Thanks. There’s a technique called “bisection” that I once learned for narrowing down the source of problems that works like this:
- Split your data in half, test which half gives the error.
- then split that half in half, test which half gives the error.
- repeat until you narrow it down to one guilty row
In your case, that’s 14 iterations:
11000
5500
2750
1375
688
344
172
86
43
21
11
5
3
1
It’s boring to do, but it’s not endless, or hopeless