Maybe your configuration is wrong and the KMS did not run correctly. Did you replace YOUR_MACHINE_PUBLIC_IP with your actual public ip in those configuration files?
@cruizba I want to create a deployment of openvidu in minikube cluster. I am doing this by using a yaml file. KMS container and other containers has been launched successfully but openvidu-server is unable to start. I got this type of logs in openvidu-server:
KMS in ws://localhost:8888/kurento is not reachable by OpenVidu Server
In logs, I find that openvidu-server is started initially with all the environment variables and due to this error, openvidu-server gets shut down.
I think may be issue due to networking in the cluster. Can you please suggest me the network configuration so that openvidu-server can reach ws://localhost:8888/kurento . And for other pods, they would be able to make an internal network between them.
Please suggest me some way so that I can make this whole deployment in one network and openvidu-server has access to kms, coturn, redis, etc and their required ports.
You need A LOT of ports bindings if you want to use KMS deployed as a Service in K8s and use the K8S DNS to access the service. So your only way to manage this excessive port binding is to run KMS in network=host.
Running KMS as network=host implies that you can not use the k8s DNS for this specific service. In minikube it may works (I don’t know i did not try it) because your cluster has only one node, and it will find what localhost is, but in real clusters, you will need to guess in what Node the KMS is deployed.
In conclustion, I don’t recomend you to deploy OpenVidu CE in a kubernetes cluster (it is not tested and we don’t support it). But in your case, I think that running all the services as network=host should do the trick.