[target-allocator] generates empty configmap for collectors.

I'm trying to deploy opentelemetry collectors using the opentelemetry operator to scrape Prometheus metrics and send metrics to remote storage using prometheusremotewrite exporter.

It worked when I used the most simple config. Just add scrap jobs to prometheus receiver and use prometheusremotewrite exporter to send metrics.

However, when I tried to set up the otel collector with target allocator, the target allocator didn't generate a proper configmap for collectors.

The configmap generated by the target allocator is called allocator-collector and this configmap only has one key "collector.yaml' with an empty value. Since the configmap is incorrect, the collector pod keeps restarting.

I suppose the target allocator will create a configmap for collectors using OpenTelemetryCollector.spec.config with receivers replaced.

Version ghcr.io/open-telemetry/opentelemetry-operator/opentelemetry-operator:v0.66.0

manifest

---
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
name: allocator
spec:
mode: statefulset
replicas: 1
serviceAccount: otel-collector-sa
targetAllocator:
enabled: true
prometheusCR:
  enabled: false
  #  serviceMonitorSelector:
  #  "release": "kube-prometheus-stack"
replicas: 1
serviceAccount: otel-collector-sa
resources:
requests:
  cpu: 200m
  memory: 512Mi
limits:
  cpu: 2
  memory: 1Gi
config: |
exporters:
  prometheusremotewrite/monitoring-center:
    endpoint: http://vminsert.monitoring/insert/0/prometheus/api/v1/write
    tls:
      insecure: true
    resource_to_telemetry_conversion:
      enabled: true
extensions:
  health_check: null
  zpages: null
processors:
  batch: null
  memory_limiter:
    check_interval: 2s
    limit_mib: 1024
receivers:
  prometheus:
    config:
      global:
        scrape_interval: 30s
      scrape_configs:
      - enable_http2: true
        follow_redirects: true
        honor_timestamps: true
        job_name: cert-manager
        kubernetes_sd_configs:
        - enable_http2: true
          follow_redirects: true
          kubeconfig_file: ""
          namespaces:
            names:
            - monitoring
            own_namespace: false
          role: endpoints
        metrics_path: /metrics
        relabel_configs:
        - action: replace
          regex: (.*)
          replacement: $1
          separator: ;
          source_labels:
          - job
          target_label: __tmp_prometheus_job_name
        - action: keep
          regex: (controller);true
          replacement: $1
          separator: ;
          source_labels:
          - __meta_kubernetes_service_label_app_kubernetes_io_component
          - __meta_kubernetes_service_labelpresent_app_kubernetes_io_component
        - action: keep
          regex: (cert-manager);true
          replacement: $1
          separator: ;
          source_labels:
          - __meta_kubernetes_service_label_app_kubernetes_io_instance
          - __meta_kubernetes_service_labelpresent_app_kubernetes_io_instance
        - action: keep
          regex: (cert-manager);true
          replacement: $1
          separator: ;
          source_labels:
          - __meta_kubernetes_service_label_app_kubernetes_io_name
          - __meta_kubernetes_service_labelpresent_app_kubernetes_io_name
        - action: keep
          regex: "9402"
          replacement: $1
          separator: ;
          source_labels:
          - __meta_kubernetes_pod_container_port_number
service:
  extensions:
  - health_check
  - zpages
  pipelines:
    metrics/monitoring-center:
      exporters:
      - prometheusremotewrite/monitoring-center
      processors:
      - memory_limiter
      - batch
      receivers:
      - prometheus
  telemetry:
    metrics:
      address: 0.0.0.0:8889

open-telemetry / opentelemetry-operator

[target-allocator] generates empty configmap for collectors. #1299