你们python api用http连接es,连得都是协调节点吗,连多台他会自动负载均衡吧?
插入数据时协调节点为什么压力会那么大老内存溢出?数据难道会先加载到协调节点?
我连4台挂了2台, 4台机器都是配置64g,绝对超不过64g的,不知道是为什么挂的,可能是并发量太大了?
我插入4亿多条数据,用mapreduce,
if len(actions) == 9000:
for i in helpers.parallel_bulk(es, actions,
thread_count=3, chunk_size=3000):
[2017-12-16T22:05:43,000][WARN ][o.e.m.j.JvmGcMonitorService] [es_0 ] [gc][110776] overhead, spent [9.7s] collecting in the last [9.8s]
[2017-12-16T22:05:43,000][DEBUG][o.e.x.m.MonitoringService] [es_0 ] monitoring execution is skipped until previous execution terminated
[2017-12-16T22:05:43,000][DEBUG][i.n.b.PoolThreadCache ] Freed 10 thread-local buffer(s) from thread: threadDeathWatcher-2-1
[2017-12-16T22:05:43,000][ERROR][o.e.t.n.Netty4Utils ] fatal error on the network layer
at org.elasticsearch.transport.netty4.Netty4Utils.maybeDie(Netty4Utils.java:140)
at org.elasticsearch.http.netty4.Netty4HttpRequestHandler.exceptionCaught(Netty4HttpRequestHandler.java:81)
at io.netty.channel.AbstractChannelHandlerContext.invokeExceptionCaught(AbstractChannelHandlerContext.java:286)
at io.netty.channel.AbstractChannelHandlerContext.invokeExceptionCaught(AbstractChannelHandlerContext.java:265)
at io.netty.channel.AbstractChannelHandlerContext.fireExc