Docker pull suddenly fails with unknown blob error from proxy repository

We have a docker-proxy repository which is proxying docker hub. This repository has been in our configuration for many years without issue. Suddenly a few days ago we started getting errors when pulling an image which had been in our repository for well over a year. This is pulling the latest version, and it was updated a couple of weeks before we started seeing the errors, but not since. I have verified I can pull the image directly from docker hub without issue. I only see the error when I try to pull it through our proxied repository. Actually, we have a docker-group repository which includes the docker-proxy repository, so it’s trying to pull it through there. I was able to pull the latest image and push it to a hosted docker repository also in the docker-group repo, but that means if the latest gets updated in docker hub we won’t automatically get it. I have also found that if I pull other versions of the same image, including one last pushed to docker hub 10 months ago, I get the same unknown blob error. This is the message from the docker pull command “error pulling image configuration: download failed after attempts=1: unknown blob”. And in the request.log on the Nexus server I see the error ““GET /repository/docker-group/v2/shiftleft/core/blobs/sha256:481fc62de1afa28dd96b7c041776f4aea58a854253cf3ba8d7d2f5295ce20936 HTTP/1.1” 404 - 156 4 “docker/20.10.17 go/go1.17.11 git-commit/a89b842 kernel/3.10.0-1160.81.1.el7.x86_64 os/linux arch/amd64 UpstreamClient(Docker-Client/20.10.17 (linux))” [qtp197910014-9986]”.
I have tried deleting the image from the docker-proxy repository, including rebuilding the index and invalidating the cache, compressing the blob, and invalidating the cache for the docker-group repo too, however this error keeps happening.
I also have another instance of Nexus RM installed on a different host for our test environment, and I’m able to pull the image there without issue.
Is there a way I can identify what is missing? I don’t know how to take the URL shown in the request.log and identify where in the filesystem it’s looking for it. Is there a way to do that? Or another way to clear out the old data so I can pull it fresh again?
We’re running OSS 3.43.0-01.
Thanks,
Jack