First of all, thanks a lot for "kube-image-keeper". It is a great project and solves many problems for me.
I would like to request for a feature, which could help the project.
When images need cached to the "kube-image-keeper's registry, the images are pulled from remote image registry.
Would it be possible to query the local cache on the nodes first to check if they already have the image ? If they do, then the image could be pulled from there instead of the remote registry.
We have a use-case where we only save 10 tags of a repository on our remote registry. However, one or more of our kubernetes clusters could be running a deployment using an image tag that is no longer present in the remote registry.
In such a case, the caching fails since it is not present on the remote registry:
$ kubectl get cachedimages
NAME STATUS CACHED RETAIN EXPIRES AT PODS COUNT AGE
<repository>-<tag> ErrImagePull 6 95m
Since the image with the needed tag is already cached on at least one of the nodes in the cluster, it would make sense to pull it from there.
This could also help reduce the traffic with the remote registry for all the image pulls.
Could you please review and let me know your thoughts on the same ?
Hi Team,
First of all, thanks a lot for "kube-image-keeper". It is a great project and solves many problems for me.
I would like to request for a feature, which could help the project.
When images need cached to the "kube-image-keeper's registry, the images are pulled from remote image registry.
Would it be possible to query the local cache on the nodes first to check if they already have the image ? If they do, then the image could be pulled from there instead of the remote registry.
We have a use-case where we only save 10 tags of a repository on our remote registry. However, one or more of our kubernetes clusters could be running a deployment using an image tag that is no longer present in the remote registry.
In such a case, the caching fails since it is not present on the remote registry:
Since the image with the needed tag is already cached on at least one of the nodes in the cluster, it would make sense to pull it from there.
This could also help reduce the traffic with the remote registry for all the image pulls.
Could you please review and let me know your thoughts on the same ?
Thanks and regards, Sivarama Raju P