ignore_metrics doesnt work? about nri-prometheus HOT 4 CLOSED

Hi @bpschmitt ignore_metrics: is part of transformations: object. Please try using the following:

transformations: 
  - description: "Description" 
    ignore_metrics:
      - prefixes:
        - "http_"

I think the Docs examples could lead you to this issue. I will propose a change there.

from nri-prometheus.

Hi @bpschmitt!

The kube_ prefix metrics that are being included can represent a huge number of metrics, as there are many metrics exposed about each one of these resources (hpas, daemonset, etc).

I recommend that you start allowing only nr_stats and apiserver_request_total and start opening up for more things little by little.

Also it's worth to mention that, depending on what there is running in the cluster, there might be other dimensional metrics being sent to New Relic. Even maybe another instance of nri-prometheus?

You can query FROM Metric select uniques(metricName) where integrationName = 'nri-prometheus' to verify.

from nri-prometheus.

Revisiting this. I'm now trying to ignore specific metrics as outlined in our docs. Here is the config being presented to my POMI pod.

/etc/nri-prometheus $ cat config.yaml
cluster_name: bschmitt-kops
require_scrape_enabled_label_for_nodes: true
ignore_metrics:
  - prefixes:
    - "http_"

POMI logs:

time="2020-08-31T21:55:15Z" level=info msg="Starting New Relic's Prometheus OpenMetrics Integration version 2.0.0"

Still seeing http_* metrics showing up for my cluster. The ignore_metrics feature still does not seem to be working properly. Halp! @jorik @douglascamata

from nri-prometheus.

I'll close this issue, docs are in process to be updated , feel free to reopen it for any other question, thanks!

from nri-prometheus.