Apps

Docker Desktop

Docker Desktop stores several layers of local app data beyond your images and volumes: Electron HTTP cache, V8 Code Cache bytecode, GPUCache shader and rendering blobs, and accumulated host and Linux VM log files used by its dashboard and background services. These files help the UI start faster and record engine activity, but they can become stale after Docker Desktop, Chromium, or graphics driver updates, or simply grow large over time. Kudu removes those rebuildable caches and older settled host and VM logs while leaving containers, images, volumes, credentials, settings, and other user data untouched.

Why it matters

Why clean Docker Desktop?

  • Invalidated Electron V8 Code Cache after a Docker Desktop update can make the dashboard open slowly or feel sluggish until bytecode is rebuilt
  • Stale GPUCache entries from a graphics driver update can cause blank panes, black windows, or flickering in the Docker Desktop interface
  • Corrupted Chromium disk cache can leave extension pages, image assets, or embedded web views loading incorrectly, with missing icons or endless spinners in the dashboard
  • Host and Linux VM logs can accumulate across engine restarts and failed launches, consuming hundreds of megabytes or more and making disk usage jump unexpectedly
  • Repeated backend errors written to host or vm logs can bury the current failure signal, so trimming older settled logs makes recent startup and networking issues easier to inspect
  • Old cached frontend assets can mismatch a newly installed Docker Desktop build, leading to odd UI behavior such as buttons not responding until the cache is regenerated
What gets cleaned

Cache paths Kudu targets

Windows

%AppData%/Docker Desktop/cache
%AppData%/Docker Desktop/Code Cache
%AppData%/Docker Desktop/GPUCache
%LocalAppData%/Docker/log/host
%LocalAppData%/Docker/log/vm

macOS

~/Library/Caches/com.docker.docker
Frequently asked

Common questions about Docker Desktop

Free & open source

Download Kudu and reclaim your disk space.

Available on Windows, macOS, and Linux. No account required, no feature gates, no telemetry without consent. All cleaning targets are open source and community-auditable.