Infrastructure

Dockerfile Standards

Container image expectations for Assistance-operated environments


Dockerfiles define what runs inside your containers. Assistance reviews and operates the platform around container images, but the application team remains responsible for what the image contains and how the process behaves unless application engineering is explicitly in scope.

What a production-ready image needs#

RequirementWhy it matters
Reproducible buildCI, rollback, and incident review need known inputs and tags
Minimal runtimeSmaller attack surface, faster pulls, and less patch noise
Non-root processReduces impact if the application is compromised
Health and shutdown behaviorEnables safe restarts, rollouts, autoscaling, and incident triage
Externalized configurationAvoids rebuilding images for environment-specific settings
Clear ownershipSomeone must approve dependency, base-image, and vulnerability changes

What Assistance operates#

AreaAssistance responsibility
ReviewDockerfile and image workflow review for managed environments
Registry integrationrepository layout, tag conventions, promotion rules, retention, and pull credentials where scoped
Security workflowscanning integration, base-image refresh guidance, and exception tracking where scoped
Runtime fitresource requests, exposed ports, health checks, volume assumptions, and deployment target compatibility
Change processplanned base-image or runtime changes, rollback notes, and production validation checks

What the customer owns#

AreaCustomer responsibility
Application codecorrectness, tests, startup behavior, graceful shutdown, and dependency compatibility
Base image choicelanguage/runtime version, license acceptance, and upgrade timing unless delegated
Vulnerability remediationfixes, risk acceptance, and release approval for image contents
Secrets and configno secrets in images, correct environment variables, and safe application consumption
Build pipelinesource provenance, build triggers, and release promotion unless CI/CD operation is in scope

Assistance review checklist#

  • pinned or intentionally managed base image
  • deterministic dependency installation
  • multi-stage build where it materially reduces risk or size
  • non-root runtime user where compatible
  • no secret material copied into layers
  • explicit ports and runtime command
  • health endpoint or documented readiness behavior
  • log output suitable for the platform collector
  • known persistent paths, if any
  • documented owner for vulnerability and dependency changes

Change workflow#

Request review when changing language runtime versions, base images, package managers, exposed ports, persistent paths, health checks, or build secrets. Emergency changes may be needed for actively exploited vulnerabilities, but Assistance still needs validation evidence before promoting the image into a managed production environment.

Not included by default#

  • rewriting application build systems
  • fixing application dependency conflicts
  • owning license/compliance decisions for packages in the image
  • guaranteeing performance for images that bypass review
  • accepting images without an owner and rollback path

Getting started#