Docker pulls fail with CloudFront EOF errors, especially for large images

0
0
Asked By MellowCedar42 On

I'm learning Docker on Windows 11 with Docker Desktop using the WSL2 backend, but I'm blocked by image-download failures. Pulling images such as mysql:8 or postgres fails with errors like `failed to copy: httpReadSeeker: failed open: failed to do request: Get "https://production.cloudfront.docker.com/...": EOF`. Small images sometimes work, but larger images usually fail partway through downloading. After reinstalling Docker Desktop, even `hello-world` began failing occasionally.

I've already reinstalled Docker Desktop, tried multiple versions, restarted my PC and Docker, run `wsl --shutdown`, reset Winsock, flushed DNS, changed DNS providers, tested a mobile hotspot, verified WSL2, tested CloudFront with curl from Ubuntu WSL, cleared Docker data, performed a clean reinstall, and used debug logging. I have plenty of disk space, am not using volumes, and the basic command I'm testing is `docker pull mysql:8`. Pulls from another registry worked for a few small images, so I'm wondering whether this is related to large layers, Docker Hub, memory, networking, or the Docker Desktop environment.

4 Answers

Answered By QuasarMink19 On

The fact that small images usually succeed while MySQL and PostgreSQL fail during later layers points more toward an interrupted long-lived transfer than an invalid image. Try comparing the same pull on another completely separate connection and from another registry. If another registry works consistently but Docker Hub fails, rate limiting, filtering, MTU issues, or a proxy may be involved. A curl test from WSL doesn’t necessarily prove that Docker Desktop’s VM and its networking path are healthy.

MellowCedar42 -

A few small images from another registry worked, but the larger Docker Hub images still fail, sometimes after several layers have downloaded.

Answered By MapleRook58 On

A clean reinstall and repeated reboots make a normal installation problem less likely. On Windows, restarting the networking services can sometimes help: `net stop winnat`, `net start winnat`, followed by restarting HNS. Also check whether security software, a VPN, proxy, firewall, or router is inspecting or timing out large HTTPS transfers. Temporarily testing without those layers can help isolate the cause.

Answered By OrbitingPanda7 On

EOF only means the connection was closed unexpectedly; it doesn’t identify the actual cause. Disk space, memory, a damaged Docker VM, or a network device terminating long downloads could all produce it. Check the free space and memory available to Docker Desktop/WSL while the pull is running, even if the Windows drive itself has plenty of room. Since volumes aren’t involved, the Docker storage allocation is still worth checking.

MellowCedar42 -

I checked the available storage and have plenty of space. I’m only running `docker pull mysql:8`, without any volumes.

Answered By VelvetComet31 On

If Docker Desktop continues to fail while native Linux works, running Docker directly inside a Linux VM or WSL installation is a practical workaround for learning. That won’t explain the original EOF, but it can confirm whether the problem is specifically Docker Desktop’s WSL2 networking or storage layer rather than the image itself. Switching operating systems is not required; you can keep Windows and use a Linux environment temporarily.

Related Questions

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.