DevOps Engineering

Connect a verified change to an observed running service and recoverable operations. Model knowledge cannot establish the live artifact, permissions, health or restore state; retrieve that evidence.

Workflow

1. Identify environment, service, artifact/revision, permissions and current operational state. Read existing pipelines/runbooks and preserve unrelated resources.
2. Define the deployment or operational outcome, health signals, failure threshold, recovery plan and actual approval boundary.
3. Inspect the relevant mechanism: reproducible build, credentials, config/state isolation, supply-chain verification, readiness, rollout, traffic or backup.
4. Make authorized changes and run focused checks. Promote the same immutable artifact where practical; test signature/attestation enforcement at the mechanism that actually enforces it.
5. Observe rollout against declared health/traffic criteria. A configured pipeline is not evidence that deployment or acceptance happened.
6. Exercise rollback/restore where required by risk. Record data/schema incompatibilities and irreversible limits; a backup's existence is not restore proof.
7. Report local validation, integration, publish/deploy and live acceptance as separate states with artifact identity and remaining uncertainty.

Optional references

Pipelines, release/rollback, deployment strategies, supply chain, containers, orchestration, IaC, environments, secrets, GitOps, observability, reliability, incidents, GitHub/Azure governance recipes.

Current user authorization applies; do not add a permission checkpoint merely because the work is operational. Retrieve current official documentation for version-sensitive commands/configuration and state unavailable evidence rather than refusing to progress without a sheet.