How to get the correct Riva Ip and hosting port while using it on aws instance

AI & Data Science Deep Learning (Training & Inference) Riva
1

https://docs.nvidia.com/deeplearning/riva/user-guide/docs/samples/weather.html

in this dock step, 3 says to modify the config.py file with the Riva IP and Host port but how to figure out which IP and port to use if the model is on was and you want to access the model from local pc.

2

Hi @a1809073 ,
You need to provide the public IP of the aws machine, if the virtual assistant is being run on local machine and riva services is hosted on aws
e.g. on the aws ec2 instance web console, they list public ip address of each machine, e.g. 65.42.3.21
then in local machine virtual assistant can have ip as 65.42.3.21:50051
on web console of ec2 instance, they need to make sure that port 50051 is open for incoming traffic

Thanks!

3

Even after changing the ip in config file to the public ip of the AWS instance machine and updating the inbound, outbound security of the instance. The server runs on 0.0.0.0:8009/rivaWeather and we even tried to change the 0.0.0.0 to the public ip of instace in our web browser but still unable to access it.

tmp_1630923704632
tmp_16309237046321200×73 14.6 KB

4

tmp_1630923696147__01
tmp_1630923696147__011144×250 84.7 KB

5

Hi @a1809073 ,
There could be two things,

  1. if the chatbot is running on local machine it will launch on 0.0.0.0:8009/rivaWeather and the 0.0.0.0 here means the IP of local machine and not the aws instance
  2. Sample Apps — NVIDIA Riva in the second step here you need to provide --net=host for the riva-samples container to be able to access external riva server.

Thanks!

1 Like
7

Thanks AakankshaS, it is working perfectly fine now after providing the public IP as the Riva server IP and changing some inbound and outbound security rules of the instance for port and --net==host worked for us.

thanks for the help.

1 Like