[ic4l]

This broke our builds since we rely on several public Docker images, and by default, Docker uses docker.io.

Thankfully, AWS provides a docker.io mirror for those who can't wait:

  FROM public.ecr.aws/docker/library/{image_name}

In the error logs, the issue was mostly related to the authentication endpoint:

▪ https://auth.docker.io → "No server is available to handle this request"

After switching to the AWS mirror, everything built successfully without any issues.

[CamouflagedKiwi]

Mild irony that Docker is down because of the AWS outage, but the AWS mirror repos are still running...

[kerblang]

Also, docker.io is rate-limited, so if your organization experiences enough growth you will start seeing build failures on a regular basis.

Also, quay.io - another image hoster, from red hat - has been read-only all day today.

If you're going to have docker/container image dependencies it's best to establish a solid hosting solution instead of riding whatever bus shows up

[pploug]

Rate limits are primarily applied to unauthenticated users, open source projects and business accounts have none/much higher tresholds

[suriya-ganesh]

based on the solution, it seems like it is quite straightforward to switchover

[firloop]

I wasn't able to get this working, but I was able to use Google's mirror[0] just fine.

Just had to change

    FROM {image_name}

to

    FROM mirror.gcr.io/{image_name} 

Hope this helps!

[0]: https://cloud.google.com/artifact-registry/docs/pull-cached-...

[ic4l]

We tried this initially

  FROM mirror.gcr.io/{image_name}

We received

  failed to resolve source metadata for mirror.gcr.io/

So it looks like these services may not be true mirrors, and just functioning as a library proxy with a cache.

If you're image is not cached on one of these then you may be SOL.

[da768]

During the last Docker Hub outage we found Google mirrors lost all image tags after a while. Image digest references would probably work

[geostyx]

public.ecr.aws was failing for me earlier with 5XX errors due to the AWS outage: https://news.ycombinator.com/item?id=45640754

[anon7000]

I manage a large build system and pulling from ECR has been flaking all day