I have a pipeline that loads multiple csv's and an xlsx file into 4 separate tables. Generally, this pipeline runs fine. However, occasionally one of the tables (always the same table) results in an invalid object. If I re-run the pipeline with no changes, all is fine. I have never had an issue with any of the other tables loaded by the pipeline. The table fields are all nvarchar(max) and the data is relatively simple. It is also a very small table. Well under 100 rows, and the data in the source rarely changes. It is definitely not changing between successful and unsuccessful runs.
I am interested in any ideas on what may be causing this periodic failure. All runs fine more than 95% of the time. When it fails, it kills the rest of the load process.