The tool I think you are trying to use expects read accessions. For a full genome, copy the genome fasta’s link at NCBI into the Upload tool. Standardize the format before attempting to use that fasta with tools. We also strongly recommend getting the annotation at the same time, and confirming all is a “match” before attempting analysis – or expect odd problems.
This forum has much QA about custom genomes and reference annotation. I added some tags to your post that will help to find.
If needed later, these tutorials cover how to get the read accessions based on other project accessions into Galaxy.
Note: Only load up a new reference genome if the server where you are working does not have it indexed already, and be very sure NOT to reuse a server-assigned “database” to a custom genome. Create a unique name to avoid conflicts.