One of my fears of homelabbing is the reliance on (3rd party)
#Docker/
#container images that could just be gone someday. I've already had it happen once, with
#Bitnami images (fuck
#Broadcom).
One way I previously thought to combat this is to manually pull the image bundle and host it in your own container registry. This works, but obviously not a reasonable effort to do and reproduce for more than ~1-2 images.
Then, the moment I discovered #Amazon/
#AWS has such a thing that addresses this - pull through cache for their
#ECR, I looked up on how I can have the same kind of setup on my
#homelab, and sure enough, there are already several options. I went with
#Zot, and it's working pretty freaking well. Now, anytime I pull any images from registries I've configured Zot to sync like
#GitHub's
ghcr.io,
docker.io,
public.ecr.aws, they'll all be pulled/cached first on my own Zot instance and stored for good there.
Man, I wish I looked into this much earlier - but better late than never.
🔗
https://github.com/project-zot/zot
🔗
https://docs.aws.amazon.com/AmazonECR/latest/userguide/pull-through-cache-creating-rule.html