娃哈哈好喝-真的！
技术够用就行，吃好喝好睡好！
当前位置：娃哈哈好喝 > 监控系统 > Loki > 正文

多节点loki读写分离+minio+grafana实现nginx日志收集展示

2022-10-10 分类：Loki 阅读(744)

系统版本：centos7.6，loki版本：2.7.1，minio：2021-11-24T23:19:33Z

环境介绍：

服务器角色备注
172.20.20.231loki-read-1loki读节点1
172.20.20.232loki-read-2loki读节点2
172.20.20.233loki-write-1loki写节点1
172.20.20.234loki-write-2loki写节点2
172.20.20.235promtail-1 nginxpromtail日志收集1，nginx服务器1
172.20.20.236promtail-2 nginxpromtail日志收集2，nginx服务器2
172.20.20.164minio-1存储1
172.20.20.165minio-2存储2
172.20.20.161nginx反代

loki下载地址：https://github.com/grafana/loki/tags

loki版本很重要，这里一定要重视，loki和promtail版本一定要相同，loki从2.4以后各版本之间配置文件还是有点差别的，所以这里坑我先踩为敬，版本很重要

如果你的环境中nginx一天的日志量没有超过100G，可以不使用这种方式，loki单节点可以扛住100G/天的日志量，但我这里的环境一天的日志量大概在300G左右，并且还在不断增加，所以多节点已经是必须的了。

网上看了很多关于loki分布式部署的文档，包括官方文档在内，没有一个可以把多节点部署讲清楚的，这个东西目前用的人不多，官方文档对关键配置也是一笑而过，这里我也没有把这个东西完全吃透，所以先上个我已经部署起来正在使用的配置来做个记录

从网上偷个图，loki读写分离的架构图，不要被那些组件吓到了

首先单节点的部署方式之前已经写过了，可以在本站找一下，下面直接上配置

1、loki配置

读节点配置：

auth_enabled: false
#读节点
target: read
​
http_prefix:
​
server:
  http_listen_address: 0.0.0.0
  grpc_listen_address: 0.0.0.0
  http_listen_port: 3100
  grpc_listen_port: 9095
  log_level: info
​
common:
  storage:
    s3:
      endpoint: 172.20.20.165:9000
      insecure: true
      bucketnames: loki
      access_key_id: Admin
      secret_access_key: 1234#abcd
      s3forcepathstyle: true
  compactor_address: http://172.20.20.161:3100
​
memberlist:
  join_members: ["172.20.20.231", "172.20.20.232", "172.20.20.233", "172.20.20.234"]
  dead_node_reclaim_time: 30s
  gossip_to_dead_nodes_time: 15s
  left_ingesters_timeout: 30s
  bind_addr: ['0.0.0.0']
  bind_port: 7946
  gossip_interval: 2s
​
ingester:
  lifecycler:
    join_after: 10s
    observe_period: 5s
    ring:
      replication_factor: 1
      kvstore:
        store: memberlist
    final_sleep: 0s
  chunk_idle_period: 1m
  wal:
    enabled: true
    dir: /data/loki/wal
  max_chunk_age: 1m
  chunk_retain_period: 30s
  chunk_encoding: snappy
  chunk_target_size: 1.572864e+06
  chunk_block_size: 262144
  flush_op_timeout: 10s
​
schema_config:
  configs:
  - from: 2020-08-01
    store: boltdb-shipper
    object_store: s3
    schema: v11
    index:
      prefix: index_
      period: 24h
​
storage_config:
  boltdb_shipper:
    shared_store: s3
    active_index_directory: /data/loki/index
    cache_location: /data/loki/tmp/boltdb-cache
​
​
limits_config:
  max_cache_freshness_per_query: '10m'
  enforce_metric_name: false
  reject_old_samples: true
  reject_old_samples_max_age: 30m
  ingestion_rate_mb: 10
  ingestion_burst_size_mb: 20
  # parallelize queries in 15min intervals
  split_queries_by_interval: 15m
​
chunk_store_config:
  max_look_back_period: 336h
​
table_manager:
  retention_deletes_enabled: true
  retention_period: 336h
​
query_range:
  # make queries more cache-able by aligning them with their step intervals
  align_queries_with_step: true
  max_retries: 5
  parallelise_shardable_queries: true
  cache_results: true
​
frontend:
  log_queries_longer_than: 5s
  compress_responses: true
  max_outstanding_per_tenant: 2048
​
query_scheduler:
  max_outstanding_requests_per_tenant: 1024
​
querier:
  query_ingesters_within: 2h
​
compactor:
  working_directory: /data/loki/tmp/compactor
  shared_store: s3

写节点配置：

auth_enabled: false
target: write
​
http_prefix:
​
server:
  http_listen_address: 0.0.0.0
  grpc_listen_address: 0.0.0.0
  http_listen_port: 3100
  grpc_listen_port: 9095
  log_level: info
​
common:
  storage:
    s3:
      endpoint: 172.20.20.165:9000
      insecure: true
      bucketnames: loki
      access_key_id: Admin
      secret_access_key: 1234#abcd
      s3forcepathstyle: true
  compactor_address: http://172.20.20.161:3100
​
memberlist:
  join_members: ["172.20.20.231", "172.20.20.232", "172.20.20.233", "172.20.20.234"]
  dead_node_reclaim_time: 30s
  gossip_to_dead_nodes_time: 15s
  left_ingesters_timeout: 30s
  bind_addr: ['0.0.0.0']
  bind_port: 7946
  gossip_interval: 2s
​
ingester:
  lifecycler:
    join_after: 10s
    observe_period: 5s
    ring:
      replication_factor: 1
      kvstore:
        store: memberlist
    final_sleep: 0s
  chunk_idle_period: 1m
  wal:
    enabled: true
    dir: /data/loki/wal
  max_chunk_age: 1m
  chunk_retain_period: 30s
  chunk_encoding: snappy
  chunk_target_size: 1.572864e+06
  chunk_block_size: 262144
  flush_op_timeout: 10s
​
schema_config:
  configs:
  - from: 2020-08-01
    store: boltdb-shipper
    object_store: s3
    schema: v11
    index:
      prefix: index_
      period: 24h
​
storage_config:
  boltdb_shipper:
    shared_store: s3
    active_index_directory: /data/loki/index
    cache_location: /data/loki/tmp/boltdb-cache
​
​
limits_config:
  max_cache_freshness_per_query: '10m'
  enforce_metric_name: false
  reject_old_samples: true
  reject_old_samples_max_age: 30m
  ingestion_rate_mb: 10
  ingestion_burst_size_mb: 20
  # parallelize queries in 15min intervals
  split_queries_by_interval: 15m
​
chunk_store_config:
  max_look_back_period: 336h
​
table_manager:
  retention_deletes_enabled: true
  retention_period: 336h
​
query_range:
  # make queries more cache-able by aligning them with their step intervals
  align_queries_with_step: true
  max_retries: 5
  parallelise_shardable_queries: true
  cache_results: true
​
frontend:
  log_queries_longer_than: 5s
  compress_responses: true
  max_outstanding_per_tenant: 2048
​
query_scheduler:
  max_outstanding_requests_per_tenant: 1024
​
querier:
  query_ingesters_within: 2h
​
compactor:
  working_directory: /data/loki/tmp/compactor
  shared_store: s3

读和写的配置文件区别就在这里：target: read和target: write

2、nginx代理配置
upstream read {
    server 172.20.20.231:3100;
    server 172.20.20.232:3100;
  }
​
  upstream write {
    server 172.20.20.233:3100;
    server 172.20.20.234:3100;
  }
​
  upstream cluster {
    server 172.20.20.231:3100;
    server 172.20.20.232:3100;
    server 172.20.20.233:3100;
    server 172.20.20.234:3100;
  }
​
  upstream query-frontend {
    server 172.20.20.231:3100;
    server 172.20.20.232:3100;
  }
​
  server {
    listen 80;
    listen 3100;
​
    location = /ring {
        proxy_pass       http://cluster$request_uri;
    }
​
    location = /memberlist {
        proxy_pass       http://cluster$request_uri;
    }
​
    location = /config {
        proxy_pass       http://cluster$request_uri;
    }
​
    location = /metrics {
        proxy_pass       http://cluster$request_uri;
    }
​
    location = /ready {
        proxy_pass       http://cluster$request_uri;
    }
​
    location = /loki/api/v1/push {
        proxy_pass       http://write$request_uri;
    }
​
    location = /loki/api/v1/tail {
        proxy_pass       http://read$request_uri;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";
    }
​
    location ~ /loki/api/.* {
        proxy_pass       http://query-frontend$request_uri;
    }
  }
3、promtail配置
server:
    http_listen_port: 9080
    grpc_listen_port: 0
​
positions:
    filename: /tmp/positions.yaml
​
#填写nginx地址
clients:
    - url: http://172.20.20.161:3100/loki/api/v1/push
​
scrape_configs:
    - job_name: nginx
      pipeline_stages:
      - replace:
          expression: '(?:[0-9]{1,3}\.){3}([0-9]{1,3})'
          replace: '***'
      static_configs:
      - targets:
         - localhost
        labels:
          job: nginx_access_log
          host: 20.235
          agent: promtail
          __path__: /data/openresty/nginx/logs/access.log
4、minio配置
#!/bin/bash
​
export MINIO_ACCESS_KEY=Admin
export MINIO_SECRET_KEY=1234#abcd
#export MINIO_PROMETHEUS_AUTH_TYPE=public #允许prometheus监控
/data/minio/minio server --config-dir /data/minio/config --console-address ":9001" \
http://172.20.20.165/data/minio/images \
http://172.20.20.165/data/minio/images2 \
http://172.20.20.166/data/minio/images \
http://172.20.20.166/data/minio/images2 \
5、grafana配置

grafana使用现成的，在添加loki数据源的时候填写nginx的地址

最后面一定要加/结尾

赞(1)
未经允许不得转载：娃哈哈好喝 » 多节点loki读写分离+minio+grafana实现nginx日志收集展示
分享到： 更多 (0)
标签：

相关推荐

热门标签

Linux (36)Nginx (22)Mysql (21)ELK (17)Shell (13)Zabbix (11)Docker (11)Python (10)Redis (10)Windows (9)Git (9)PHP (8)Prometheus (7)Loki (5)mac (5)Ansible (4)Jenkins (4)VMware (4)Saltstack (3)Canal (3)Svn (2)Kafka (2)Minio (2)Grafana (2)Kubernetes (1)Nacos (1)

网站统计

  • 日志总数：211
  • 标签总数：32
  • 页面总数：0
  • 分类总数：39
  • 链接总数：0
  • 最后更新：2024-11-04