EPISODE #2

BY THE ORDERs OF EXCELLENCE AND CONTINUITY

NEED FOR SPEED - THE CONTINUOUS QUALITY GATEKEEPER

>

GitHub Actions

>

GitHub Actions

>

Java

>

Java

>

JUnit

>

JUnit

>

Selenium

>

Selenium

MISSION BRIEF

Last episode, we turned a 26-hour test marathon into a 20-minute parallel showcase.

But one-off speed was just the beginning.

 

This time? We brought discipline.


Every test unified under a single framework - clean structure, crystal-clear intent, ironclad reliability.

No more guessing, no more tangled dependencies.

We optimized the entire suite - every test refined and aligned.

Then we took the smart approach: selecting the most meaningful, high-impact tests to power the pipeline runs.

 

Every deploy now triggers these curated checks automatically.

Pipelines verify every release - results shipped, risks exposed, decisions backed by real data.


Our ROA framework forges trust.

Our pipeline enforces it.

OP 01: THE INITIATION PROTOCOL

Unification wasn't optional - it was mission critical.


We walked into fragmented suites, tribal coding styles, and accidental state-sharing. Regression wasn't strategy - it was folklore.

So we went line by line, cutting through noise and symmetry-forging structure.


The initial phase was unification of the tests - making them easy to understand, simple to maintain, and fully optimized for parallel execution.

For this, we deployed our soon-to-be-released ⏳ E2E Test Framework - The Ring of Automation 💍 (ROA).


Every test gained:

  • Explicit lifecycle management - setup and teardown couldn't leak

  • Isolation by design - no side effects, no dependencies, no hidden state

  • Deterministic outcomes - failure with answers, not excuses

  • Sharp readability - one way to do everything, no distractions

 

From their massive set of regressions, we cut through the noise.


They wanted a pipeline, but a pipeline demands unified, consistent tests, written in a format anyone can understand.

So we dissected their 2,000 regression cases, extracted the most critical scenarios, and reforged them into 100 high-priority tests. These became the foundation of their smoke testing, instantly revealing whether a release was worthy to advance.


Lean. Focused. Deadly.

Legacy flakiness vanished.


Tests didn't just "pass," they communicated: when they failed, we knew exactly why.

With a unified core, the suite became reliable, maintainable, and ready for real automation.


The chaos went silent.

Order took the lead.

>

The Code Behind the Protocol

>

The Code Behind the Protocol

OP 02: THE PIPLINE CORE

We architected a pipeline that lives inside their application repo, designed to sense every codebase heartbeat and respond with surgical test execution. No ceremony. No waste. Just intelligent automation that thinks before it acts.


Our Target:

  • Test batch execution: under 2 minutes

  • Complete pipeline: under 5 minutes


The Architecture. Pipeline operates in four synchronized phases:

  • Merge Detection → Code hits main branch

  • Deploy Signal → Automatic deployment to stable environment

  • Smart Scoping → Codebase detection exposes targeted test scope

  • Targeted Execution → Triggers scoped reusable test workflow


The Intelligence. Instead of running the entire regression test on every commit, we built a detection system that maps code changes to test coverage. Frontend changes run UI tests. Backend changes run integration tests. Documentation changes run nothing.


It scopes. It targets. It executes only what matters.


The Execution. Once code merges to main, the deployment automatically triggers and at the moment it completes, our codebase detector awakens. It detects the commit diff, identifies affected service codebase, and exposes precise test tags. Then it hands off to our reusable test workflow from Episode 1 - the same parallel execution engine, now fed with targeted accuracy. The workflow fires. Tests execute against the fresh deployment. Results generate into Allure format. The report publishes directly to GitHub Pages under the application repo.


The Result. Fast feedback, cut to the core. The pipeline doesn't just run tests - it thinks about which tests to run, executes them faster than humanly possible, and delivers results with forensic clarity.

>

GitHub Pipeline Setup

>

GitHub Pipeline Setup

OP 03: ORCHESTRATION IN THE WILD

Now they wanted the entire supply chain. Real-time PR gating.


The Drama.


We run the pipeline on pull request. Merge? Earn it.


Deploy spun up an on-demand k8s environments, the grid roared as the PRs lit up red. Tests started revealing real weaknesses and now at full disclosure, broadcasting organization-wide. Failed PRs piling up the queue. DevOps weren’t happy. Resource consumption spiked.

Developers started to bargain and promoted a new religion called "Please-Let-Me-Merge." We found TODO comments like:


"Tests failing but business critical - can we merge anyway?"

"PR has been open for 3 days, tests keep failing, deadline is tomorrow!"


Cute. Also no.


It was time to give everyone space to breathe but keep the quality gate strong.

 

The Solution.


We switched the trigger to push to main. Post-merge execution on a stable, persistent DEV environment.

Suddenly, everyone was happy. Developers could merge, but failures became immediately visible and directly traceable to specific commits. No ambiguity. No blame games. Just clear, actionable feedback pointing to exact changes that broke the system.


Testers saw the efficiency and started leaning into ROA. They began hardening the gate - writing fewer tests, but with higher signal and better throughput fading away the legacy coverage.


And we? We watched intelligent orchestration do what brute force never could: make testing immediate, faster and smarter.

MISSION REPORT

Total pipeline: under 5 minutes execution

Core validation: 100 high-priority tests under 2 minutes

Tests crafted for precision: ROA-built, efficient, deterministic

Smart triggering: Change detection gates runs - no match, no spend


The pipeline doesn’t just run fast - it targets only what moved. Everyone gets immediate, relevant feedback. No noise. No waiting.


The 100-test core became their smoke checkpoint - the definitive answer to "is this release ready?" in under 2 minutes.


Testers started shifting towards ROA framework over their legacy system. Why? Because every test now had immediate visibility and direct impact on the development cycle.


The system that started as a 26-hour overnight ritual became a 2-minute precision instrument that is actually trusted.