The cost of provisioning by hand
At MetaZenIT I wanted to offer hosting for Frappe applications. Bringing a site online meant preparing its environment, installing the application and making it reachable at its domain. My goal was to set that work in motion from a self-service panel.
My responsibility
As founder and technical lead, I adapted the software and operated the servers it ran on. I maintained Docker images, investigated failed deployments and changed Frappe components when a fix belonged in their code.
From the panel to a running site
Frappe Press managed applications, sites and domains. Frappe Agent carried out infrastructure operations on remote servers. Docker images supplied the application environments. DNS, a reverse proxy and TLS connected the customer’s domain to the site.
Tracing failures across components
A deployment error could originate inside the agent. I investigated a failure during proxy initialization in Frappe Agent and fixed how it accessed its configuration lock. The upstream project merged the change.
Adapting and deploying the platform
I built images, adapted components for Ubuntu 24.04 and integrated the agents with Press. I maintained changes in forks and worked on both the server configuration and the code that coordinated operations.
Production outcome
The platform provisioned and hosted Frappe applications in production. It served one paying customer during that period at MetaZenIT.
What I learned
Following a provisioning request from the panel through to its domain gave me a way to narrow down incidents. The site’s state, the agent operation and the proxy configuration each answer a different question about the same deployment.