Web UI not showing Coloured (Scores) Overlay for bushfire.tif

[greg-aiml]

Summary

Coloured overlay not shown after processing has completed. System Log on web-ui has some errors but they're not very informative.

Steps to Reproduce

1. Deploy locally using Docker (WSL2 on Win 11 (Chrome web browser) )
2. Upload bushfire.tif test image
3. Wait 1hr for processing to complete

Issue(s)

1. Processing is very slow and no way of telling when it will complete.
2. After processing completes, the log shows some errors and the UI doesn't display the overlay

System Log:

Web UI not showing coloured overlay:

[anjithask]

The error appears to occur when attempting to view the source file from the report while the image is still processing. Under other circumstances, the functionality works without issues. The files stored in the Google bucket are not corrupted, and both the source and results can be properly viewed on QGIS. The error seems to be caused by GDAL, though the exact cause is not clear, it is being investigated.

[greg-aiml]

Re-starting the local-run in Docker resulted in the processed image being correctly displayed:

[greg-aiml]

Deleting the result using the web UI results in the following errors:

[greg-aiml]

I'm unable to retry the processing because I cannot upload any other images (the WebUI show the Delete dialog whenever I try to upload a new image).

[greg-aiml]

How do I clean up the local test system so that I can upload another test image and re-attempt the processing?

[anjithask]

Deleting the result using the web UI results in the following errors:

The file not found error indicates that the file record was not present in the database. The Docker container restart likely caused the issue because it reinitializes the environment, potentially losing the file records from the database.

[anjithask]

How do I clean up the local test system so that I can upload another test image and re-attempt the processing?

From the above image, to solve the error, you can refresh the webpage to upload files.

To clean and restart fully, remove any cache and files from mounted volume. Based on the docker-compose.yml these include database data, model and NATS data.

Use the following command to prune and delete all associated data with local deployment.
sudo docker system prune -a; sudo rm -r /tmp/dra/; sudo rm -r ~/.cache/dra; sudo rm -r db/data/; sudo rm -r queue/data

[greg-aiml]

I've given it a few runs and all seems to be working (the overlay is displayed on the test image once the processing completes).

The log of the most recent successful run (for reference):

Steps to Reproduce

1. Clone release v1.0.0 from Git
2. Follow readme to set up for local testing
3. In ~/gards run docker compose up
4. Start a new new job by uploading the bushfire.tif test image
5. Wait for the system to download runtime requisites (model checkpoints etc)
6. Repeat tests of similar images are then quite fast (less than 1min)

[greg-aiml]

Actually. That small test image takes about 10s