Devsolved
The Symptom
Symptom
Developers complained that GitLab CI pipelines were taking over 45 minutes just to push to the registry. EC2 nodes in the staging cluster were constantly running out of disk space (`No space left on device`).
Root Cause Analysis
Root Cause
The Dockerfile was poorly structured. It copied the entire directory (including `node_modules` and `.git`), ran `npm install`, and then ran `npm run build`. Every single file change invalidated the Docker layer cache, forcing a complete re-download of all dependencies. Furthermore, the base image was the bloated `node:18` (which is almost 1GB) instead of Alpine, and the `.dockerignore` file was completely missing.
Code
dockerfile
FROM node:18
# The fatal flaw: Copying everything before installing invalidates cache!
COPY . .
RUN npm install
RUN npm run build
CMD ["npm", "start"]Resolution
Resolution
We implemented a Multi-stage build using `node:18-alpine`. We explicitly copied `package.json` first to leverage layer caching, installed dependencies, built the app, and then copied ONLY the `dist/` folder into a pristine, tiny production image.
Code
dockerfile
FROM node:18-alpine AS builder
WORKDIR /app
COPY package*.json ./
RUN npm ci
COPY . .
RUN npm run build
FROM node:18-alpine AS production
WORKDIR /app
COPY --from=builder /app/dist ./dist
CMD ["node", "dist/main.js"]
Discussions 0