Skip to content

BUG_EXERCISES.md

Latest commit

 

History

History
134 lines (101 loc) · 6.18 KB

BUG_EXERCISES.md

File metadata and controls

134 lines (101 loc) · 6.18 KB

Bug-Fix Exercises (Debugging Practice)

This document lists real bug scenarios in this repository as hands-on exercises.

How to use this exercise set

  • Reproduce the issue first.
  • Write a failing test (or script) that proves the bug.
  • Fix with the smallest safe change.
  • Re-run go test ./... and go build ./....

Exercise 1 — Runtime panic from short container IDs

  • Category: Runtime panic / edge case
  • Area: Monitoring
  • Location: /home/runner/work/basic-docker-engine/basic-docker-engine/monitor.go (ContainerMonitor.GetMetrics)
  • Symptom: Program can panic with slice bounds out of range when container ID length is less than 8.
  • Repro hint: Run monitoring code against a short container ID (for example "abc").
  • What to debug: String slicing assumptions.
  • Done when: Monitoring no longer panics for short IDs and returns metrics safely.

Exercise 2 — Nil pointer risk from deferred close before error check

  • Category: File I/O correctness
  • Area: Layer metadata persistence
  • Location: /home/runner/work/basic-docker-engine/basic-docker-engine/main.go (saveLayerMetadata)
  • Symptom: defer file.Close() is scheduled before confirming os.Create(...) succeeded.
  • Repro hint: Force create failure (permissions/path issue) and observe behavior.
  • What to debug: Ordering of error handling and cleanup.
  • Done when: File handles are only closed when file creation succeeds.

Exercise 3 — Wrong PID tracked for a running container

  • Category: State inconsistency / observability bug
  • Area: Container lifecycle
  • Location: /home/runner/work/basic-docker-engine/basic-docker-engine/main.go (runWithoutNamespaces)
  • Symptom: Stored PID reflects the CLI process instead of the spawned container command process.
  • Repro hint: Run a container and compare metadata PID with actual child process PID.
  • What to debug: Process lifecycle timing (cmd.Start/cmd.Run) and which PID should be recorded.
  • Done when: Metadata PID maps to the actual container command process.

Exercise 4 — ps/exec status flow relies on PID file that is never written

  • Category: Feature integration bug
  • Area: CLI status and exec path
  • Location:
    • Reader: /home/runner/work/basic-docker-engine/basic-docker-engine/main.go (getContainerStatus, execCommand)
    • Writer path gap: /home/runner/work/basic-docker-engine/basic-docker-engine/main.go (runWithoutNamespaces)
  • Symptom: Status/exec behavior can be incorrect because PID file expectations are not met.
  • Repro hint: Run a container, inspect files under /tmp/basic-docker/containers/<id>/, then try ps/exec.
  • What to debug: Contract mismatch between producer and consumer of runtime state.
  • Done when: PID file lifecycle is consistent and dependent commands behave correctly.

Exercise 5 — Deferred close in loop can leak descriptors under many layers

  • Category: Resource leak / scalability bug
  • Area: Image pulling
  • Location: /home/runner/work/basic-docker-engine/basic-docker-engine/image.go (Pull)
  • Symptom: Layer readers are closed with defer inside loop, so descriptors stay open until function returns.
  • Repro hint: Pull image with many layers and watch file descriptor count/pressure.
  • What to debug: Resource lifetime in loops.
  • Done when: Each layer reader is closed promptly after extraction.

Exercise 6 — Hardcoded image directory bypasses shared runtime configuration

  • Category: Configuration consistency bug
  • Area: Image management
  • Location:
    • /home/runner/work/basic-docker-engine/basic-docker-engine/image.go (ListImages, Pull, LoadImageFromTar)
    • /home/runner/work/basic-docker-engine/basic-docker-engine/main.go (listImages)
  • Symptom: Multiple code paths hardcode /tmp/basic-docker/images instead of consistently using shared directory variables.
  • Repro hint: Change runtime base directory and observe path inconsistencies.
  • What to debug: Single source of truth for storage paths.
  • Done when: Image paths are consistently derived from shared config.

Exercise 7 — Potential path traversal / unsafe extraction behavior

  • Category: Security bug
  • Area: Image extraction
  • Location: /home/runner/work/basic-docker-engine/basic-docker-engine/image.go (extractLayer, LoadImageFromTar)
  • Symptom: Tar extraction trusts archive entries and may write outside intended rootfs depending on archive content/tool behavior.
  • Repro hint: Create a crafted tar with traversal-like paths and evaluate extraction safety.
  • What to debug: Archive extraction safety boundaries.
  • Done when: Extraction guarantees writes stay within target rootfs.

Exercise 8 — Error handling is partially ignored in state transitions

  • Category: Reliability / silent failure
  • Area: Container state persistence
  • Location: /home/runner/work/basic-docker-engine/basic-docker-engine/main.go (runWithoutNamespaces)
  • Symptom: Calls to UpdateContainerState(...) are not checked for errors.
  • Repro hint: Make state file path unwritable and run container lifecycle.
  • What to debug: Error propagation in critical state updates.
  • Done when: Failures to persist state are surfaced and handled predictably.

Exercise 9 — Network state management uses global mutable state

  • Category: Concurrency and maintainability risk
  • Area: Network management
  • Location: /home/runner/work/basic-docker-engine/basic-docker-engine/network.go
  • Symptom: Global networks slice is read/written without synchronization.
  • Repro hint: Simulate concurrent network operations (or parallel tests) and look for races.
  • What to debug: Shared state ownership and synchronization strategy.
  • Done when: Concurrent operations are safe and deterministic.

Suggested debugging workflow (work-like)

  1. Reproduce and capture exact failure output.
  2. Add a failing test that isolates one bug.
  3. Fix one bug at a time (avoid broad refactors).
  4. Re-run:
    • go test ./...
    • go build ./...
  5. Add a short note in your commit message: root cause + prevention idea.