Alertmanager pod is stuck in pending state if the masters are not labeled with "node-role.kubernetes.io/worker"

[pratik705]

Describe the bug Following affinity is set for Alertmanager[1]. Additionally, nodeSelector is also set to the controller[2]. If the controllers/masters are not labeled as node-role.kubernetes.io/worker, Alertmanager will be stuck in a pending state, as expected. To resolve the issue, I had to remove the affinity. [1]

    affinity:
      nodeAffinity:
        requiredDuringSchedulingIgnoredDuringExecution:
          nodeSelectorTerms:
            - matchExpressions:
                - key: node-role.kubernetes.io/worker
                  operator: In
                  values:
                    - worker

[2]

      nodeSelector:
        openstack-control-plane: enabled

To Reproduce Steps to reproduce the behavior:

1. Dont label the master/controlelrs with node-role.kubernetes.io/worker.
2. Try deploying the prometheus following the instructions

Current behavior

• Alertmanager will be stuck in a pending state.

  kubectl get pod -A |egrep -i prom
  prometheus       alertmanager-kube-prometheus-stack-alertmanager-0           0/2     Pending   0             109s
  prometheus       kube-prometheus-stack-kube-state-metrics-7db65fb76b-fjldc   1/1     Running   0             18h
  prometheus       kube-prometheus-stack-operator-b87456bcf-k2rnj              1/1     Running   0             18h
  prometheus       kube-prometheus-stack-prometheus-node-exporter-5kv9p        1/1     Running   0             18h
  prometheus       kube-prometheus-stack-prometheus-node-exporter-7dfj7        1/1     Running   0             18h
  prometheus       kube-prometheus-stack-prometheus-node-exporter-bl4rf        1/1     Running   0             18h
  prometheus       kube-prometheus-stack-prometheus-node-exporter-f6bct        1/1     Running   0             18h
  prometheus       kube-prometheus-stack-prometheus-node-exporter-g9hzj        1/1     Running   0             18h
  prometheus       kube-prometheus-stack-prometheus-node-exporter-q8twv        1/1     Running   0             18h
  prometheus       prometheus-kube-prometheus-stack-prometheus-0               2/2     Running   0             6m44s

Expected behavior

• Alertmanager should be in running state.