To be honest, I wore myself out trying to manage everything solely through the GCP GUI.
IAM, enabling APIs, permission errors. Going back and forth between screens,
- Losing track of what I configured and where
- Error messages being abstract, obscuring the root cause
- No history of click operations left behind
Repeating "Maybe it's this?" in this state, I realized a significant amount of time had passed.
The Turning Point: Opening Cloud Shell
Halfway through, it suddenly hit me,
Wait, wouldn't it be faster to launch Cloud Shell and hit gcloud?
Thinking that, I launched Cloud Shell within GCP.
From there, the situation changed instantly.
- Instantly confirm which project I'm touching with gcloud config get-value project
- Explicitly execute API enablement with gcloud services enable xxx.googleapis.com
- Even for IAM assignment errors, CLI error messages make more sense
The biggest factor was, being able to paste CUI logs to AI for consultation right there.
Compared to GUI screenshots, command + error message is overwhelmingly faster to discuss.
Reflections
Looking back now,
- GUI is for "people who know the settings"
- CUI is overwhelmingly stronger for troubleshooting
- It's better for mental health to assume Shell from the start with GCP
That was my conclusion.
For those dying in the GCP GUI, I think it's better to launch Cloud Shell once, and proceed while throwing:
- gcloud
- Logs
- Error messages
directly to AI. It will likely be faster, more accurate, and safer in the end.
This is entirely a failure story I realized after taking the long way around.
[Important] Precautions when sending logs to AI Escaping to CUI is efficient, but please be very careful about security. Make sure never to paste sensitive information like API keys, service account keys (JSON content), or passwords to AI. Please make it a habit to check that no sensitive information is included before pasting logs.
Cloud Vision OCR Series Articles
This article is part of a series on setting up and operating OCR using Cloud Vision API.
- [Summary] Cloud Vision OCR x CLI x GCP Setup & Troubleshooting Log](/archives/ocr-series-summary-en)
- Setup: Why Cloud Vision API OCR Failed with IAM and How I Finally Got It Working
- Operation: Why Running Cloud Vision OCR via CLI Was the Easiest Method
- Workflow: Why GCP x VSC x CLI is the Correct Route for Frequent OCR
- Tools: Explanation of Practical Tools Created to Run Cloud Vision OCR via CLI
- Code: Free Code: Minimal Script to Safely Run Cloud Vision OCR via CLI
- Extra: When Stuck in GCP GUI, Escaping to Cloud Shell + CUI Was Faster (Failure Story)
