#### 故障描述 启动机器上部署的服务时，会报错 Cannot allocate memory - fork(2) #### 故障分析 1. 检查dmesg日志 2. 检查message日志 3. 检查内存硬件错误 #### 日志信息 # less /var/log/message Aug 24 16:36:49 bjsjs_71_73 puppet-agent[74662]: Certificate Re quest fingerprint (SHA256): DF:AF:04:3E:66:F5:98:02:F2:17:11:19 :64:27:DC:F1:DE:15:49:14:27:3B:BD:1E:80:76:7D:C7:35:41:44:3A Aug 24 16:36:50 bjsjs_71_73 puppet-agent[74662]: Caching certif icate for ca Aug 24 16:37:26 bjsjs_71_73 puppet-agent[77199]: Caching certif icate for nm003.shopping.hadoop.sjs.ted Aug 24 16:37:26 bjsjs_71_73 puppet-agent[77199]: Caching certif icate_revocation_list for ca Aug 24 16:37:26 bjsjs_71_73 puppet-agent[77199]: Caching certif icate for nm003.shopping.hadoop.sjs.ted Aug 24 16:37:26 bjsjs_71_73 puppet-agent[77199]: Unable to fetc h my node definition, but the agent run will continue: Aug 24 16:37:26 bjsjs_71_73 puppet-agent[77199]: Error 400 on S ERVER: Cannot allocate memory - fork(2) Aug 24 16:37:26 bjsjs_71_73 puppet-agent[77199]: Retrieving plu ginfacts Aug 24 16:37:26 bjsjs_71_73 puppet-agent[77199]: Retrieving plu gin #### 结论 1. 这边检查时，无法复现。（未重复，故障消失） 2. 很有可能就是应用占用内存过多，没释放，导致新申请内存失败 3. 不排除某些内核参数设置不适合当前应用，导致内存充足，但应用申请内存失败