Docker Cleaning Up Cli Lab

Lab 1: The “Junk Generator” (Visualize the problem)

1. Check Space: docker system df (Note the numbers).
2. Create Junk:
   • Run a container: docker run -d --name junk-app alpine sleep 1000
   • Stop it: docker stop junk-app
   • Pull an image: docker pull busybox
3. Check Space Again: docker system df (Reclaimable space increased).
4. Clean Up: docker system prune -f
5. Verify: docker ps -a (The stopped container is gone). docker images (Busybox is still there! Why? Because it is not “dangling”, just unused).
6. Deep Clean: docker system prune -a -f
7. Verify: docker images (Busybox is gone).

Lab 2: The “Orphan Volume” Hunt

1. Create Volume: docker volume create orphan-vol
2. Check List: docker volume ls
3. Prune: docker volume prune -f
4. Check List: docker volume ls (It’s gone).

Lab 1: The “Time Machine” Clean (Using Filters)

Goal: Automate cleanup safely by only deleting old junk.

1. Create “Old” and “New” Junk:
   • We can’t actually fake time easily, so we will use the logic.
   • Pull an image: docker pull hello-world (This is “New”).
2. The Command:
   • docker system prune -a --filter "until=24h"
   • Explanation: This command looks for junk created more than 24 hours ago.
   • Result: Since you just pulled hello-world, it is preserved.
3. The Comparison:
   • Now run: docker system prune -a (Without filter).
   • Result: hello-world gets deleted because it is not currently running.
   • Lesson: Use until=24h in your cron jobs (automated scripts) to avoid deleting images you just built but haven’t started yet.

Lab 2: The “Dependency Hell” (Why can’t I delete this?)

Goal: Understand image layers and why Docker sometimes refuses to delete.

1. Create Base Image:
   • Create file Dockerfile.base: FROM alpine
   • Build: docker build -t my-base:v1 -f Dockerfile.base .
2. Create Child Image:
   • Create file Dockerfile.child: FROM my-base:v1
   • Build: docker build -t my-child:v1 -f Dockerfile.child .
3. Attempt Deletion:
   • docker rmi my-base:v1
4. Result:
   • Error: image is being used by ...
   • Why? Docker protects you. If you delete the base, the child breaks.
5. Fix:
   • Delete the child first: docker rmi my-child:v1
   • Now delete the base: docker rmi my-base:v1
   • Success!

Lab 3: The “Silent Disk Killer” (Log Explosion)

Goal: Simulate a container filling the disk with logs and fix it without restarting.

1. Create the Problem:
   • Run a container that spams logs endlessly.
   • docker run -d --name spammer alpine sh -c "while true; do echo 'CRITICAL ERROR: Disk is filling up fast...'; sleep 0.01; done"
2. Observe the Growth:
   • Let it run for 30 seconds.
   • Find the log file: LOG_PATH=$(docker inspect --format='{{.LogPath}}' spammer)
   • Check size: ls -lh $LOG_PATH (It will grow by MBs every second).
3. Attempt Prune (Fail):
   • Run docker system prune.
   • Result: Does nothing. The container is running, so logs are “active.”
4. The Architect Fix (Truncate):
   • You cannot delete the file while open, but you can empty it.
   • truncate -s 0 $LOG_PATH
   • Check size again: ls -lh $LOG_PATH (It is now 0 bytes!).
5. The Permanent Fix (Log Rotation):
   • Stop the spammer: docker rm -f spammer
   • Run with limits: docker run -d --log-opt max-size=1m --log-opt max-file=3 alpine ...
   • Result: Docker will now automatically delete old logs when they reach 1MB.