Repository navigation
Failed in Random Forest Classification for Sentinel 2 Images #3269
Replies: 1 comment 2 replies
|
The installed Windows build doesn't save sidecar logs to a file, which is why you can't find them anywhere. Here's where they go and how to capture them. Why you can't find them When a Whitebox job fails, the sidecar logs the full traceback to its stderr ( The Diagnostics dialog doesn't help either. It records only the frontend's network requests and browser errors, and the Python traceback never reaches it. (It also shows "No error diagnostics captured" partly because the 0 errors filter is selected; your 500 network requests are logged, but they only contain the same generic message.) How to capture the log Start GeoLibre from Command Prompt (cmd, not PowerShell) with its output redirected to a file:
The redirect works for a GUI app because cmd passes the file handles to the program it starts, so whatever it writes lands in the file. Likely causes to check meanwhile The four
If you post the traceback from the log, we can find the exact cause. |


The error at the end is a timeout, not a crash in the classification:
GeoLibre gives each Whitebox tool run at most 1 hour by default. When that time runs out, it stops the process and reports the generic "Tool execution failed" message. Your four
progresslines mean the classification was still working. It just needed more than an hour.Option 1: raise the time limit. The limit can be set with
GEOLIBRE_WHITEBOX_RUN_TIMEOUT_SECS, in seconds (whitebox.py). The desktop app passes your environment variables on to the sidecar. In the same cmd window as before: