It fails. We have had issues where the release process is successful, deployed to nexus, but fails with changes in the release or hotfix conflicting with changes in develop. We suspect our issues were related more to space/tab issues and reformatted code, but the potential exists.
When it fails, you can fix the conflict and continue and it works, but that is NOT an automated step. I am more likely to perform releases manually to be able to unwind issues if there is a problem. Or I can perform it locally in "dry run" mode (no Nexus or Git push options) and if it works we can run the job. But then what is the point of the job? I have also finished releases from a console on the Jenkins slave where the job failed.