You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Docker Desktop 4.95.0 on Windows — WSL2 virtualization/backend intermittently loses Engine API connectivity #755
Docker Desktop 4.95.0 on Windows intermittently loses access to the Linux Docker Engine API. The Desktop UI can report that Docker is running while CLI requests fail with HTTP 500 or the Docker Engine named pipe is missing.
Steps to reproduce
Start or reopen Docker Desktop after Windows starts.
Wait until Docker Desktop reports that it is running.
Run docker info and docker ps -a.
Inspect existing containers or retrieve their logs.
Repeat the checks after the Engine API temporarily becomes available.
Expected behavior
The Docker Engine API should remain consistently accessible. Commands such as docker info, docker ps, docker inspect, and docker logs should work reliably.
Actual behavior
The issue occurs intermittently:
Docker API requests return HTTP 500 for /v1.56/info, /v1.56/containers/json, and /v1.56/containers/<container>/json.
At other times, the CLI reports that npipe:////./pipe/dockerDesktopLinuxEngine cannot be found.
Docker Desktop backend logs repeatedly report connect tcp 192.168.65.7:2376: no route to host.
Diagnostics report that dockerVirtualization is missing and that WSL commands such as uname, dmesg, free, and ps time out with context deadline exceeded.
Requests to the internal diagnostic service also time out.
A previous diagnostic snapshot recorded an OOM kill of beam.smp and I/O errors on loop1. It is unclear whether these are related.
Restarting Docker Desktop or shutting down WSL has sometimes restored Engine API access temporarily, but the local Supabase development stack remains partially unhealthy.
Activity