It has to do with the way new data files are “sniffed” to detect the datatype.
It is seeing columns with content that resembles chromosome names and coordinates. I’m investigating what may have changed and if/why the servers are acting differently. We have some tutorials with warnings but I thought this was addressed. We had some recent updates that may be impacting this.
For now, you can assign the datatype directly. Or, if you are moving entire histories, the files will restore on the other server however they were labeled on the origin server. This is one reason why I prefer that method – the original file name, datatype, collection structure – all is preserved.
More soon. Thanks for alerting us about the oddness. 