We have a large (80,000+ nodes) site that is currently on 4.7.2 and I attempting to upgrade to 4.11.3. After the upgrade we are getting directed to the no nodes page and republishing the entire site is not fixing it.

After some investigation and finding this issue it appears this is due to some errors in our database where about 300 nodes have two entries in the cmsDocument table that are flagged as published. This results in a Dictionary duplicate key exception be thrown by the umbraco.content.LoadContentFromDatabase() method as the Sql query here has been modified (between 4.10.x and 4.11) with a join to the cmsDocument table to only retrieve published documents in order to build the xml cache. Because of this the xml cache is not being built and the site breaks.

Obviously this is a fault in our data that occured in an earlier version of Umbraco and this needs to be fixed by ourselves, however, I do have a few questions:

Should this error cause such a fatal application error when previous versions have not been affected by this?

If so should I raise this as a new issue or look to get the existing issue re-opened (i'm happy to do necessary work to fix this as a simple clause to check it the key already exists should suffice and maybe a log warning listing any duplicate nodes in this table)?

Also any suggestions as to the best way to fix this in the database would be very welcome, I was thinking of matching the published flag with the newest flag but where duplicates exist?