Closed exArax closed 4 years ago
Prometheus scrapes node_exporter from all nodes using internal IPs from the node_exporter pods/services.
When you say "down status" you mean in Prometheus Targets?
@carlosedp yes exactly
First of all, I'd advise against having your nodes exposed to the internet with valid IPs, but it shouldn't affect the usage.
We'd need more information:
curl -k
one of the IPs in port 9100? It will fail since there is no authentication but that's fine.When I am doing curl from the host I am getting unauthorized message, if I perform it from another device I am getting timed out message. I will try to find answers to the other two questions and I will post them here, thank you @carlosedp
Apparently there is something blocking the access between the hosts on the needed port since curl didn't work from outside.
In the ufw of the devices I have allowed the required ports. These are the logs for node-exporter and kube-rbac-proxy:
I changed the port of the node-exporters and now it works !!! @carlosedp thank you for your time :)
I am setting up a cluster with public ip. The k3s master is a cloud vm and I have two rpis as workers all of them they have public ip, so they are not in the same LAN. The node exporters in the rpis have a down status and they show a context deadline exceeded error. I cannot find what's going wrong. The weird fact is that the cadvisors of rpis have up statuses, and all the required ports are open.