Skip to content

ModelHub

ModelHub is the internal UI + API surface for browsing and running deployed ML models.

modelhub.png

Access

  • Primary entrypoint (via Cortex Nginx): https://cortex.shore.mbari.org/modelhub/

Routing / backend

On cortex.shore.mbari.org, Nginx proxies:

  • /modelhub/ → http://maximilian.shore.mbari.org:8080/modelhub/

This is configured in playbooks/files/nginx/cortex/nginx.conf.

What you should see

  • A web UI under /modelhub/
  • API requests from the UI should stay under the same /modelhub/ prefix

Troubleshooting quick checks

  • Proxy path issues: if the UI loads but API calls 404, confirm the UI is configured for a /modelhub/ base path and that Nginx is proxying /modelhub/ (note the trailing slash behavior).
  • Large requests: Nginx sets client_max_body_size 100m for /modelhub/ (see nginx.conf). If uploads still fail, check upstream limits on the backend service.
  • Backend reachability: from the Cortex host, verify maximilian.shore.mbari.org:8080 is reachable and that the backend service is running.