DevOps QA — testing in a continuous-delivery world
The NZ-localised guide to testing when releases ship many times a day.
In a DevOps shop the old model — a test phase, a sign-off, a quarterly release — is gone. Code goes to production continuously, runs in containers on Kubernetes, hides behind feature flags, and rolls out a few users at a time. The quality question changes from “did the test phase pass?” to “is this release safe to keep promoting, right now, based on what production is telling us?” This track teaches you to test in that world.
From the container to the canary
Container & Kubernetes Testing
Testing Docker images and Kubernetes deployments. Health and readiness probes, config and secret handling, and spinning up ephemeral test environments that match production.
~30 min read · ~70 min with exercises · DevOps QA
Lesson 2Feature Flags & Progressive Delivery
Decoupling deploy from release. Testing both states of a flag, flag combinations, and the stale-flag risk that quietly turns into a defect months later.
~30 min read · ~70 min with exercises · DevOps QA
Lesson 3Canary & Blue-Green Validation
Automated canary analysis, rollback triggers, and synthetic smoke checks. Comparing the canary against the baseline, and gating promotion on what observability tells you.
~30 min read · ~70 min with exercises · DevOps QA
Lesson 4Azure DevOps for Testers
Azure Pipelines (YAML), publishing test results, branch policies and PR validation builds, quality gates, and slotting Playwright, Selenium, or k6 into a pipeline — plus when an NZ shop picks Azure DevOps over GitHub Actions or GitLab CI.
~30 min read · ~70 min with exercises · DevOps QA
Testing moves out of a phase and into the pipeline
For most of testing’s history, quality had a place in time: a test phase between development and release. DevOps removed that place. When Te Whatu Ora or a bank deploys many times a day, there is no week to test and no gate a person stands at. Testing has to happen inside the pipeline, and the pipeline has to be trustworthy enough to ship without a human in the loop on every change.
That does not make testing smaller — it makes it different. The system under test is now a container image, not a build on a shared server. A feature can be deployed to production and still switched off, so “released” and “deployed” stop meaning the same thing. A change reaches one percent of users before it reaches everyone, and the decision to promote it is made by comparing live metrics, not by reading a test report. Each of those shifts is a new place a tester adds value — and a new place a defect can hide.
This track teaches those three shifts in the NZ context. By the end you will be able to test a containerised application and its Kubernetes deployment, build a test strategy around feature flags, and validate a canary or blue-green release using the same observability signals that drive an automated rollback — and write the evidence that shows the release was safe to promote.
Other specialised tracks
API Testing
Contract testing, status codes, and idempotency — the checks a canary smoke test leans on.
SpecialisedPerformance Testing
Latency, throughput, and the baselines a canary is measured against.
SpecialisedChaos Testing
Deliberately breaking things to prove the system, and its rollbacks, recover.