Closed markoa closed 10 months ago
Hi @markoa I found this issue when investigating why I'm getting ff3a5c916c92: Retrying in 20 seconds
messages when pulling images in CI jobs. I found out about the undocumented internal semaphore mirror by digging into the config...
Some questions:
Hi @schliflo, Please let me answer Your questions:
These "retrying" messages can manifest when a docker registry proxy doesn't have the image requested and is working on fetching it before serving it or the docker registry for some other reason cannot serve the request at the moment.
If you find that "retrying" message on several occasions please submit a support request.
I ran into the Retrying
message multiple times before. We use custom images (which are pre-built & published to DockerHub) in our CI jobs.
The issue seems to be solved by using the google proxy:
sudo sed -i 's@https://dockerhub.semaphoreci.com/@https://mirror.gcr.io@g' /etc/docker/daemon.json
sudo restart docker
I still think that the default configuration using the semaphore proxy should be documented.
Closing as stale.
We recently introduced a Docker Hub mirror which runs inside the same DC as all Semaphore jobs. This should be prominently visible on the "Working with Docker" page.
For comparison, Google Cloud's docs about their DH mirror: link.