[bmscomp]

Why

scripts/ still holds shell versions of jobs the kates CLI now does. Two copies drift apart. The perf scripts start raw kafka-*-perf-test Jobs that Kates never records. The chaos scripts write their own ChaosEngines instead of going through the resilience API.

What changes

• Removed: 12 scripts. Each one maps to a single kates command (table below).
• Makefile:
  • make test now runs kates test create --type LOAD --records 1000000 --wait.
  • make test-load|stress|spike|endurance|volume|capacity run kates test create --type <TYPE> --wait.
  • make kafka-detect runs kates detect.
  • The target names and help text stay the same, so the book's Makefile tables still hold.
• Docs:
  • scripts/CHAOS_TESTS.md: the single-fault runs now point at kates resilience run -f with examples 02, 03, 05 and 07 of kates/docs/tutorials/resilience.json. The Game Day section and the observability notes stay.
  • resilience.json _usage.script_alternatives: only full_gameday is left.
  • setup-kafka-chaos.sh's closing hint and dashboards/kafka-performance/README.md name the CLI.
• Not edited: docs/kafka-multi-version-deploy-plan.md and docs/mirror-maker2-cli-migration-plan.md still name test-perf-*.sh. They are records of past plans, so I left them as written.

One behaviour change: the make test* targets now need the Kates API deployed and reachable (kates ports). The scripts only needed the Kafka cluster.

Every script in scripts/

Script	Verdict	Reason
test-kafka-performance.sh	removed	kates test create --type LOAD --records 1000000 --wait
test-perf-load.sh	removed	kates test create --type LOAD --wait (--producers, --consumers, --throughput, --record-size cover its env vars)
test-perf-stress.sh	removed	kates test create --type STRESS --wait
test-perf-spike.sh	removed	kates test create --type SPIKE --wait
test-perf-endurance.sh	removed	kates test create --type ENDURANCE --wait
test-perf-volume.sh	removed	kates test create --type VOLUME --wait
test-perf-capacity.sh	removed	kates test create --type CAPACITY --wait
test-chaos-broker-delete.sh	removed	kates resilience run -f, resilience example 02 (POD_DELETE)
test-chaos-broker-cpu-hog.sh	removed	kates resilience run -f, example 03 (CPU_STRESS)
test-chaos-disk-io-stress.sh	removed	kates resilience run -f, example 05 (IO_STRESS)
test-chaos-network-partition.sh	removed	kates resilience run -f, example 07 (NETWORK_PARTITION)
kafka-cluster-report.sh	removed	kates detect: same 3-AZ compatibility report, plus JSON/PDF output and generated values. Exit codes differ: the script exits 1 on incompatible, kates detect --fail-on-error exits 2
start-cluster.sh	kept, partial	kates deploy creates a kind cluster only during a deploy and only when no cluster is reachable. The CLI has no standalone create, no --clear-proxy / --clear-proxy-ca, no proxy reconcile on an existing cluster, and does not read proxy/proxy.conf
destroy.sh	kept, partial	kates clean uninstalls releases and namespaces. It does not delete the kind cluster, the kind-registry container and volume, or the kind network
port-forward.sh	kept, partial	kates ports does the same job on other local ports (API 8080, Grafana 3000, Kafka UI 8085, Apicurio 8081). The script uses 30083/30080/30081/30082, which the book and tutorial 13 document. kates ports has no Litmus UI or Headlamp forward
deploy-kates.sh	kept, partial	kates deploy installs Kates only with the whole stack, in its own namespaces. It has no single-component install, which make kates-deploy needs
deploy-kafka.sh	kept, partial	same: no Kafka-only install in the CLI (the book and charts/kafka-cluster/INSTALL.md use it)
deploy-cert-manager.sh	kept, partial	kates deploy installs cert-manager only with the stack, into cert-manager (the script uses kafka). Needed by make cert-manager
deploy-kafka-ui.sh	kept, partial	stack-only in the CLI. Needed by make ui and make deploy-all
deploy-monitoring.sh	kept, partial	stack-only in the CLI, which installs into monitoring (the script uses kafka)
deploy-jaeger.sh	kept, partial	stack-only in the CLI (inside --with-monitoring). Needed by make jaeger and make deploy-all
ensure-kafka-user.sh	kept, partial	kates deploy creates the kates-backend KafkaUser only during a full deploy. make calls the script alone
gameday.sh	kept, partial	kates resilience run covers baseline + one fault + recovery. It does not do the pre-flight and post-flight checks, the Prometheus firing-alert check, or apply a ChaosEngine file
test-chaos-gameday.sh	kept, partial	load across a 3-fault sequence. The CLI runs load + one fault (resilience run) or several faults without load (disruption run), not both together
verify-kafka-policies.sh	kept, partial	kates security netpol audits NetworkPolicies without touching the cluster, and kates kyverno violations lists violations. Neither runs the live probe pod that tries to open a socket to the brokers
setup-kafka-chaos.sh	kept	registers chaos infrastructure with a Litmus portal and installs the experiments and RBAC. No CLI equivalent; test-chaos-gameday.sh needs it
test-common.sh	kept	sourced by test-chaos-gameday.sh
download-charts.sh	kept	fetches litmus-core into the checkout and builds kafka-cluster deps. kates charts pull bundles published charts, which is a different job
deploy-velero.sh	kept	the CLI does not install Velero
load-images-to-kind.sh, pull-images.sh, setup-registry.sh, registry-status.sh, cleanup-registry.sh	kept	local registry and kind image preloading: no CLI equivalent
deploy-kafka-generic.sh, deploy-apicurio.sh	kept	CI calls them (ci-*.yml, integration.yml)
build-legacy-kafka-image.sh, test-mm2-migration.sh, mm2-kafka-cli.sh	kept	ci-mirror-maker2.yml calls the first two, and the migrate code refers to all three. kates migrate image build / migrate lab are the CLI counterparts, but CI still runs the scripts
connect-smoke-test.sh, native-smoke-test.sh, native-e2e.sh, full-local-test.sh	kept	CI or make test harnesses
install-kates.sh	kept	the installer for the CLI itself (release + lint jobs)
connect-plugin-loader.sh	kept	init-container plugin downloader; not a CLI job
common.sh, retry.sh	kept	shared helpers; retry.sh is used by CI
check-*.sh, check-*.py, gen-chart-table.sh, gen-version-matrix.sh, gen-dashboards.py, book_metrics.py, chart_bumps.py, prepare-book-render.sh, test_*.py, book-metrics-baseline.json	kept	CI gates and generators
convert-callouts.sh, debug-kafka-connection.sh, debug-resilience.sh, deploy-headlamp.sh, deploy-kafka-kates.sh, download-apicurio-chart.sh, restart-docker.sh, run-it-tests.sh, debug-connect-values.yaml, patch.yaml, patch-mongo.yaml, krafter-values.yaml.bak	kept, out of scope	nothing in the repo references them, but none duplicates a CLI command (deploy-kafka-kates.sh is a Kafka-only install, same gap as deploy-kafka.sh). Candidates for a separate dead-file cleanup

Checks

• shellcheck -x -S warning -e SC2317,SC2034,SC2155 scripts/*.sh kates-ci.sh passes (Lint · shellcheck).
• make check-help: all 156 targets are documented.
• make -n test test-load test-capacity kafka-detect expands to the kates commands above.
• ResilienceTutorialTest: 30/30 pass against the edited resilience.json.
• git grep finds no reference to a removed script outside the two plan docs above.
• Not run: the make test* targets against a live cluster.