需求

收集 ES 的指标, 并进行展示和告警;

现状

  1. ES 通过 docker compose 安装
  2. 所在环境的 K8S 集群有 Prometheus 和 AlertManager 及 Grafana

方案

复用现有的监控体系, 通过: Prometheus 监控 ES.

Prometheus 监控 ES 架构

具体实现为:

采集端 elasticsearch_exporter

可以监控的指标为:

Name Type Cardinality Help
elasticsearch_breakers_estimated_size_bytes gauge 4 Estimated size in bytes of breaker
elasticsearch_breakers_limit_size_bytes gauge 4 Limit size in bytes for breaker
elasticsearch_breakers_tripped counter 4 tripped for breaker
elasticsearch_cluster_health_active_primary_shards gauge 1 The number of primary shards in your cluster. This is an aggregate total across all indices.
elasticsearch_cluster_health_active_shards gauge 1 Aggregate total of all shards across all indices, which includes replica shards.
elasticsearch_cluster_health_delayed_unassigned_shards gauge 1 Shards delayed to reduce reallocation overhead
elasticsearch_cluster_health_initializing_shards gauge 1 Count of shards that are being freshly created.
elasticsearch_cluster_health_number_of_data_nodes gauge 1 Number of data nodes in the cluster.
elasticsearch_cluster_health_number_of_in_flight_fetch gauge 1 The number of ongoing shard info requests.
elasticsearch_cluster_health_number_of_nodes gauge 1 Number of nodes in the cluster.

...

可以直接在 github 上找到完整的

展示端 基于Grafana

?️ Reference:

ElasticSearch dashboard for Grafana | Grafana Labs

Grafana ES 仪表板

告警指标 基于prometheus alertmanager

?️ Reference:

ElasticSearch:https://awesome-prometheus-alerts.grep.to/rules.html#elasticsearch-1

Prometheus ES Alert 界面

实施步骤

以下为手动实施步骤

Docker Compose

docker pull quay.io/prometheuscommunity/elasticsearch-exporter:v1.3.0

docker-compose.yml 示例:

? Warning:

exporter 在每次刮削时都会从 ElasticSearch 集群中获取信息,因此过短的刮削间隔会给 ES 主节点带来负载,特别是当你使用 --es.all--es.indices 运行时。我们建议你测量获取/_nodes/stats/_all/_stats对你的ES集群来说需要多长时间,以确定你的刮削间隔是否太短。

原 ES 的 docker-copmose.yml 示例如下:

version: '3'
services:
  elasticsearch:
    image: elasticsearch-plugins:6.8.18
    ...
    ports:
      - 9200:9200
      - 9300:9300
    restart: always

增加了 elasticsearch_exporter 的yaml如下:

version: '3'
services:
  elasticsearch:
    image: elasticsearch-plugins:6.8.18
    ...
    ports:
      - 9200:9200
      - 9300:9300
    restart: always
  elasticsearch_exporter:
      image: quay.io/prometheuscommunity/elasticsearch-exporter:v1.3.0
      command: 
      - '--es.uri=http://elasticsearch:9200'
      - '--es.all'
      - '--es.indices'
      - '--es.indices_settings'
      - '--es.indices_mappings'
      - '--es.shards'
      - '--es.snapshots'
      - '--es.timeout=30s'      
      restart: always
      ports:
      - "9114:9114"    

Prometheus 配置调整

prometheus 配置

Prometheus 增加静态抓取配置:

scrape_configs:
  - job_name: "es"
    static_configs:
      - targets: ["x.x.x.x:9114"]

说明:

x.x.x.x 为 ES Exporter IP, 因为 ES Exporter 通过 docker compose 和 ES部署在同一台机器, 所以这个 IP 也是 ES 的IP.

Prometheus Rules

增加 ES 相关的 Prometheus Rules:

groups:
  - name: elasticsearch
    rules:
      - record: elasticsearch_filesystem_data_used_percent
        expr: 100 * (elasticsearch_filesystem_data_size_bytes - elasticsearch_filesystem_data_free_bytes)
          / elasticsearch_filesystem_data_size_bytes
      - record: elasticsearch_filesystem_data_free_percent
        expr: 100 - elasticsearch_filesystem_data_used_percent
      - alert: ElasticsearchTooFewNodesRunning
        expr: elasticsearch_cluster_health_number_of_nodes < 3
        for: 0m
        labels:
          severity: critical
        annotations:
          description: "Missing node in Elasticsearch cluster\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
          summary: ElasticSearch running on less than 3 nodes(instance {{ $labels.instance }}, node {{$labels.node}})
      - alert: ElasticsearchDiskSpaceLow
        expr: elasticsearch_filesystem_data_free_percent < 20
        for: 2m
        labels:
          severity: warning
        annotations:
          summary: Elasticsearch disk space low (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "The disk usage is over 80%\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchDiskOutOfSpace
        expr: elasticsearch_filesystem_data_free_percent < 10
        for: 0m
        labels:
          severity: critical
        annotations:
          summary: Elasticsearch disk out of space (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "The disk usage is over 90%\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchHeapUsageWarning
        expr: (elasticsearch_jvm_memory_used_bytes{area="heap"} / elasticsearch_jvm_memory_max_bytes{area="heap"}) * 100 > 80
        for: 2m
        labels:
          severity: warning
        annotations:
          summary: Elasticsearch Heap Usage warning (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "The heap usage is over 80%\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchHeapUsageTooHigh
        expr: (elasticsearch_jvm_memory_used_bytes{area="heap"} / elasticsearch_jvm_memory_max_bytes{area="heap"}) * 100 > 90
        for: 2m
        labels:
          severity: critical
        annotations:
          summary: Elasticsearch Heap Usage Too High (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "The heap usage is over 90%\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchClusterRed
        expr: elasticsearch_cluster_health_status{color="red"} == 1
        for: 0m
        labels:
          severity: critical
        annotations:
          summary: Elasticsearch Cluster Red (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elastic Cluster Red status\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchClusterYellow
        expr: elasticsearch_cluster_health_status{color="yellow"} == 1
        for: 0m
        labels:
          severity: warning
        annotations:
          summary: Elasticsearch Cluster Yellow (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elastic Cluster Yellow status\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchHealthyDataNodes
        expr: elasticsearch_cluster_health_number_of_data_nodes < 3
        for: 0m
        labels:
          severity: critical
        annotations:
          summary: Elasticsearch Healthy Data Nodes (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Missing data node in Elasticsearch cluster\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchRelocatingShards
        expr: elasticsearch_cluster_health_relocating_shards > 0
        for: 0m
        labels:
          severity: info
        annotations:
          summary: Elasticsearch relocating shards (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elasticsearch is relocating shards\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchRelocatingShardsTooLong
        expr: elasticsearch_cluster_health_relocating_shards > 0
        for: 15m
        labels:
          severity: warning
        annotations:
          summary: Elasticsearch relocating shards too long (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elasticsearch has been relocating shards for 15min\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchInitializingShards
        expr: elasticsearch_cluster_health_initializing_shards > 0
        for: 0m
        labels:
          severity: info
        annotations:
          summary: Elasticsearch initializing shards (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elasticsearch is initializing shards\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchInitializingShardsTooLong
        expr: elasticsearch_cluster_health_initializing_shards > 0
        for: 15m
        labels:
          severity: warning
        annotations:
          summary: Elasticsearch initializing shards too long (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elasticsearch has been initializing shards for 15 min\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchUnassignedShards
        expr: elasticsearch_cluster_health_unassigned_shards > 0
        for: 0m
        labels:
          severity: critical
        annotations:
          summary: Elasticsearch unassigned shards (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elasticsearch has unassigned shards\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchPendingTasks
        expr: elasticsearch_cluster_health_number_of_pending_tasks > 0
        for: 15m
        labels:
          severity: warning
        annotations:
          summary: Elasticsearch pending tasks (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "Elasticsearch has pending tasks. Cluster works slowly.\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"
      - alert: ElasticsearchNoNewDocuments
        expr: increase(elasticsearch_indices_docs{es_data_node="true"}[10m]) < 1
        for: 0m
        labels:
          severity: warning
        annotations:
          summary: Elasticsearch no new documents (instance {{ $labels.instance }}, node {{$labels.node}})
          description: "No new documents for 10 min!\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

并重启生效.

?Warning:

  • ElasticsearchTooFewNodesRunning 告警的条件是 es 集群的node 少于 3个, 对于单节点 ES 会误报, 所以按需开启rule或按需屏蔽(slience).
  • ElasticsearchHealthyDataNodes 告警同上.

AlertManager 告警规则及收件人配置

按需调整, 示例如下:

'global':
  'smtp_smarthost': ''
  'smtp_from': ''
  'smtp_require_tls': false
  'resolve_timeout': '5m'
'receivers':
  - 'name': 'es-email'
    'email_configs':
      - 'to': 'sfw@example.com,sdfwef@example.com'
        'send_resolved': true
'route':
  'group_by':
    - 'job'
  'group_interval': '5m'
  'group_wait': '30s'
  'routes':
    - 'receiver': 'es-email'
      'match':
        'job': 'es'

并重启生效.

Grafana 配置

导入 json 格式的 Grafana Dashboard:

{
    "__inputs": [],
    "__requires": [
        {
            "type": "grafana",
            "id": "grafana",
            "name": "Grafana",
            "version": "5.4.0"
        },
        {
            "type": "panel",
            "id": "graph",
            "name": "Graph",
            "version": "5.0.0"
        },
        {
            "type": "datasource",
            "id": "prometheus",
            "name": "Prometheus",
            "version": "5.0.0"
        },
        {
            "type": "panel",
            "id": "singlestat",
            "name": "Singlestat",
            "version": "5.0.0"
        }
    ],
...

可以直接在 Grafana 上找到完整的

?️ 参考文档

三人行, 必有我师; 知识共享, 天下为公. 本文由东风微鸣技术博客 EWhisper.cn 编写.