Docker "error during connect ... docker_engine" (Windows Named Pipe) in CI

On Windows, the Docker CLI talks to the engine over a named pipe (//./pipe/docker_engine). This error means the CLI could not open that pipe - the engine is not running, or the CLI is pointed at the wrong context.

What this error means

Any docker command on a Windows runner fails immediately with error during connect: ... open //./pipe/docker_engine: The system cannot find the file specified. No work happens because the CLI never reaches the daemon.

error during connect: Get "http://%2F%2F.%2Fpipe%2Fdocker_engine/v1.45/version":
open //./pipe/docker_engine: The system cannot find the file specified.

Common causes

The Docker engine/Desktop is not running

On a Windows runner the engine may not be started, or Docker Desktop has not finished initializing, so the named pipe does not exist yet.

The CLI is set to the wrong context or engine mode

A context pointing at a non-existent engine, or Linux-vs-Windows container mode mismatch, leaves the pipe unreachable.

Using a Windows runner where a Linux daemon was assumed

A workflow written for Linux Docker fails on a windows-latest runner where the daemon (or container mode) differs.

How to fix it

Start the engine and confirm the pipe is up

Ensure the daemon is running before any docker command.

# PowerShell on a Windows runner
Start-Service docker
docker version   # should now reach the engine

Point at the right context, or use a Linux runner

1. Run docker context ls and select the context for the running engine.
2. Switch container mode if the workflow targets Linux vs Windows containers.
3. If the job only needs Linux containers, run it on a Linux runner instead.

How to prevent it

• Ensure the Docker engine/service is started as an early step on Windows runners.
• Pin the correct docker context and container mode for the workflow.
• Use Linux runners for Linux-container workloads unless Windows is required.

Related guides