Troubleshooting an fme.exe crash

How to Troubleshoot an fme.exe Crash

You may come across the following error when running your translation:

When this happens the log file may not necessarily indicate what the exact problem is. This is because FME processes feature by feature (unless there is a group-based transformer in the workspace), but the log reports transformer by transformer. A reader, writer, or transformer not handling a feature successfully is a likely cause of this error. To narrow down where in the workspace and what feature specifically is the cause of the error we recommend following these steps as a general rule-of-thumb to troubleshoot:
 

Step 1: Check if the problem is with the reader(s)

Open the source datasets in the FME Inspector application. If the FME Inspector application throws an error please submit a support request at Safe Support . If all source datasets are read successfully then the problem is not with the reader (go to step 2).

Step 2: Check if the problem is with the writer(s)

Check Writers|Redirect to Inspector Application and run the workspace (this turns off the writers). If the crash does not occur the problem is with a writer, so please submit a support request at Safe Support . If the fme.exe crash still occurs the problem is in a transformer (go to step 3).
 

Step 3: Look at the last message in the log file

This may not point you to exactly what the problem is, but it will likely give you a clue as to where in your workspace to start looking. For example, take the lines below:
 

Example 1:

INFORM|Reading table/feature class 'Parcels'

- this could indicate the problem is somewhere in the Parcels pipeline. Enable just this source feature type and see if you still get a crash. If so disable groups of transformers to narrow down which section causes the error, then disable transformers individually until you find the problem transformer.
 

Example 2:

INFORM|Sorter2(SortFactory): Done sorting 28131 feature(s)

- this indicates the problem is somewhere after Sorter2. Confirm this by disabling the connection line from Sorter2 to the next transformer. If it no longer crashes disable groups of transformers, or one transformer at a time until the crash occurs again and you've found the problem transformer.
 

Step 4: Create a reproduction package for Safe support

A reproduction package, at a minimum, consists of:

- source data
- workspace
- log file

Ideally you would send us the smallest amount of data in a small workspace as a reproduction.

To create a small workspace:

If you are troubleshooting up to this point in a large workspace, or if your source data is in a database, then write the data to an FFS file by placing a Recorder transformer right before the problem transformer. Be sure to disable the connection to the problem transformer. Run the workspace, which results in an FFS file being written to disk. Create a new workspace which contains an FFS reader to read this file and the problem transformer only.

To send a smaller dataset:

In the small workspace use the "Max Features to Read" and "Start Feature" parameters on the FFS reader to narrow down approximately what feature is causing the problem. Place another Recorder in this workspace (disabling the problem transformer) and write a smaller number of features to a new FFS file and send this to us rather than the bigger FFS file.