root@longhorn-manager:/# curl 10.43.0.3:9500/metrics

do not return all metrics for example

longhorn_instance_manager_cpu_usage_millicpu

or

longhorn_node_cpu_capacity_millicpu

Should I fetch it from other pod? on which port ?

For example, you can

kl exec -it longhorn-manager-xxxxxx -- curl <http://longhorn-backend:9500/metrics>

ref: https://github.com/longhorn/longhorn/issues/2817

@icy-agency-38675 the ip

10.43.0.3

is longhorn-manager so it is almost the same as

curl <http://longhorn-backend:9500/metrics>

I do something like this

root@longhorn-manager-g5nck:/# while true ; do curl -s  <http://longhorn-backend:9500/metrics> | grep longhorn_node_cpu_usage_millicpu ; done

but the output is empty metric

longhorn_node_cpu_usage_millicpu

is not exported for some reason