CLI: Never created the inference.json file --> Hanging

[gleithoernchen]

Hi,
First thanks for the input.
I was trying to run the CLI example, and it is hanging in a while loop and never gets any further.
Inside the monitor_progress(), in the while loop for checking if the inference.json file exists...
Somehow that file is not created... Export of an image ran through.
Any idea why? Is that the right place for that question? How can I check where the script hangs as I can not pipe the output of the main() somewhere
Thanks,
Daniel

[stevejluke2]

Are you running the script from the interactive console in workstation or running it as a script from your scripts folder?

If the latter you will need to add code to consume the standard out and standard error streams. What happens when you don't is these streams fill and python will fill with output. When you run the script in the console Workstation will consume the output but when you use a standalone script you need to to put the output elsewhere.

In the predict_selected python script line 108 reads from the CLI standard out, and line 113 feeds it to the runner's standard out. Instead you should put it someplace else, like a log file, or get a handle to Java's System.out and print it there. You probably will want to do the same for the prediction_progress.stderr which is not handled in this example.

[gleithoernchen]

Thanks for the input. I ran it through console.
I tried again this morning and now it is running trough and the output is visible. The never ending loop that I had was happening in line 149, so after what you had mentioned.
Anyway, I don't have that problem anymore but don't know why...
Thanks for the project. The microservice solution is quite useful...
Daniel

[sluke-nuix]

Ok, thanks for the update, and I am glad it is now working for you.

The never ending loop that I had was happening in line 149, so after what you had mentioned

Just an FYI: The progress monitoring code (monitor_progress(): lines 140-170) and the code used to do logging (and do the external processing) (execute_scoring() and get_process_monitor() lines 102-137) will run in parallel, on two different prcesses. The monitor_progress (where you see the problem) occurs inside Nuix Workstation, while the scoring happens on another, external Python processes, and the process_monitor runs in Nuix Workstation on a different thread. So overall there are 2 processes, and in process 1 (Nuix) there are 2 threads, for a total of 3 parallel(ish) actions.

1. Scoring: External Python process, started on line 132, executing the cli/predict_from_folder.py script
2. Logging (get_process_monitor().monitor()): Nuix Workstation process new thread, started on line 134, copies output from the scoring process to the Nuix process
3. Progress Monitoring (monitor_progress()): Nuix Workstation process, scripting thread, started on line 298, watches the status file for completion

The loop where you saw the problem comes contextually after the logging code (where I think there may be a problem) but it will be run at the same time.