Non-negative number required error while appending tables

1. One of the ways we found around this bug (it's not an issue) was by changing the order of the Appended Queries, for eg:

If you're appending Table 2, 3 and 4 to Table 1 (one being the main) - change the order to append Table 1, 3 and 4 to Table 2. 

That solved our issue in one of our reports - we are appending more than 100m rows between 4 different datasets. This is not consistent - sometimes it still gives the error but eventually, we were able to make it through. 

2. If the error happens around the same row count - for eg: the error always happens around 19m rows - and you know that on 19m rows your dataset is changing from Table 1 to Table 2, then there might be something wrong a part of your Table 2. This is how we bypassed this bug on another case. In theory, nothing should be wrong with the dataset - we are 100% sure of that - but somehow, somewhere, Power BI is interpreting the data the wrong way. Once we deleted the first millions on Table 2 the query ran as expected.

This shouldn't be happening and it appears to be happening a lot - only on large datasets apparently. 

3. On another thread this came up:

"Realizing this problem is bigger than me, I decided to try and get around it by creating the final form of the data table on the SQL Server side, thus removing the heavy workload (Which was mostly merged operations) from the mashup engine. This resulted in a successful report refresh."

We tried this but PowerBI couldn't handle the 200m rows dataset.

Hope this helps.