Clive - Technical Viability Check: Using Gemini...
Technical Viability Check: Using Gemini API + Make.com to self-heal a WordPress server. Is this loop robust?
Hi everyone, I am a non-technical founder validating the technical viability of a simple self-healing WordPress hosting setup using free tiers. Before I spend weeks trying to build it via prompts, I want to make sure I'm not missing an obvious architectural flaw.
The Proposed Stack:
Hosting: Oracle Cloud Free Tier (Ubuntu VM running WordPress & MySQL via Docker Compose + Caddy Server).
Brain: Google AI Studio (Gemini 2.5 Flash API) via system instructions to read error logs.
Glue: Make.com webhook watches for a 5xx error β Sends last 50 lines of Nginx logs to Gemini β Gemini chooses a predefined command (e.g., docker restart mysql-container) β Calls a Google Cloud Run Function to execute it via SSH.
My Questions:
Will Gemini 2.5 Flash reliably stick to predefined commands without hallucinating bad syntax that breaks the container?
Is using an SSH bridge from a Google Function to an OCI instance robust enough for basic self-healing?
Note: Since I am non-technical and will be building this using No-Code tools, please explain any major risks as simple business logic problems rather than raw code.
Appreciate any feedback!
Hi everyone, I am a non-technical founder validating the technical viability of a simple self-healing WordPress hosting setup using free tiers. Before I spend weeks trying to build it via prompts, I want to make sure I'm not missing an obvious architectural flaw.
The Proposed Stack:
Hosting: Oracle Cloud Free Tier (Ubuntu VM running WordPress & MySQL via Docker Compose + Caddy Server).
Brain: Google AI Studio (Gemini 2.5 Flash API) via system instructions to read error logs.
Glue: Make.com webhook watches for a 5xx error β Sends last 50 lines of Nginx logs to Gemini β Gemini chooses a predefined command (e.g., docker restart mysql-container) β Calls a Google Cloud Run Function to execute it via SSH.
My Questions:
Will Gemini 2.5 Flash reliably stick to predefined commands without hallucinating bad syntax that breaks the container?
Is using an SSH bridge from a Google Function to an OCI instance robust enough for basic self-healing?
Note: Since I am non-technical and will be building this using No-Code tools, please explain any major risks as simple business logic problems rather than raw code.
Appreciate any feedback!
