# development


09/27/2021, 6:35 AM
Topic: Docker backend At work we mirror quite a number of images from docker hub to private docker registry. Sometimes with some customisations, often as-is. I’m considering whether to have a dedicated target for the purpose of the action of mirroring images between registries, but think it may be possible (and even better) to simply beef up the
target to handle this use case easily. (cont. in thread)
The main use case is to allow deploying applications in an air-gapped environment. The mirroring is done in three steps. 1. Docker pull (or Docker build in case of customisations) 2. Docker tag (in case of a pull, as build can tag it in step 1) 3. Docker push When running Docker pull, the image name may not reflect the desired image name to push 1 to 1. There are two main things I want to make this flexible. 1. docker push (publish) and 2. inline dockerfile commands For the plain mirroring, the Dockerfile would simply by a single
line. It would be much clearer to be able to have the information for the base image, directly in the
The biggest win for re-using
for mirroring, is when you want customisations, it becomes a regular Docker build, as for plain mirroring it simply is a no-op build with re-tagging and push.
Hmm.. realize this should probably be in a issue instead, that I can refer to from a PR… 😉


09/27/2021, 1:04 PM
i kinda did something like this a long time ago, but i used the repo names as CLI args to a
goal I wrote:
👍 1


09/27/2021, 1:20 PM
Cool. Will look at for inspiration.