For this kind of issue, sometimes recreating the extract solves it.
However if it does not, I suggest opening a case with our support team as they will most likely need to go through the logs.
Is the goal to load an empty extract and then have the server 'build / load' the extract during a scheduled refresh? It could be that the 'bin' calculation doesn't have any data to build itself. I haven't used bin's in our extract yet, but you could try a different empty extract method and see if it'll bypass the error - if not, I agree with Ciaran that you might need to open a ticket.
This is how we build all our empty extracts. On our dev environment, we build the data source based on a test company with a tiny amount of records.
Here's how we do it:
Create a parameter (like below)
Create a calculated field on that parameter
Right click on the data source > edit data source filters > add > select the calculated field (BEE_FIN in my case)
Select the 'True' value and check the 'Exclude' box in the bottom right.
Click ok to apply it. Now your data source will load an empty extract when the parameter is set to 'YES' or 'True'.
Right click on the data source > extract data. This should build an empty extract.
Important: Change the parameter to 'No' or 'False' before publishing
After generating the empty extract, change the parameter to 'No' and then publish the data source. Refresh the data source on the server. The server will then do all the heavy lifting of loading the extract.