Uploaded image for project: 'OpenShift Logging'
  1. OpenShift Logging
  2. LOG-2294

[Vector] Vector internal metrics are not exposed via HTTPS due to which OpenShift Monitoring Prometheus service cannot scrape the metrics endpoint.

XMLWordPrintable

    • Logging (Core) - Sprint 216, Logging (Core) - Sprint 217

      Vector TP

      OCP Version: 4.10.0-0.nightly-2022-02-26-230022

      Description of the problem:

      Vector Collector metrics exported by the Prometheus sink are bind to the port 24231 and only accessible via HTTP. 

      [sinks.prometheus_output]
type = "prometheus_exporter"
inputs = ["internal_metrics"]
address = "0.0.0.0:24231"
default_namespace = "collector"

      The OpenShift Monitoring Prometheus service expects the Vector Collector metrics endpoint listening on HTTPS port 24231 (port: metrics).  

      oc get servicemonitors.monitoring.coreos.com collector -o yaml
apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  creationTimestamp: "2022-03-02T11:00:17Z"
  generation: 1
  name: collector
  namespace: openshift-logging
  ownerReferences:
  - apiVersion: logging.openshift.io/v1
    controller: true
    kind: ClusterLogging
    name: instance
    uid: a000e25d-d6dd-444e-843d-f827c2bc79cf
  resourceVersion: "141175"
  uid: 995e45dc-4ac9-4d6e-b857-0fb383ea6513
spec:
  endpoints:
  - bearerTokenSecret:
      key: ""
    path: /metrics
    port: metrics
    scheme: https
    tlsConfig:
      ca: {}
      caFile: /etc/prometheus/configmaps/serving-certs-ca-bundle/service-ca.crt
      cert: {}
      serverName: collector.openshift-logging.svc
  - bearerTokenSecret:
      key: ""
    path: /metrics
    port: logfile-metrics
    scheme: https
    tlsConfig:
      ca: {}
      caFile: /etc/prometheus/configmaps/serving-certs-ca-bundle/service-ca.crt
      cert: {}
      serverName: collector.openshift-logging.svc
  jobLabel: monitor-collector
  namespaceSelector:
    matchNames:
    - openshift-logging
  selector:
    matchLabels:
      logging-infra: support

      The Prometheus service is not able to scrape the metrics endpoint and the Collector target is shown as down.

      Steps to reproduce the issue:

      1 Deploy ClusterLogging instance with Vector as collector.

      apiVersion: "logging.openshift.io/v1"
kind: "ClusterLogging"
metadata:
  name: "instance" 
  namespace: "openshift-logging"
spec:
  managementState: "Managed"  
  logStore:
    type: "elasticsearch"  
    retentionPolicy: 
      application:
        maxAge: 7d
      infra:
        maxAge: 7d
      audit:
        maxAge: 7d
    elasticsearch:
      nodeCount: 3 
      storage: {}
      resources: 
          requests:
            memory: "1Gi"
      proxy: 
        resources:
          limits:
            memory: 256Mi
          requests:
            memory: 256Mi
      redundancyPolicy: "SingleRedundancy"
  visualization:
    type: "kibana"  
    kibana:
      replicas: 1
  collection:
    logs:
      type: "vector"  
      vector: {}

      2 Check that the Prometheus Monitoring targets on the OpenShift Console - Observe - Targets - Metrics target are shows as down for Collector.

      3 Access the Collector metrics endpoint via HTTPS and HTTP. It fails via HTTPS but is accessible via HTTP. 

      $ oc exec collector-fmvf5 -c collector -- curl -k -H "Authorization: Bearer `oc sa get-token prometheus-k8s -n openshift-monitoring`" -s -H "Content-type: application/json" https://collector.openshift-logging.svc:24231/metrics
command terminated with exit code 35

$ oc exec collector-fmvf5 -c collector -- curl -k -H "Authorization: Bearer `oc sa get-token prometheus-k8s -n openshift-monitoring`" -s -H "Content-type: application/json" http://collector.openshift-logging.svc:24231/metrics
# HELP vector_adaptive_concurrency_averaged_rtt adaptive_concurrency_averaged_rtt
# TYPE vector_adaptive_concurrency_averaged_rtt histogram
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="-inf"} 0 1646216619065
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="0.015625"} 31 1646216619065
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="0.03125"} 72 1646216619065
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="0.0625"} 85 1646216619065
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="0.125"} 86 1646216619065
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="0.25"} 87 1646216619065
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="0.5"} 87 1646216619065
vector_adaptive_concurrency_averaged_rtt_bucket{component_kind="sink",component_name="default",component_type="elasticsearch",le="0"} 87 1646216619065

      4 Extract and check the Vector config.

      # Logs from containers (including openshift containers)
[sources.raw_container_logs]
type = "kubernetes_logs"
auto_partial_merge = true
exclude_paths_glob_patterns = ["/var/log/pods/openshift-logging_collector-*/*/*.log", "/var/log/pods/openshift-logging_elasticsearch-*/*/*.log", "/var/log/pods/openshift-logging_kibana-*/*/*.log"][sources.raw_journal_logs]
type = "journald"[sources.internal_metrics]
type = "internal_metrics"[transforms.container_logs]
type = "remap"
inputs = ["raw_container_logs"]
source = '''
  level = "unknown"
  if match(.message,r'(Warning|WARN|W[0-9]+|level=warn|Value:warn|"level":"warn")'){
    level = "warn"
  } else if match(.message, r'Info|INFO|I[0-9]+|level=info|Value:info|"level":"info"'){
    level = "info"
  } else if match(.message, r'Error|ERROR|E[0-9]+|level=error|Value:error|"level":"error"'){
    level = "error"
  } else if match(.message, r'Debug|DEBUG|D[0-9]+|level=debug|Value:debug|"level":"debug"'){
    level = "debug"
  }
  .level = level  .pipeline_metadata.collector.name = "vector"
  .pipeline_metadata.collector.version = "0.14.1"
  ip4, err = get_env_var("NODE_IPV4")
  .pipeline_metadata.collector.ipaddr4 = ip4
  received, err = format_timestamp(now(),"%+")
  .pipeline_metadata.collector.received_at = received
  .pipeline_metadata.collector.error = err
 '''[transforms.journal_logs]
type = "remap"
inputs = ["raw_journal_logs"]
source = '''
  level = "unknown"
  if match(.message,r'(Warning|WARN|W[0-9]+|level=warn|Value:warn|"level":"warn")'){
    level = "warn"
  } else if match(.message, r'Info|INFO|I[0-9]+|level=info|Value:info|"level":"info"'){
    level = "info"
  } else if match(.message, r'Error|ERROR|E[0-9]+|level=error|Value:error|"level":"error"'){
    level = "error"
  } else if match(.message, r'Debug|DEBUG|D[0-9]+|level=debug|Value:debug|"level":"debug"'){
    level = "debug"
  }
  .level = level  .pipeline_metadata.collector.name = "vector"
  .pipeline_metadata.collector.version = "0.14.1"
  ip4, err = get_env_var("NODE_IPV4")
  .pipeline_metadata.collector.ipaddr4 = ip4
  received, err = format_timestamp(now(),"%+")
  .pipeline_metadata.collector.received_at = received
  .pipeline_metadata.collector.error = err
 '''
[transforms.route_container_logs]
type = "route"
inputs = ["container_logs"]
route.app = '!((starts_with!(.kubernetes.pod_namespace,"kube")) || (starts_with!(.kubernetes.pod_namespace,"openshift")) || (.kubernetes.pod_namespace == "default"))'
route.infra = '(starts_with!(.kubernetes.pod_namespace,"kube")) || (starts_with!(.kubernetes.pod_namespace,"openshift")) || (.kubernetes.pod_namespace == "default")'
# Rename log stream to "application"
[transforms.application]
type = "remap"
inputs = ["route_container_logs.app"]
source = """
.log_type = "application"
"""
# Rename log stream to "infrastructure"
[transforms.infrastructure]
type = "remap"
inputs = ["route_container_logs.infra","journal_logs"]
source = """
.log_type = "infrastructure"
"""
[transforms.pipeline_0_]
type = "remap"
inputs = ["application","infrastructure"]
source = """
.
"""
# Adding _id field
[transforms.default_add_es_id]
type = "remap"
inputs = ["pipeline_0_"]
source = """
index = "default"
if (.log_type == "application"){
  index = "app"
}
if (.log_type == "infrastructure"){
  index = "infra"
}
if (.log_type == "audit"){
  index = "audit"
}
."write-index"=index+"-write"
._id = encode_base64(uuid_v4())
"""[transforms.default_dedot_and_flatten]
type = "lua"
inputs = ["default_add_es_id"]
version = "2"
hooks.process = "process"
source = """
    function process(event, emit)
        if event.log.kubernetes == nil then
            return
        end
        dedot(event.log.kubernetes.pod_labels)
        -- create "flat_labels" key
        event.log.kubernetes.flat_labels = {}
        i = 1
        -- flatten the labels
        for k,v in pairs(event.log.kubernetes.pod_labels) do
          event.log.kubernetes.flat_labels[i] = k.."="..v
          i=i+1
        end
        -- delete the "pod_labels" key
        event.log.kubernetes["pod_labels"] = nil
        emit(event)
    end    function dedot(map)
        if map == nil then
            return
        end
        local new_map = {}
        local changed_keys = {}
        for k, v in pairs(map) do
            local dedotted = string.gsub(k, "%.", "_")
            if dedotted ~= k then
                new_map[dedotted] = v
                changed_keys[k] = true
            end
        end
        for k in pairs(changed_keys) do
            map[k] = nil
        end
        for k, v in pairs(new_map) do
            map[k] = v
        end
    end
"""[sinks.default]
type = "elasticsearch"
inputs = ["default_dedot_and_flatten"]
endpoint = "https://elasticsearch.openshift-logging.svc:9200"
index = "{{ write-index }}"
request.timeout_secs = 2147483648
bulk_action = "create"
id_key = "_id"
# TLS Config
[sinks.default.tls]
key_file = "/var/run/ocp-collector/secrets/collector/tls.key"
crt_file = "/var/run/ocp-collector/secrets/collector/tls.crt"
ca_file = "/var/run/ocp-collector/secrets/collector/ca-bundle.crt"
[sinks.prometheus_output]
type = "prometheus_exporter"
inputs = ["internal_metrics"]
address = "0.0.0.0:24231"
default_namespace = "collector"

      Additional notes:

      With Fluentd we bind the metrics endpoint via HTTPS:

      # Prometheus Monitoring
<source>
  @type prometheus
  bind "#{ENV['POD_IP']}"
  <ssl>
    enable true
    certificate_path "#{ENV['METRICS_CERT'] || '/etc/fluent/metrics/tls.crt'}"
    private_key_path "#{ENV['METRICS_KEY'] || '/etc/fluent/metrics/tls.key'}"
  </ssl>
</source><source>
  @type prometheus_monitor
  <labels>
    hostname ${hostname}
  </labels>
</source>

       

            prangupt Pranjal Gupta (Inactive)
            rhn-support-ikanse Ishwar Kanse
            Ishwar Kanse Ishwar Kanse
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Created:
              Updated:
              Resolved: