When You're the Whole Platform Team: A Solo DevOps Operating Model

It's a common and stressful pattern: a small company hands one person — often a junior — the keys to everything platform. Deployments, observability, networking, wildcard domains, SSL, the Traefik config nobody else understands. There's no team, no operating model, just a growing pile of responsibilities and the quiet dread that if you're sick, the whole thing is unowned. You can't hire a team by Friday, but you can build an operating model that keeps "solo platform owner" from meaning "single point of failure."

The real problem isn't workload — it's that it's all in your head

A solo owner can do the work. The fragility is that the knowledge is undocumented and the processes are improvised. Fix those two and the load becomes survivable (and you become un-blocking-able). Three moves do most of it.

Move 1: standardize deploys so they're boring

Every deploy done a slightly different way is a separate thing to remember. Collapse them into one path:

• One repeatable deploy method, with the exact image digest tested and promoted rather than rebuilt per environment.
• A rollback that's pre-written and rehearsed, so a bad deploy is a 60-second action, not a research project.
• Health checks that confirm the new release is actually serving, not just that a container started.

When deploys are uniform, you stop holding a dozen special cases in working memory.

Move 2: externalize the knowledge into runbooks

The single highest-leverage thing a solo owner can do is write the symptom-named runbooks that turn tribal knowledge into a checklist:

• One page per common failure ("SSL expired," "Traefik 502s," "deploy stuck"), with the literal commands and file paths.
• The Traefik/edge config documented — especially the Cloudflare-and-Traefik seam where wildcard SSL and routing actually break.
• Enough that a competent stranger (or future-you at 2am) could recover without you.

Runbooks are how a one-person platform survives that person being away.

Move 3: automate the edge so it doesn't page you

The recurring solo-owner fires — expired certs, a renewal that didn't reload — should be automated and monitored, not manually babysat:

• Wildcard SSL via DNS-01 with persisted state, renewal that reloads the proxy.
• A cert-expiry alert that warns days ahead — monitoring beyond a basic ping.
• Alerts scoped and routed so you're paged for real incidents, not noise — the alerting last-mile.

See everything in one place

A solo owner can't watch ten dashboards. Consolidate the platform's health into one view so a glance tells you what's wrong.

A single operational view in ServerCompass — for a one-person platform team, one pane over deploys, SSL, and health beats juggling separate tools.

Operating-model checklist

• One standardized, rehearsed deploy + rollback path.
• Symptom-named runbooks covering SSL, Traefik, deploys.
• Edge (wildcard SSL, renewal) automated and monitored, not manual.
• Alerts scoped/routed to real incidents.
• One consolidated health view.

Takeaway

Being the only platform owner is workable; being the only person who knows how it works is the risk. Standardize deploys, write the runbooks, and automate the edge — and you turn a fragile pile of responsibilities into a durable operating model that survives you taking a day off.