec: 300
# safetyBound: 500
# 2. Inspect the cluster without creating anything
configmax discover --config my-config.yaml
# 3. Run the enabled benchmarks
configmax run --config my-config.yaml
# 4. Re-read a saved result or render a report
configmax results --file benchmarks/results/<result>.json
configmax report --file benchmarks/results/<result>.json --format html
Results print as a summary table and persist as JSON under benchmarks/results/, including the
per-batch curve and the stop reason, which is the same data the operator reports in
.status.results. The config file accepts the same mode, discovery, and distressOverrides
fields as the CRD overrides shown on the operator page.