Docker push to container registry never completes on a new runner

After switching to a new dedicated server for my GitLab runner, pushing to containers never completes and just has an endless stream of messages like:
bbdcc09b9436: Waiting
856be9b83b46: Waiting
454b1f880d9e: Waiting
fd674058ff8f: Waiting
bbdcc09b9436: Waiting
61ccb076d3e2: Waiting
4416e5d0ee0a: Waiting
22a0038f868f: Waiting

What’s weird is this only happens on two of my projects, my third project works fine on this new server. My build script is exactly the same and worked fine on my old runner, using the CI variables for docker login
docker login $Env:CI_REGISTRY -u $Env:CI_REGISTRY_USER -p $Env:CI_REGISTRY_PASSWORD

I’ve also tried switching the login to using a deploy token with read/write registry access to see if it was an issue with the automatic registry login, but that didn’t work either.

Any ideas how to fix this?

I tried adding a docker logout $Env:CI_REGISTRY before calling docker login to see if it was a cache issue between different projects, but that didn’t work either. I’m at a loss on how to fix this.

Sounds like you need to look at the logs on the server.

Command line on the server of gitlab-ctl tail registry will monitor what the registry process is doing and you might find an error message in there to help narrow this down.

I’m on gitlab.com, so I don’t believe I would have access to those logs?