Page 21 - Docker

Teams that regularly hit Docker Hub rate limits in CI pipelines can cut pull times and avoid failures with a local registry acting as a pull-through cache, without manually mirroring images.

Starting Composer and npm from scratch costs minutes on every single build. The good news is that both package managers ship with robust local caches that can persist across Docker builds without inflating the image size. The key lies in correctly separating the lock file layer, the vendor layer and the cache mount, and in understanding which cache type is the right one for each situation.

The dind service shows up in nearly every GitLab CI guide on building container images, but rarely is it explained why it requires a privileged runner and what risk that carries.

If you only ever read Docker logs with docker logs, you lose the overview fast in production. Structured logging over stdout and stderr, the right log driver, and a centralized log management system are what make container logs observable, searchable, and alertable.

Most GitLab CI guides show caching using composer.lock or package-lock.json and stop there. Anyone who needs multiple independent caches per job or wants finer-grained invalidation has to go deeper into cache:key.

A slow Magento Docker environment drags down every development cycle. With the right volume strategy, Mutagen sync, correctly configured OPcache, and a running Hyva watcher, perceived speed improves dramatically, without giving up the isolation benefits of Docker.

Solving Docker performance problems starts with understanding why the three platforms have fundamentally different architectures. On Linux, Docker runs natively; on macOS and WSL2, it runs through a virtualization layer, with measurable consequences for builds, bind mounts, and I/O-heavy applications.

A correctly configured PhpStorm environment for Magento 2 saves hours every day: a Docker interpreter for precise code completion, Xdebug without manual steps, CLI runners for Magento commands right inside the IDE, and file templates that create new ViewModels, plugins and repositories in seconds.

docker system prune only cleans up locally; the growth of a self-hosted Docker registry with thousands of pushed images is completely unaffected by it. Registry-side garbage collection and thoughtful retention policies are the actual lever against runaway storage consumption.

Anyone maintaining a library or tool for multiple PHP or Node versions can hardly avoid test matrices. This article shows how parallel: matrix: in GitLab CI turns a single job definition into several parallel test runs, and how to balance runtime cost against test coverage.