There are a lot of duplicate CoverageRecords from the time when the OA content server offered per-collection coverage. There was also a brief period when IntegrationClient's had data_sources that were used to offer coverage / licensepools, and it's all still floating around in the database.

It may be more trouble than its worth right now, but it would be nice to remove the IntegrationClient datasources entirely (and connect their dependents to Collection-based datasources or whatever is in-line with the current state of the code instead). My fear is that inconsistencies will create painful or confusing debugging later on.

At first glance, it appears CoverageRecord and LicensePool have had the most problems with this, though some types of metadata (Hyperlink, for example) may also be affected.

┆Issue is synchronized with a GitHub issue
┆Repository Name: metadata_wrangler
┆Issue Number: 151