Don't fail jobs when docker images already exist
This MR covers the case where we try to push already existing images.
This could happen if we retry a pipeline that has partially failed.
This MR also includes a few other small changes. There are already a few MRs based on this one, so for simplicity's sake I won't branch off any more.
- Use the correct operator docker image repository for pulling, since the connect namespace is used only for pushing
- Bump the GitLab Runner version used in the operator to v13.10.0. This is usually done during the release and will be so, in the future
- All instances of VERSION in the Makefile assume it's prefixed with
v
- bundle.Dockerfile is generated and updated automatically by the operator-sdk tool
How to test:
GITLAB_RUNNER_OPERATOR_REGISTRY=docker.io/gngeorgiev make VERSION=0.0.5 CERTIFIED=false build-and-push-operator-image
Make sure that this is working. However, DockerHub allows for multiple versions of the same image, so it doesn't return an error.
Another way is to push to GitLab Registry and observe that the jobs won't fail:
Edited by Georgi N. Georgiev