enable metrics_collector

[yanqin001]

On the management UI web, if click the "Channels" button, it will show "Stats in management UI are disabled on this node“

i add "management_agent.disable_metrics_collector = false" config in the rabbitmq yaml like this

apiVersion: rabbitmq.com/v1beta1
kind: RabbitmqCluster
metadata:
  name: kb-rabbitmq
  namespace: kb
spec:
  replicas: 3
  resources:
    requests:
      cpu: 2
      memory: 5Gi
    limits:
      cpu: 2
      memory: 5Gi
  rabbitmq:
    additionalConfig: |
      default_user=admin
      default_pass=9zPd2h8A
      cluster_partition_handling = pause_minority
      vm_memory_high_watermark_paging_ratio = 0.99
      disk_free_limit.relative = 1.0
      collect_statistics_interval = 10000
      management_agent.disable_metrics_collector = false
  persistence:
    storageClassName: longhorn
    storage: "100Gi"
  affinity:
    podAntiAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
      - labelSelector:
          matchExpressions:
            - key: app.kubernetes.io/name
              operator: In
              values:
              - kb-rabbitmq
        topologyKey: kubernetes.io/hostname

but the config dose not work, how the channels info can be seen from the webpage.

[mkuratczyk]

There is no mention of the image in your definition. What image are you using?

[yanqin001]

operator image is rabbitmqoperator/cluster-operator:2.0.0
rabbitmq image is rabbitmq:3.9.26

[qibint]

@mkuratczyk Having the same issue with enabling the metrics collector through this additionalConfig.
image: rabbitmq:3.11.10
rabbitmqoperator/cluster-operator:2.1.0
Can you help? Thanks

[qibint]

Hey @mkuratczyk.
We're using RabbitMQ Cluster Kubernetes Operator and create the rabbitmq instances from it following - https://www.rabbitmq.com/kubernetes/operator/using-operator.html#advanced-config. I will definitely need the flexibility to turn on/off this metric through this configuration file.

[mkuratczyk]

Please provide your YAML.

PS No need to mention me in every post. :)

[qibint]

Thank you for looking into this.
The config is simple:

apiVersion: rabbitmq.com/v1beta1
kind: RabbitmqCluster
metadata:
  name: amqp-workers
spec:
  image: rabbitmq:3.11.10
  service:
    type: ClusterIP
  replicas: 3
  persistence:
    storage: 20Gi
    storageClassName: gp2
  rabbitmq:
    additionalConfig: |
      # https://www.rabbitmq.com/configure.html
      # management metrics can be disabled altogether
      management_agent.disable_metrics_collector = false

[mkuratczyk]

Ok, I see the problem. Yes, the rabbitmq image contain a config file which is alphabetically later than all files generated by the Operator and therefore it "wins". I think the easiest solution would be to rename this file in the image so that it plays nicely with the Operator. The image already contains 10-defaults.conf, so the pattern is there. It could be renamed to something like 15-disable_metrics_collector.conf in these places:

https://github.com/docker-library/rabbitmq/blob/master/Dockerfile-alpine.template#L298-L299
https://github.com/docker-library/rabbitmq/blob/master/Dockerfile-ubuntu.template#L292-L295

and apply the templates (./update.sh).

I recommend you submit a PR to that repo (it's maintained by Docker).

[qibint]

Thanks

[mkuratczyk]

I'm closing this issue since it doesn't belong here. I consider this an issue in https://github.com/docker-library/rabbitmq.
One could argue it's an issue in https://github.com/rabbitmq/cluster-operator, but not rabbitmq-server for sure.