I have a workflow that takes a multi-page document generated by Snapshot, uses Pattern Matching (RegEx) to extract a unique identifier for each page, extracts the page, looks up the identifier in a separate SQL database to get more information about the page, then creates a new document, moves that page into the document and uses the SQL data to insert metadata, name the document, and determine the path to put it into our Laserfiche repository. There is a starting rule that watches a folder, then triggers the workflow when a document is created there.

If we create a document in the watch folder, the workflow is triggered; then while that instance of workflow is still running, a second document is created in the watch folder. The result is that pages from the second document will get mixed in with the first document. Where each page from the source should be a single page in the resulting documents, some of the resulting documents have 2 pages (one page from each of the source documents), so we're losing pages from the second document because they're mixed in with the first, and don't have the right metadata attached and are in the wrong folder.

Is this a known bug or is there a setting I missed somewhere?