推荐订阅源

1、Young GC导致集群master重新选举，一台server fail

[2016-12-10 07:38:24,546][WARN ][transport ] [BFRD_1] Received response for a request that has timed out, sent [239708ms] ago, timed out [119708ms] ago, action [discovery/zen/fd/ping], node [[BFRD_0][Z0Uq2P24TqGU073EpEosMQ][bfas-u-app-0218][inet[/10.20.2.18:9300]]], id [471700783]
[2016-12-10 07:51:18,521][WARN ][monitor.jvm ] [BFRD_1] [gc][young][390494][6365] duration [12.8m], collections [1]/[12.8m], total [12.8m]/[42.8m], memory [63.1gb]->[52.1gb]/[98.5gb], all_pools {[young] [11.4gb]->[116.5mb]/[12gb]}{[survivor] [1.1gb]->[324.8mb]/[1.5gb]}{[old] [50.5gb]->[51.6gb]/[85gb]}

相关配置如下，30s一次心跳ping，每次ping超时2m，尝试6次

discovery.zen.fd.ping_timeout: 120s
discovery.zen.fd.ping_retries: 6
discovery.zen.fd.ping_interval: 30s

2、一台server fail后，其余节点为啥写不进

3、集群启动过程中出现大量INITIALIZING、UNASSIGNED分片

{
  "cluster_name" : "elasticsearch",
  "status" : "yellow",
  "timed_out" : false,
  "number_of_nodes" : 4,
  "number_of_data_nodes" : 4,
  "active_primary_shards" : 60,
  "active_shards" : 116,
  "relocating_shards" : 0,
  "initializing_shards" : 6,
  "unassigned_shards" : 8
}