← All writing

Essential Docker CLI Commands & Flags Used Daily in Production with Real-World Scenarios

· Sourabh G Kulkarni

Master essential Docker CLI commands and flags: buildx, exec, logs, cp, inspect, stats, networks, and prune for debugging and operating production systems.

Almost every engineer starts with the same four Docker commands: docker build, docker run, docker ps, and docker stop.

And when everything is working smoothly on your local machine, that basic quartet feels sufficient. But when a production database container begins thrashing CPU at 3:00 AM, a CI/CD build fails due to a missing environment variable, or an orphaned network interface locks up a host port, those basic commands fall short.

Operating containers in production requires command-line fluency: knowing how to filter JSON metadata using Go templates, stream live memory telemetry, surgically extract files from crashed containers, and debug virtual network bridges.

Here are the essential Docker CLI commands and flags used daily by DevOps and platform engineers, accompanied by real-world operational scenarios.


1. Surgical Inspection: Extracting Exact Metadata with --format

When debugging, running docker inspect <container> dumps hundreds of lines of raw JSON to your terminal. Manually scrolling through that wall of text to find a container's IP address or restart count is slow and inefficient.

Docker CLI natively supports Go template formatting via the -f / --format flag.

Scenario A: Extract the Container's Internal Bridge IP Address

docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' my-app
# Output: 172.17.0.4

Scenario B: Extract Real-Time Health Check Status & Failure Logs

docker inspect -f '{{.State.Health.Status}} - Failed: {{.State.Health.FailingStreak}}' my-app
# Output: unhealthy - Failed: 4

Scenario C: Format docker ps into a Clean, Readable Table

Instead of truncating columns across narrow terminal windows, output only the fields you care about:

docker ps --format "table {{.ID}}\t{{.Names}}\t{{.Status}}\t{{.Ports}}"

2. Real-Time Resource Monitoring: docker stats

When a host server is under high load and you need to immediately identify which container is leaking memory or maxing out CPU cores:

docker stats --no-stream

Output:

CONTAINER ID   NAME         CPU %     MEM USAGE / LIMIT     MEM %     NET I/O          BLOCK I/O
a1b2c3d4e5f6   postgres     2.41%     412MiB / 8GiB         5.03%     12.4MB / 8.2MB   1.2GB / 400MB
f6e5d4c3b2a1   node-api     184.20%   3.81GiB / 4GiB        95.25%    84.1MB / 92.4MB  45MB / 12MB
  • In one glance, you immediately see that node-api has pegged two CPU cores (184%) and is within 200MB of triggering an OOMKilled termination (95.25% MEM).

Pro Tip: Stream Specific Format for Scripting

docker stats --format "{{.Name}}: CPU {{.CPUPerc}} | MEM {{.MemPerc}}"

3. Production Log Troubleshooting: docker logs

Using raw docker logs <container> on a service that has been running for six months will dump gigabytes of text into your terminal, freezing your shell session.

The Essential Flags

  • -f (Follow): Stream live logs in real time (like tail -f).
  • --tail 100: Print only the last 100 lines.
  • --since 15m: Show only logs generated within the last 15 minutes (or pass an ISO timestamp 2026-10-31T01:00:00).
  • -t (Timestamps): Prepend an exact UTC timestamp to every log entry (vital for correlating events with Prometheus/Grafana alerts).
docker logs -f --tail 200 -t --since 30m payment-service

4. Disaster Recovery & File Extraction: docker cp

Imagine an application container crashes with a segmentation fault on startup. You need to inspect a crash dump file or a local SQLite database stored inside the container, but because the container is stopped, you cannot docker exec into it.

docker cp works even when the container is completely stopped:

# Copy a file from a stopped container to your host machine
docker cp crashed-container:/app/logs/crash.dump ./crash.dump

# Copy an updated configuration file from your host directly into a container
docker cp ./nginx.conf web-server:/etc/nginx/nginx.conf

5. Non-Destructive Interactive Debugging: docker exec

When entering a running container, you almost always need an interactive TTY session:

docker exec -it <container> /bin/sh

Scenario: Running as Root Inside an Unprivileged Container

For security reasons, production containers often run as non-root users (USER node or USER 1001). If you need to install debugging tools (like curl or tcpdump) to triage a live network issue, use the -u 0 (root) override flag:

docker exec -u 0 -it api-server apt-get update && apt-get install -y tcpdump

6. Network Triage: docker network inspect

When two containers fail to communicate over internal Docker bridge networks (e.g., web throws Connection Refused trying to reach db):

  1. Inspect the Bridge Network to Verify Both Containers Are Connected:
    docker network inspect bridge
    
  2. Create an Isolated Custom Bridge Network: Containers on the default bridge network cannot resolve each other by container name. Custom bridge networks provide automatic internal DNS resolution:
    # 1. Create a custom network
    docker network create app-network
    
    # 2. Connect running containers on the fly
    docker network connect app-network web
    docker network connect app-network db
    
    Now web can reach db simply by addressing http://db:5432.

7. Storage Hygiene: Safe vs Deep Prunes

# Level 1: Safe Prune (Removes stopped containers & dangling untagged images)
docker system prune

# Level 2: Comprehensive Prune (Also removes all unused images, not just dangling ones)
docker system prune -a

# Level 3: Deep BuildKit Cache Purge (Frees gigabytes of intermediate build layers)
docker builder prune -a --force

# Level 4: Volume Cleanup (CAUTION: Permanently purges unattached persistent volumes)
docker volume prune

Summary Reference Table

TaskCommand
Inspect Specific JSON Fielddocker inspect -f '{{.State.Pid}}' <container>
Stream Tail of Recent Logsdocker logs -f --tail 100 --since 15m -t <container>
Live Resource Telemetrydocker stats --no-stream
Extract File from Stopped Containerdocker cp <container>:/path/to/file ./local/
Debug as Root Userdocker exec -u 0 -it <container> /bin/sh
Connect Container to Networkdocker network connect <network> <container>
Reclaim Host Disk Spacedocker system prune -a && docker builder prune -a