postgres-operator: postgres-exporter - pgBackRest query fails in untrusted TLS environment

Describe the bug I have a postgres cluster is working on Azure with metrics enabled. Everything worked fine except for postgres-exporter, it only worked on replicas deployment

To Reproduce Steps to reproduce the behavior:

1. My own cluster is deployed with this command(be noticed that all requirements are prepared already: s3, tls, prometheus,…):

pgo create cluster infra \
    --pgbouncer \
    --service-type=LoadBalancer \
    --server-ca-secret=postgresql-ca \
    --server-tls-secret=infra-tls-keypair \
    --pgbouncer-tls-secret=infra-tls-keypair \
    --replication-tls-secret=infra-rep-tls-keypair \
    --pgbouncer-replicas=2 \
    --replica-count=1 \
    --pvc-size=10Gi \
    --pgbackrest-storage-type=s3 \
    --pgbackrest-s3-bucket=<S3-BUCKET> \
    --pgbackrest-s3-endpoint=<S3-ENDPOINT> \
    --pgbackrest-s3-key=<S3-REGION> \
    --pgbackrest-s3-key-secret=<S3-PROJECT> \
    --pgbackrest-s3-region=eu-cenral-1 \
    --pgbackrest-s3-uri-style=path \
    --pgbackrest-s3-verify-tls=false \
    --metrics

1. I haved added 2 service monitors from my prometheus operator, they are:

leader-postgres-exporter:

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  annotations:
    meta.helm.sh/release-name: prometheus
    meta.helm.sh/release-namespace: cicd
  labels:
    app: prometheus-operator-prometheus
    app.kubernetes.io/managed-by: Helm
    chart: prometheus-operator-8.13.8
    heritage: Helm
    pg-cluster-role: master
    release: prometheus
  name: psql-infra-master
  namespace: cicd
spec:
  endpoints:
  - port: postgres-exporter
  jobLabel: psql-infra-master
  namespaceSelector:
    matchNames:
    - pgo
  selector:
    matchLabels:
      name: infra
      pg-cluster: infra

replica-postgres-exporter:

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  annotations:
    meta.helm.sh/release-name: prometheus
    meta.helm.sh/release-namespace: cicd
  labels:
    app: prometheus-operator-prometheus
    app.kubernetes.io/managed-by: Helm
    chart: prometheus-operator-8.13.8
    heritage: Helm
    pg-cluster-role: replica
    release: prometheus
  name: psql-infra-replica
  namespace: cicd
spec:
  endpoints:
  - port: postgres-exporter
  jobLabel: psql-infra-replica
  namespaceSelector:
    matchNames:
    - pgo
  selector:
    matchLabels:
      name: infra-replica
      pg-cluster: infra

1. From my prometheus, I see leader-postgres-exporter is always unaccessible. Even if I switched role between them, it is always unaccessible.

1. Finally I looked at how differences between theirs k8s manifests. I see database sidecar from leader pod missed these configurations, exporter pods are same:

I have also tried to run curl inside leader exporter pod, and the result is failed:

Expected behavior I am not sure it’s feature or bug, I just want to have 2 dedicated metrics for both of them.

Screenshots (Included already)

Please tell us about your environment:

• Operating System: Linux
• Where is this running ( Local, Cloud Provider): Azure
• Storage being used (NFS, Hostpath, Gluster, etc): Azure persistent volumes
• Container Image Tag: registry.developers.crunchydata.com/crunchydata/crunchy-postgres-exporter:centos8-4.6.0
• PostgreSQL Version: registry.developers.crunchydata.com/crunchydata/crunchy-postgres-ha:centos8-13.1-4.6.0
• Platform (Docker, Kubernetes, OpenShift): Kubernetes
• Platform Version: 1.17

Additional context Add any other context about the problem here.