Infrastructure rules¶

Category code: INF

Total rules: 12

Rules¶

Starting every service by default in local or default compose profiles wastes local compute and encourages unnecessary background activity.

Restarting noncritical services aggressively can create needless churn and repeated work.

Single-stage Docker builds often ship build tooling and temporary artifacts into runtime images, increasing image size and transfer cost.

Heavy base images increase transfer, storage, and patching footprint without improving runtime value.

Large build contexts increase build time, cache churn, and unnecessary transfer to the Docker daemon.

Leaving package indexes and caches behind increases image size with no runtime benefit.

Runtime images should not carry compilers, package managers, or build-only tooling that is not needed after build.

Containers without realistic requests and limits create waste, contention, or throttling.

Workloads that can scale horizontally but do not use autoscaling can waste cluster capacity or under-serve demand.

Very frequent probes create unnecessary traffic and container work, especially at scale.

Sidecars add useful capabilities, but they also add CPU, memory, storage, and network overhead.

Floating tags make deployments less predictable and can increase repeated pulls and churn.