Checkpoint Lab — CI/CD Performance Gates in GitHub Actions, GitLab CI, and Jenkins

1. Exact assumptions and ceilings

2. Freeze inputs

Hash JMX, properties, SLO, evaluator, launcher and fixture. The launcher writes an input manifest automatically. Do not edit them between the two phases.

3. Predictions

• Both runs: exactly 100 JTL Work rows and 100 target events.
• Degraded: engine execution 0, p95≈180 ms, error=0%, gate exit10.
• Baseline: engine 0, p95≈40–60 ms, error=0%, gate exit0.
• Both dashboards/logs survive; degraded evidence is never overwritten.

4. Provision and inspect

python .\tools\provision_jmeter.py --dest .tools
$env:JMETER_HOME = (Resolve-Path ".\.tools\apache-jmeter-5.6.3").Path
& "$env:JMETER_HOME\bin\jmeter.bat" -v
java -version
python --version
Get-Content .\policy\slo.json

5. Execute expected SLO failure

python .\tools\run_performance_gate.py `
  --fixture-mode degraded `
  --run-id p28-check-degraded `
  --out .\results\p28-check-degraded
$DegradedExit = $LASTEXITCODE
if ($DegradedExit -ne 10) {
  throw "Expected SLO exit 10, got $DegradedExit"
}

6. Preserve and inspect before repair

Get-Content .\results\p28-check-degraded\launcher.json
Get-Content .\results\p28-check-degraded\gate.json
Get-Content .\results\p28-check-degraded\target-summary.json
Get-Item .\results\p28-check-degraded\results.jtl,
         .\results\p28-check-degraded\jmeter.log,
         .\results\p28-check-degraded\dashboard\index.html,
         .\results\p28-check-degraded\target-events.jsonl

Require engine_exit_code=0, gate_exit_code=10, 100 JTL, 100 target, p95>120 and zero errors.

7. Prove the exit-code trap

launcher.json:
  engine_exit_code: 0
  gate_exit_code: 10
  final_exit_code: 10

gate.json:
  status: SLO_FAIL
  validity_failures: []
  slo_failures:
    - p95 ... > 120ms

This is the deliberately broken case that a JMeter-exit-only pipeline would falsely pass.

8. Restore only the fixture

python .\tools\run_performance_gate.py `
  --fixture-mode baseline `
  --run-id p28-check-baseline `
  --out .\results\p28-check-baseline
if ($LASTEXITCODE -ne 0) {
  throw "Expected PASS exit 0"
}

9. Before/after verification

10. Map the same command to CI

GitHub Actions:

name: jmeter-smoke-gate
on:
  pull_request:
permissions:
  contents: read
jobs:
  performance-gate:
    runs-on: ubuntu-latest
    timeout-minutes: 10
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-java@v5
        with:
          distribution: temurin
          java-version: '17'
      - name: Provision JMeter 5.6.3
        run: python3 tools/provision_jmeter.py --dest .tools
      - name: Run portable performance gate
        env:
          JMETER_HOME: ${{ github.workspace }}/.tools/apache-jmeter-5.6.3
        run: >
          python3 tools/run_performance_gate.py
          --fixture-mode baseline
          --run-id pr-${{ github.run_id }}
          --out results/p28-ci
      - name: Upload performance evidence
        if: ${{ always() }}
        uses: actions/upload-artifact@v4
        with:
          name: jmeter-performance-evidence
          path: results/p28-ci/
          retention-days: 7

GitLab CI:

stages:
  - test
performance_gate:
  stage: test
  timeout: 10m
  script:
    - java -version
    - python3 tools/provision_jmeter.py --dest .tools
    - export JMETER_HOME="$CI_PROJECT_DIR/.tools/apache-jmeter-5.6.3"
    - python3 tools/run_performance_gate.py --fixture-mode baseline --run-id "gitlab-$CI_PIPELINE_ID" --out results/p28-ci
  artifacts:
    when: always
    expire_in: 7 days
    paths:
      - results/p28-ci/

Jenkins:

pipeline {
  agent { label 'java17-python3' }
  options {
    timeout(time: 10, unit: 'MINUTES')
  }
  stages {
    stage('Performance gate') {
      steps {
        sh 'java -version'
        sh 'python3 tools/provision_jmeter.py --dest .tools'
        sh '''
          export JMETER_HOME="$WORKSPACE/.tools/apache-jmeter-5.6.3"
          python3 tools/run_performance_gate.py \
            --fixture-mode baseline \
            --run-id "jenkins-${BUILD_NUMBER}" \
            --out results/p28-ci
        '''
      }
    }
  }
  post {
    always {
      archiveArtifacts artifacts: 'results/p28-ci/**',
                       allowEmptyArchive: true,
                       fingerprint: true
    }
  }
}

The real PR job should use the approved baseline/real authorized target mode, not permanently run synthetic degraded mode. A dedicated gate-self-test can deliberately expect exit10.

11. Evidence packet

12. Validity statement

13. Checklist

• Only 127.0.0.1:8028.
• JMeter5.6.3 / Java17 / no plugins.
• Same JMX/properties/SLO hashes.
• 100 configured and 100/100 achieved each phase.
• Degraded engine0/gate10; baseline engine0/gate0.
• Only target delay 180→40 changes.
• p95≤120/error≤1 policy explicit.
• Artifacts survive failure in all provider patterns.
• No baseline auto-update or retry-until-green.

14. Cleanup / rollback

1. Launcher terminates its fixture in finally; confirm no listener remains on port 8028.
2. Keep degraded/baseline evidence until review.
3. Remove local .tools/apache-jmeter-5.6.3 only if desired; reprovisioning re-verifies SHA-512.
4. Delete result directories after retention policy.
5. No production/shared target, real credential, cloud CI resource, RMI engine, container cluster or OS/JVM global tuning was modified.

15. Production operating-model addition and Chapter 29 bridge

Chapter 28 adds a CI performance-gate contract: pinned runtime, common launcher, exact workload, JTL/target validity, explicit percentile/error formula, reviewed SLO/baseline, separate engine/gate/invalid-run semantics, runner/generator evidence, unconditional artifact retention, gate cadence/severity and baseline/secret governance.

Chapter 29 moves to Latency Percentiles, Throughput, Errors, Saturation, and Bottleneck Diagnosis, where the metrics behind these gates are interpreted causally rather than as isolated threshold numbers.

Knowledge check

Official references and version notes

• Apache JMeter downloads — JMeter 5.6.3 and Java 8+ requirement.
• JMeter execution guidance — GUI for authoring/debug and CLI/non-GUI for load.
• JMeter Dashboard Report — -e -o/-g -o and dashboard CSV requirements.
• GitHub Actions build/test examples — artifact upload with always() after test failure.
• GitHub setup-java — stable Java setup action; example uses v5 with Temurin 17.
• GitLab CI YAML reference — artifacts:when: always semantics.
• Jenkins Declarative Pipeline syntax — post/always, timeout, agent behavior.
• Jenkins tests and artifacts — archiveArtifacts in an always-run post block.