outside of localhost can not access API

[brainnnnup]

📚 Documentation

When I use the example image_classifier。The API can be accessed locally，other devices can not access API。

Non local：{ "code": 404, "type": "ModelNotFoundException", "message": "Model not found: densenet161" }

local：{ "tabby": 0.4666191339492798, "tiger_cat": 0.46449020504951477, "Egyptian_cat": 0.06614047288894653, "lynx": 0.0012924382463097572, "plastic_bag": 0.00022909714607521892 }

[PushpakBhoge512]

I think you swapped the above cases, are you running the server at your local personal laptop or desktop?
in that case it won't work for other devices If you want it there are some router configuration you can do and them expose the ports but I think even then it will be only accessible to the devices connected to router

[msaroufim]

It's hard to debug without repro instructions do you mind filling up the issue template so I can help?

[brainnnnup]

I think you swapped the above cases, are you running the server at your local personal laptop or desktop?
in that case it won't work for other devices If you want it there are some router configuration you can do and them expose the ports but I think even then it will be only accessible to the devices connected to router

Sorry, I didn't reply in time.I don't quite understand setting the routing configuration.It seems that there is no such step when configuring torch serve. Can you explain it in detail. Thank you. @PushpakBhoge512

[brainnnnup]

It's hard to debug without repro instructions do you mind filling up the issue template so I can help?
Sorry, I didn't reply in time. There are no repro instructions. Another host in the LAN can't access the inference API of torch serve.
@msaroufim

[brainnnnup]

I think you swapped the above cases, are you running the server at your local personal laptop or desktop?
in that case it won't work for other devices If you want it there are some router configuration you can do and them expose the ports but I think even then it will be only accessible to the devices connected to router

I got it, thank you. I change the default inference_address=https://127.0.0.1:8443 to inference_address=https://0.0.0.0:8443, the problem is solved.

[Zitong-Xu]

Hello, I came across the same error, I was running in a VM, this is my TorchServe model server:

However, when i ran the inference api, it shows 404 error.

I also changed the inference_address from 127.0.0.1 to 0.0.0.0, do you have any idea on how to solve it? Thanks a lot!

[Zitong-Xu]

Solved! I forget to register the model before inference, both 127.0.0.1 and 0.0.0.0 are works for me