Docker Architecture and Components

Overview

Describe the Docker client–daemon path and the roles of dockerd, containerd, and the OCI runtime so you can debug “where did my request fail?”

You talk to the Docker CLI; it calls the Docker Engine API on dockerd. The daemon uses containerd and runc to create containers on the host kernel.

This is a core tutorial in Module 1 · Container Fundamentals of the REBASH Academy Docker for Cloud & DevOps Engineers series — written for Cloud, DevOps, Platform, and SRE engineers.

Prerequisites

• Introduction to Containers and Docker

Learning Objectives

By the end of this tutorial, you will be able to:

• Trace CLI → dockerd → containerd → runc
• Distinguish image, container, and registry
• Know what docker context selects
• Relate namespaces/cgroups to isolation

Architecture

This topic’s control points and relationships are shown below.

Theory

What

Docker’s architecture separates the CLI, the Engine (dockerd), containerd, and an OCI runtime such as runc, all sitting on the host kernel. Images, containers, networks, and volumes are API objects the daemon manages. Registries store images remotely.

Why

Troubleshooting “Docker is broken” means knowing which layer failed: CLI talking to the wrong context, daemon down, pull failing in containerd, or runtime/kernel limits. Production platforms (and Kubernetes) reuse pieces of this stack — especially containerd and OCI — so the mental model transfers.

How it works

You type docker commands; the CLI calls the Engine API (local Unix socket or TCP). dockerd orchestrates networks, volumes, and higher-level UX. It delegates image pull and container lifecycle to containerd, which invokes runc to start the process in namespaces/cgroups. An image is immutable layers plus config; a container adds a writable layer and process state; a registry (Docker Hub, GitHub Container Registry, Amazon Elastic Container Registry) is the remote store.

Component	Role
Docker CLI	User interface (docker)
dockerd	Engine API, networks, volumes
containerd	Image pull, container lifecycle
runc	OCI runtime — starts the process
Host kernel	Namespaces, cgroups, filesystems

Key concepts

• Client–daemon — CLI is not the container runtime
• Contexts — CLI can point at remote engines
• Rootful vs rootless — privilege model of the daemon
• Shim / supervisors — keep containers reparented correctly after daemon restarts

Common pitfalls

• Exposing the Docker socket without understanding it is root-equivalent
• Debugging inside the CLI when dockerd is the failing component
• Assuming Desktop networking equals Linux Engine networking
• Confusing image ID, digest, and tag

Hands-on Lab

Objective

Prove the Docker client talks to a healthy Engine by collecting version, info, context, and disk-usage evidence files.

Prerequisites

• Docker Engine or Docker Desktop running
• Permission to run docker commands

Lab environment

Workspace: ~/rebash-docker/module-01-arch

Local Docker daemon. Evidence files only — no long-running containers required.

Terminal

mkdir -p ~/rebash-docker/module-01-arch && cd ~/rebash-docker/module-01-arch

Real-world scenario

A developer reports “Docker is broken” after switching laptops. Before you restart services, you capture client versus server versions, active context, storage driver details, and disk usage — the same artefacts SREs attach to incident tickets.

Step-by-step tasks

Task 1 – Split client and server version evidence

The CLI and daemon can differ; record both sides explicitly.

Terminal

cd ~/rebash-docker/module-01-arch
docker version | tee docker-version.txt
docker version --format 'Client={{ "{{" }}.Client.Version{{ "}}" }} Server={{ "{{" }}.Server.Version{{ "}}" }}' | tee version-split.txt
grep -q 'Client:' docker-version.txt
grep -q 'Server:' docker-version.txt

Expected output

docker-version.txt shows Client and Server blocks; version-split.txt has both version strings on one line.

Task 2 – Engine info and storage driver

docker info reveals runtime, cgroup driver, and storage driver — common root causes when containers fail to start.

Terminal

cd ~/rebash-docker/module-01-arch
docker info | tee docker-info.txt
docker info --format 'StorageDriver={{ "{{" }}.Driver{{ "}}" }} CgroupDriver={{ "{{" }}.CgroupDriver{{ "}}" }}' | tee info-drivers.txt
grep -q 'Storage Driver' docker-info.txt
test -s info-drivers.txt

Expected output

docker-info.txt is multi-line; info-drivers.txt names the storage and cgroup drivers.

Task 3 – Context and disk footprint

Contexts route the CLI; docker system df shows image/container/volume pressure on the node.

Terminal

cd ~/rebash-docker/module-01-arch
docker context ls | tee docker-contexts.txt
docker system df | tee docker-system-df.txt
grep -q 'CURRENT' docker-contexts.txt
grep -E 'Images|Containers|Local Volumes' docker-system-df.txt

Expected output

docker-contexts.txt marks the current context with *; docker-system-df.txt lists Images, Containers, and Local Volumes rows.

Validation steps

• docker-version.txt and version-split.txt prove client and server are reachable
• info-drivers.txt records storage and cgroup drivers
• docker-contexts.txt and docker-system-df.txt capture context and disk summary

Common errors and fixes

Error	Cause	Fix
Cannot connect to the Docker daemon	dockerd stopped or wrong context	docker context ls; start Engine or switch context
Server section missing in version	Daemon unreachable	Check Docker Desktop or systemctl status docker
permission denied	Socket access	Fix docker group membership or use rootless mode

Challenge exercise

Switch context temporarily (if a second context exists), re-run docker context ls, then switch back and append a one-line note to docker-contexts.txt.

Terminal

cd ~/rebash-docker/module-01-arch
ALT="$(docker context ls --format '{{ "{{" }}.Name{{ "}}" }}' | grep -v "$(docker context show)" | head -n 1 || true)"
if [ -n "$ALT" ]; then
  docker context use "$ALT"
  docker context ls | tee docker-contexts-alt.txt
  docker context use default 2>/dev/null || docker context use "$ALT"
fi
echo "Active context after lab: $(docker context show)" | tee -a docker-contexts.txt

Expected output

If an alternate context exists, docker-contexts-alt.txt shows the switch; the final line names the restored active context.

Learning outcomes

• Separated Docker CLI client output from Engine server metadata
• Identified storage and cgroup drivers from docker info
• Documented active context and node disk usage for troubleshooting handovers

Cleanup

No containers were created. Remove evidence files if you do not need them:

Terminal

cd ~/rebash-docker/module-01-arch
rm -f docker-version.txt version-split.txt docker-info.txt info-drivers.txt \
  docker-contexts.txt docker-contexts-alt.txt docker-system-df.txt 2>/dev/null || true

Validation

• Lab commands run under ~/rebash-docker/module-01-arch/
• Evidence files prove client, Engine, context, and disk usage were captured
• You can explain each Theory section in your own words
• You can describe one production failure mode for this topic

Code Walkthrough

Production practice for Docker Architecture and Components always combines:

1. Inspect before you change (status, plan, logs, dry-run)
2. Prefer reversible, documented changes (Git, IaC, drop-ins, version pins)
3. Capture evidence (command output, pipeline logs) for handovers
4. Prefer current tools and APIs over legacy shortcuts
5. Least privilege — escalate credentials only when required

Keep runbooks short enough to follow under pressure. Automate checks; keep humans for judgement.

Security Considerations

• Treat credentials and tokens for docker as privileged — never commit them
• Prefer short-lived auth (OIDC, roles, SSO) over long-lived keys
• Validate blast radius before apply/deploy/delete operations
• Restrict who can approve production changes
• Collect audit logs; limit who can read sensitive traces

Common Mistakes

Exposing the Docker socket without understanding it is root-equivalent

Validate assumptions against the Theory section and official docs before changing production.

Debugging inside the CLI when dockerd is the failing component

Lab shortcuts (open security groups, admin roles, skip approvals) must not ship unchanged.

Changing production without a rollback path

Always know how to revert (previous artefact, prior release, state rollback, DNS failback).

Best Practices

• Encode Docker Architecture and Components changes as code and review them in pull requests
• Pin versions (images, modules, actions, provider plugins)
• Separate environments with clear promotion gates
• Alert on symptoms with runbooks attached
• Destroy lab resources; tag everything with owner and expiry where possible

Troubleshooting

Symptom	Likely cause	Fix
Auth / permission denied	Wrong identity, policy, or scope	Check caller identity, roles, and least-privilege policies
Timeout / no route	Network, DNS, security group, or endpoint	Trace path, DNS, and allow-lists before retrying
Drift / unexpected plan	Manual change or wrong state/workspace	Reconcile desired vs actual; avoid click-ops on managed resources
Pipeline/job red	Flaky step, cache, or missing secret	Read failing step logs; bisect recent workflow/config changes
Cost spike	Idle load balancer, NAT, oversized compute	Inventory billable resources; stop/delete labs promptly

Summary

Docker Architecture and Components is essential for Cloud and DevOps engineers working with docker. Practise the lab until the inspection and change path is muscle memory, then continue the track.

Interview Questions

1. Role of dockerd versus the CLI?
2. What storage driver concerns matter on Linux?
3. How do containerd/runc fit the stack?
4. Why does architecture knowledge help troubleshooting?
5. What is the difference between create and run?

Sample answer — question 2

Use docker info and inspect to see driver/runtime details.

Sample answer — question 4

Limit who can talk to the daemon socket.

Related Tutorials

• Course overview
• Docker Installation and Setup

References

• Docker Engine architecture