You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Remove "ghost" papers from MongoDB where GCS file is deleted #545
We currently have several "ghost" records in our MongoDB database. These are paper records where the underlying PDF file was deleted (or failed to upload) to our Google Cloud Storage bucket, but the database record was never removed.
Because the file_url points to a missing object, Google Cloud returns a 404 NoSuchKey XML error when users try to view or download them.
So users get infinte loading screen and download button not working
The userfacing part is handled on staging since #544
the viewer checks the file first and shows "Paper Not Found" instead of loading forever, and downloads show "This paper is no longer available"
whats left is deleting the ghost records themselves which needs access to the production database and bucket. If it helps, I can write a script that lists papers whose file is missing (dry run by default) for a maintainer to run
Activity
Dx-Alz-xD commentedon Nov 11, 2026
The userfacing part is handled on staging since #544
the viewer checks the file first and shows "Paper Not Found" instead of loading forever, and downloads show "This paper is no longer available"
whats left is deleting the ghost records themselves which needs access to the production database and bucket. If it helps, I can write a script that lists papers whose file is missing (dry run by default) for a maintainer to run
Rithish-2914 commentedon Nov 12, 2026
The user-facing side is handled the viewer shows "Paper Not Found" and downloads block gracefully instead of hanging.
What's left: a maintenance script to audit and hard-delete MongoDB records whose GCS file no longer exists. you can surely work on this