Server error cuOpt

AI & Data Science NVIDIA cuOpt cuOpt
1

Hi,
When I try the following
sudo docker run --network=host -it --gpus all --rm nvcr.io/nvidia/cuopt/cuopt:22.12

I get the following error:

INFO:     Started server process [1059]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:5000 (Press CTRL+C to quit)
INFO:     127.0.0.1:38724 - "GET / HTTP/1.1" 404 Not Found
INFO:     127.0.0.1:33238 - "GET / HTTP/1.1" 404 Not Found
INFO:     127.0.0.1:57024 - "GET / HTTP/1.1" 404 Not Found
INFO:     127.0.0.1:43460 - "GET / HTTP/1.1" 404 Not Found

PS: Even with Python SDK, I have the same kind of issue.

Best,
Naveen

2

What kind of request are you sending it to the server?

And are you running on WSL?

the apis are available in http://0.0.0.0:5000/cuopt/

for docs : http://0.0.0.0:5000/cuopt/redoc or http://0.0.0.0:5000/cuopt/docs

For examples on how to form the request data follow examples in cuOpt-Resources/notebooks/routing/microservice at branch-22.12 · NVIDIA/cuOpt-Resources · GitHub

Make sure the request is goinf as HTTP and not HTTPS.

3

I just started with cuOpt and run the first command mentioned under NVIDIA CUOPT CONTAINER Setup — cuOpt 22.12.00 documentation … So, I haven’t sent any request.
However, Jupyter notebook works fine as I can open it in the browser and run.

4

May be then something trying to ping port 5000 and hitting this server, since request is not proper it is just giving error.

5

Agree with above, it looks to me like the request is just wrong. Here is my output, and the two commands that I sent. As you can see I got the 404 from the second request and the log message matches.

INFO:     Started server process [1060]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:5000 (Press CTRL+C to quit)
INFO:     127.0.0.1:42646 - "GET /cuopt/health HTTP/1.1" 200 OK
INFO:     127.0.0.1:40266 - "GET / HTTP/1.1" 404 Not Found

$ curl 127.0.0.1:5000/cuopt/health
{"status":"RUNNING","version":"22.12"}

# This one does not work because just / is not an endpoint for cuopt
$ curl 127.0.0.1:5000
{"detail":"Not Found"}
6

If you go here on github, there are utility scripts for running cuopt and in the cuopt-api subdirectory there is a sample program for posting a problem to the server.

Update, fixed:
Actually one mistake in post-cuopt.py which I will fix, line 101 should use "set_cost_matrix’ instead of ‘add_cost-matrix’