I have a python script that built database relates in ArcMap in a filegeodatabase. They are many-to-many relates between three tables using a relationship table in the middle. This is the design from the source in PostGIS that I want to replicate. The keys between the tables are indexed integer fields. This all worked well in ArcMap and building the relates took a few minutes.
Now in ArcGISPro the same script takes many hours! What has happened?
[As a side-issue, BigInteger keys cause a crash, they must be 32 bit Integers]
The tables are medium sized, typically 2M records. Note that you cannot rename any tables afterwards, you have to rebuild from scratch. All tables must be in the same filegeodatabase. The relation table built is a semi-hidden table, hard to delete in a script.
Has anyone else noticed this? Do you have any suggestions to restore the speed? Maybe remove the indexes before building the relates and re-creating afterwards.
