Essentially if there's a lot of data coming back for a particular query or there are a lot of queries, then the time to process can exceed the polling frequency. In this case the file can be processed more than once. Once the first process is completed, the file will be deleted. The impact on the end result is that the acknowledgement file will record multiples of the same file.
There are details here of someone else who faced a similar issue.
http://mule.1045714.n5.nabble.com/File-transport-read-large-files-multiple-times-td2680566.html
|