Adapter failing to scrape Prometheus metrics

[zalando-incubator]

Adapter failing to scrape Prometheus metrics

[szuecs]

We have no idea what banzai cloud does nor what an istio Prometheus is.

Can you explain what you think this all relates to this project?

This projects can execute Prometheus queries to provide values for metrics-server that is used by the HPA.

[mwm5945]

banzai is just a helm chart to install the kube-metrics-adapter, and istio prometheus is just a prometheus pod that istio sends its metrics to (like total requests, destination pods names, etc). I'm trying to get the kube-metrics-adapter to scrape the metrics from that particular Prometheus pod for use in a horizontal pod autoscaler. However, im getting the no plugin found error when the adapter starts up and loads the HPAs.

Essentially, it installs kube-metrics-adapter and sets the prometheus-server arg to point to the prometheus that belongs to istio.

I've also tried applying the roles/services/etc available in /docs, but to no avail (same error).

[mikkeloscar]

@mwm5945 can you link directly to how you HPA manifest looks like? Based on the error message from kube-metrics-adapter, it sounds like the HPA is not correctly defined. I'm just not sure how it looks like?

[mwm5945]

After adding some logs/building myself, i realized that the prometheus args weren't getting to the CLI, for some reason specifying the entrypoint in the dockerfile and the args in the k8's deployment was causing issues, and no prometheus URL was being sent to the CLI. Closing, thanks!