Docker error "Couldn't load integration instances from config file"

[arielsalvo]

Hi

I kept getting the following error when trying out docker image newrelic/k8s-nri-haproxy:2.1.0:

"Couldn't load integration instances from config file" component=PluginRegistry configFile=haproxy-config.yml error="Integration definition not found" integration=com.newrelic.haproxy

Had to import the definition to another directory for newrelic-infrastructure to pick it up: /var/db/newrelic-infra/custom-integrations/haproxy-definition.yml This should be changed in: https://github.com/newrelic/nri-haproxy/blob/10f84727b7dd6ed6185f9b9c893461470b2f12c8/Dockerfile#L11

Also, the path to the sidecar in the already included definition appears not to work unless changed from ./bin/nri-haproxy to the absolute path /nri-sidecar/newrelic-infra/newrelic-integrations/bin/nri-haproxy

After these changed, it worked for me. Is there a better way to do this?

Thanks! --Ariel

[camdencheek]

Hi @arielsalvo -- The generally recommended way of doing this is to use a ConfigMap. It should look something like the following:

apiVersion: v1
kind: ConfigMap
metadata:
  name: haproxy-newrelic-integrations-config
  namespace: default
data:
  config.yaml: |
    integration_name: com.newrelic.haproxy
    instances:
      - name: haproxy
        # Command can be all_data, metrics, or inventory
        command: all_data
        arguments:
          stats_url: http://127.0.0.1:7777/stats
          username: cacheuser
          password: password
          cluster_name: mmacias
  definition.yaml: |
    name: com.newrelic.haproxy
    description: Reports status and metrics for haproxy service
    protocol_version: 2
    os: linux
    commands:
      all_data:
        command:
          - ./bin/nr-haproxy
        prefix: config/haproxy
        interval: 15
      metrics:
        command:
          - ./bin/nr-haproxy
          - --metrics
        interval: 15
      inventory:
        command:
          - ./bin/nr-haproxy
          - --inventory
        prefix: config/haproxy
        interval: 15

[arielsalvo]

Hi, @camdencheek

You are correct but my tests were not yet in K8s, just docker. Still, shouldn't the definition be added to the image in the right place and with the correct path to the bin? Otherwise, why go through the effort of adding the file to the image?

[camdencheek]

Hi @arielsalvo -- sorry for the slow reply. This one dropped off my radar. I'm not terribly familiar with the design behind the docker images, but my understanding is that they were built specifically for monitoring in kubernetes as a sidecar. Running in a standalone docker container is not currently supported

[arielsalvo]

Hi, @camdencheek

No problem.

What I'm trying to say is that the file haproxy-definition.yml that is being added to the image is useless in both use cases (k8s and standalone). If you changed the path and location as I stated in my first comment, you wouldn't need to add it as a config_map and it would work out of the box in both environments only adding a config_map for config.yaml (that contains the config for what you are monitoring).

In any case, since this is for convenience and can easily be fixed by adding the config_map, you can close this issue but I honestly think this should and could be modified.

[camdencheek]

Thanks for the input -- I've passed it on to the team that maintains our Kubernetes offering