API concurrency issue

Python
1

"I have a FastAPI backend deployed on Fly.io that handles PDF processing for my React frontend. When a user selects a component, two things should happen independently:

  1. Display a PDF file from a direct URL in an iframe
  2. Process this PDF on the server for future Q&A functionality

The issue: When deployed on Fly.io, the PDF doesn’t display until the server finishes processing it, even though these should be independent operations. The PDF viewer is using a direct URL that shouldn’t depend on server processing.

Interestingly, when running the same code locally, everything works as expected - the PDF displays immediately while processing happens in the background.

My fly.toml configuration:

toml

Copy

[http_service]
  internal_port = 8000
  force_https = true
  auto_stop_machines = 'stop'
  auto_start_machines = true
  min_machines_running = 0
  processes = ['app']

[[vm]]
  memory = '1gb'
  cpu_kind = 'shared'
  cpus = 1

This seems like a concurrency issue where Fly.io might be queuing all requests while processing the PDF, even though the PDF display should be independent. Has anyone encountered similar issues with long-running processes blocking other requests on Fly.io?"

2

Try adding more CPUs? Is your web server also responsible for the PDF processing? You should create another app to handle those tasks async otherwise you’ll block your web server.

3

Makes sense. Thanks a lot!