elasticsearch 压测优化后的配置

来源:互联网 发布:linux程序权限 编辑:程序博客网 时间:2024/06/06 23:55

配置说明

一、JVM配置:

Jdk1.8es要求最低jdk1.7

最大堆最小堆均为32G

GC类型:-XX:+UseG1Gc -XX:MaxGCPauseMillis=200 

其他默认。

二、Es集群配置:

 

cluster.name: es-test

node.name: "test-169-71"

 

node.master: false

node.data: true

 

index.store.type: niofs                                读写文件方式

index.cache.field.type: soft                         缓存类型

 

bootstrap.mlockall: true                             禁用swap

 

gateway.type: local                                    本地存储

 

gateway.recover_after_nodes: 3                   3个数据节点开始恢复

 

gateway.recover_after_time: 5m                  5分钟后开始恢复数据

 

gateway.expected_nodes: 4                         4es节点开始恢复

 

cluster.routing.allocation.node_initial_primaries_recoveries:8           并发恢复分片数

cluster.routing.allocation.node_concurrent_recoveries:2            同时recovery并发数

 

indices.recovery.max_bytes_per_sec: 250mb                            数据在节点间传输最大带宽

indices.recovery.concurrent_streams: 8                             同时读取数据文件流线程

 

 

discovery.zen.ping.multicast.enabled: false                        禁用多播

discovery.zen.ping.unicast.hosts:["192.168.169.51:9300", "192.168.169.52:9300"]

 

discovery.zen.fd.ping_interval: 10s                     节点间存活检测间隔

discovery.zen.fd.ping_timeout: 120s                   存活超时时间

discovery.zen.fd.ping_retries: 6                          存活超时重试次数

 

http.cors.enabled: true                                       使用监控

 

index.analysis.analyzer.ik.type:"ik"                   ik分词

 

#####thread pool setting#################

threadpool.index.type: fixed                              写索引线程池类型

threadpool.index.size: 64                                   线程池大小(建议2~3cpu数)

threadpool.index.queue_size: 1000                            队列大小

 

threadpool.search.size: 64                                  搜索线程池大小

threadpool.search.type: fixed                             搜索线程池类型

threadpool.search.queue_size: 1000                    队列大小

 

threadpool.get.type: fixed                                  取数据线程池类型

threadpool.get.size: 32                                      取数据线程池大小

threadpool.get.queue_size: 1000                         队列大小

 

threadpool.bulk.type: fixed                                批量请求线程池类型

threadpool.bulk.size: 32                                    批量请求线程池大小

threadpool.bulk.queue_size: 1000                       队列大小

 

threadpool.flush.type: fixed                               刷磁盘线程池类型

threadpool.flush.size: 32                                          刷磁盘线程池大小

threadpool.flush.queue_size: 1000                      队列大小

 

#indices.store.throttle.type: merge

indices.store.throttle.type: none                                 写磁盘类型

indices.store.throttle.max_bytes_per_sec:500mb        写磁盘最大带宽

 

index.merge.scheduler.max_thread_count: 8               索引merge最大线程数

index.translog.flush_threshold_size:600MB               刷新translog文件阀值

原创粉丝点击