Closed yeldarby closed 1 year ago
I pushed an update to our backend API to fix this issue & verified the download script works again.
The underlying problem was that we added error checking to our API to ensure that all images in a version had finished being generated before allowing the dataset to be exported and archived into a zip file.
This error checking was too strict and didn't handle the edge case where there were some corrupted images that failed to generate properly, which made our pip package wait indefinitely for the progress to reach 100% (which would never happen due to the corrupted images).
I added a timeout to prevent failed images from preventing exporting the rest of the dataset.
Propagating this issue from Hacker News:
I think this likely is related to this PR merged a few hours ago. One of the datasets looks to have progress marked as greater than 100% which is causing the script to think it’s not completed yet & wait indefinitely.