Open
Description
In the BQ import script, I have witnessed a bit of data inconsistency for large collections (of 5 million docs +)
This was running it with multiple threads, and with collection group query enabled.
For a collection of about 6.5 million documents I saw a discrepancy of plus or minus about 1000 documents.
I can't say how consistent this discrepancy was, as I only had a chance to run the script a couple of times on this large collection.