I'm having trouble setting up etcd auth. I have an etcd server running using a Bitnami Helm chart, and I can log in via a separate pod using etcdctl. However, my meta node is erroring with an authentication failed message. The environment variables for the username and password are correctly set in the meta pod. I've verified that the risingwave-etcd-auth secret has the correct keys and they are working by using a separate pod. How can I debug this further to resolve the authentication issue?
Cameron Braid
Asked on Feb 22, 2023
I've checked the helm values and the YAML configurations for the service and statefulset, and everything seems to be set up correctly. The secret risingwave-etcd-auth is being used properly for the ETCD_USERNAME and ETCD_PASSWORD environment variables. I've tried to reproduce the issue locally using the provided helm values. I also suggested getting a shell in the container using kubectl exec to increase the logging level by setting the RUST_LOG environment variable to debug. This revealed an error message stating that the user name is empty, which is confusing since ETCD_USERNAME is set to root. It turns out there was a breaking change in the risingwave project two weeks ago that might be causing this issue. I'll check the operator side and see if we can fix it soon.
