Step 4: Test the Customer Journey and Release Process¶
A successful internal build is only the baseline. Exercise onboarding as a customer, verify the application and promised lifecycle, and make the release process repeatable before launching.
Exercise the Deployment as a Customer¶
Test the correct environment and customer subscription, not just an internal administrator's ability to create an instance. Non-production portals are organization-only. If your Omnistrate account uses SSO, set a separate password as documented to sign in to a non-production portal.
| Deployment model | Customer-journey test |
|---|---|
| Hosted SaaS | Subscribe to the test plan and create an instance in the provider account. Test your chosen tenancy and isolation. |
| BYOC account or VPC | Use the intended subscription, connect the customer account, complete bootstrap, confirm readiness, and deploy in the selected region and network. |
| BYOC K8s | Generate the customer onboarding kit, confirm the kubectl context, install the agent, and wait for account readiness. Use the guide's local k3d smoke-test path before testing a production-like cluster. |
| Air-gapped | Configure AWS artifact hosting and prepare a reachable customer image registry, Helm, kubectl, and valid kubeconfig. Generate and transfer the installer and images. Test installation, application readiness, and an update without internet or control-plane access; retain internal cluster and registry connectivity. |
Use the Customer Portal guide to test plan selection, subscriptions, deployment creation, endpoints, and customer access controls.
Portal walkthroughs¶
- Customer Portal walkthrough
- Admin Portal walkthrough
- Generated Customer Portal demonstration
- Day-2 operations demonstration
Define Successful Onboarding¶
- Verify that the application works through its advertised endpoint with expected authentication, not merely a
RUNNINGstatus. - Confirm that storage persists and required network restrictions and tenant isolation hold.
- Test the lifecycle you promise: a representative upgrade, scaling where supported, backup and restore where configured, and deletion in a disposable environment.
- Record recovery steps and data-retention behavior. Obtain approval before destructive tests.
- Keep a reproducible specification, pinned artifacts, plan/version/environment identifiers, test results, and a deployment overview describing responsibilities and access paths.
Debug with Evidence¶
Use the omnistrate-sre skill on failed or stalled deployments. Start with status and workflow events, then inspect rendered configuration, registry or IAM errors, probes, and workload logs. Change one cause at a time; follow the troubleshooting workflow.
Escalate in your shared Omnistrate Slack channel with environment, instance and workflow IDs, timestamps, a sanitized error, and the last change. Never paste credentials.
Make Releases Repeatable¶
Once the baseline passes, keep the specification in Git and automate validation, builds, development deployment, smoke tests, and an approved production promotion.
- Release as Active for testing; promote a tested version to Preferred for new usage.
- Treat existing-instance upgrades as a separate rollout rather than assuming Preferred-version promotion upgrades the fleet.
- Adapt the GitHub Actions example to your specification type and use scoped, rotated CI secrets.
- Follow Continuous Development and Version Rollout and Upgrades to make promotion and rollout reproducible.