Five Client Properties on Self-Managed Infrastructure
Five WordPress properties for real clients, running on a self-managed VPS behind Traefik with CI/CD, snapshot-safe deploys, and written runbooks, operated with postmortem discipline.
Status
Client DeliveryDomain
InfrastructureHeadline result
5 live client properties; documented runbooks; incident-to-postmortem operating culture
Demonstrates
Representative stack
Edge
- Traefik reverse proxy
- TLS automation
Hosting
- OVH VPS
- 5 WordPress properties
- Containerized isolation
Operations
- CI/CD deploys
- Snapshot-safe rollback
- Written runbooks
Situation
Five WordPress properties for real clients, five different owner-led organizations in five different lines of business, consolidated onto self-managed infrastructure: an OVH VPS fronted by Traefik, with containerized isolation per property.
Problem
Small-business websites usually live in one of two bad places: overpriced managed hosting that still breaks, or a hand-configured server that only one person (who left) ever understood. Either way, deploys are scary, rollbacks are prayers, and when something breaks at 9pm nobody knows the procedure.
Approach
Run client hosting like production infrastructure, scaled to small-business economics. Every property deploys through CI/CD rather than FTP-and-hope. Snapshots gate risky changes so rollback is a procedure, not an emergency. And every operational task that happened twice got written into a runbook.
Architecture and key decisions
- Traefik at the edge. One reverse proxy handles routing and TLS for all properties. Certificates renew themselves, and adding a property is configuration, not surgery.
- Isolation per property. One client’s plugin disaster cannot take a neighbor down.
- Snapshot-safe deploys. Every risky change is preceded by a restore point; rollback is rehearsed, not improvised.
- Runbooks as deliverables. Operations knowledge lives in documents, not in one person’s memory. The property survives personnel.
- Postmortems over blame. A billing-logic incident in an adjacent system became a written postmortem and a permanent audited-ledger rule. Failures become rules; rules become culture.
What shipped
The operated platform itself: five live client properties, the Traefik/Docker hosting architecture, CI/CD pipelines, snapshot procedures, and the runbook library.
Outcome
Five client properties serving real traffic on infrastructure with documented procedures, rehearsed rollback, and an incident culture that turns failures into written rules.
What this demonstrates
Operational maturity is a transferable habit, not a tool choice. The same discipline, isolate, automate deploys, rehearse rollback, write the runbook, postmortem the failure, is what I install when a client’s infrastructure is one resignation away from being unmanageable.