Storm异常宕掉问题解决

问题描述

搭建Strom测试环境后，进程、UI界面一切正常，但奇怪的是，下次登录服务器重新检查进程发现整个Storm shut down，查看日志如下：

 

Nimbus.log

2017-09-15T15:06:16.690+0800o.a.s.z.ClientCnxn [INFO] Socket connection established to e3base1/172.21.8.64:11001,initiating session

2017-09-15T15:06:16.698+0800o.a.s.z.ClientCnxn [INFO] Session establishment complete on servere3base1/172.21.8.64:11001, sessionid = 0x15e5a6ab4630023, negotiated timeout =20000

2017-09-15T15:06:16.698+0800 o.a.s.c.f.s.ConnectionStateManager[INFO] State change: CONNECTED

2017-09-15T15:06:16.726+0800 b.s.d.nimbus[INFO] Starting Nimbus server...

2017-09-15T16:43:26.548+0800 b.s.d.nimbus[INFO] Shutting down master

2017-09-15T16:43:26.590+0800o.a.s.z.ZooKeeper [INFO] Session: 0x15e5a6ab4630023 closed

2017-09-15T16:43:26.590+0800o.a.s.z.ClientCnxn [INFO] EventThread shut down

2017-09-15T16:43:26.591+0800 b.s.d.nimbus[INFO] Shut down master

 

Supervisor.log

2017-09-15T15:01:22.921+0800o.a.s.z.ClientCnxn [INFO] Socket connection established toe3base1/172.21.8.64:11001, initiating session

2017-09-15T15:01:22.928+0800o.a.s.z.ClientCnxn [INFO] Session establishment complete on servere3base1/172.21.8.64:11001, sessionid = 0x15e5a6ab4630022, negotiated timeout =20000

2017-09-15T15:01:22.929+0800o.a.s.c.f.s.ConnectionStateManager [INFO] State change: CONNECTED

2017-09-15T15:01:23.001+0800b.s.d.supervisor [INFO] Starting supervisor with idabda54aa-1c8a-49d3-bf53-9db3d32ba5dd at host e3base3

2017-09-15T16:43:26.546+0800b.s.d.supervisor [INFO] Shutting down supervisorabda54aa-1c8a-49d3-bf53-9db3d32ba5dd

2017-09-15T16:43:26.551+0800 b.s.event[INFO] Event manager interrupted

2017-09-15T16:43:26.551+0800 b.s.event[INFO] Event manager interrupted

2017-09-15T16:43:26.563+0800o.a.s.z.ZooKeeper [INFO] Session: 0x15e5a6ab4630022 closed

2017-09-15T16:43:26.563+0800o.a.s.z.ClientCnxn [INFO] EventThread shut down

 

问题解决

分析了Storm日志、Zookeeper日志没有发现异常信息；

Storm启动采用如下方式：

storm nimbus > /dev/null 2>&1 &

storm supervisor > /dev/null 2>&1 &

 

尝试将启动方式修改为

nohup storm nimbus > /dev/null 2>&1 &

nohup storm supervisor > /dev/null 2>&1 &

 

修改为上述启动命令后问题解决。