It would be useful if you could somehow narrow it down to a specific file, the error message itself seems rather mundane.

One thing you can try is use a WorkspaceRunner from a second workspace to run the one you're having trouble with on a per-file basis, as the WorkspaceRunner has Succeeded and Failed outputs it should help you narrow down where the issue is.

Yeah workspace runner per file is a good idea to help track down specific data issues.

An error like that with no error messages does seem to indicate a crash. It could also be memory related if you really have a lot of features there but usually FME is pretty good at trying to deal with that. If you're running FME just in desktop then check that you're not running with feature caching in case it's memory related.

Another option to try and debug which is a lot less fun and can be pretty painful is to turn on feature caching and use partial runs to run parts of the workspace progressively - for crashes I would look focus on the following areas.

• Check the readers can work on all data (run just the reader(s))
• If you have any geometric transformers (especially ones which have a tolerance in them)
• Writers

Thanks guys, I've just ran it on the file folder where it lat crashed and it crashed again on the same file.

The workbench is essentially a file validation one, so if files throw it you would assume something is off in the file - is there a way to flag it and skip the file so the workbench keeps running?

For context, I'm receiving lots of files from various parties, and the idea is to send back a log when things are off. I suppose I could hide that file and re-run the workbench and manualy mark the file but that's not ideal.

Also there are lots of files so the workbench will take around an hour.. so continuity is important.

Are you able to pin point where in the workspace it's crashing? if so you might be able to change the workspace so that it no longer crashes and perhaps catch the issue.

If you expect to run the process on multiple files then it might make sense to run each file one-by-one through the workspace with a WorksapceRunner. This will let you know which files processed successfully and which failed.

The parent process with the WorkspaceRunner inside of it (calling your workspace) wont crash itself so you can keep processing the next file.

Not sure if this batch processing would work for your data but it might be worth considering

I fixed it by tracing to the last file, i removed from the dataset and it ran OK.

Anything special about that last file that you can see?