I am getting a similar message for a workflow, while using Knime 3.7.1 and Windows 10. The file workflow.knime, when opened in Notepad++, is full of NUL characters - which I guess is binary garbage, as noted previously.
The circumstances in which this corruption happened, was that this fairly simple workflow was open (the only one open), in the background, while I worked on some Excel files in the foreground. I think I had saved the Knime workflow, in the previous few minutes, but I could be wrong.
The Knime instance was set to take 8 GB of RAM, on a 16 GB RAM computer, which maybe did not help, regarding what happened next.
Which was that, with the 5 Excel files that I had open, plus 1 or 2 other programmes (MySQL Workbench, and probably a browser), the computer locked up, while trying to carry out an Excel task (saving a large file to our Azure server in the cloud, on a relatively slow connection).
In order to try to recover, I closed down all programmes apart from Excel, to give it more room. This did not help, so I looked in Task Manager, and saw that Knime was still visible as a process taking up memory, even tho I had closed it. So I killed it via Task Manager. Excel did not recover, so eventually (after Task Manager stopped responding) I had to just turn the machine off.
After restarting I opened Knime and the workflow was now showing the message:
java.io.IOException: Unable to parse xml: line=1: Content is not allowed in prolog.
xml: URI=java.io.BufferedInputStream@65435c77
dtd: URI=null
Regarding XML, it might be notable that I had MySQL Workbench open during the crash (and subsequently had lost all the remote connections that I had defined, tho not the localhost one), and that my Excel (that seemed to cause the crash) had an add-on for MySQL, that gave an error message regarding XML after the crash.
So I suspect some kind of general XML corruption might be associated with the crash… but not sure.
The above is just for info, in case useful (I’m resigned to just re-creating the corrupted Knime workflow). I have reduced Knime’s memory footprint to 5 GB for the time being though.