DELL 服务器Kernel 提示Package power limit notification 的解决方法

love_china

IDC机房内有客户反应CPU温度过高，高达81度，但是dell 服务器没有报警，机房室内温度正常， 机身温度也正常。
显示器显示：

1. CPU13: Package power limit notification (total events = 1)
   
2. CPU1: Package power limit notification (total events = 1)
   
3. CPU17: Package power limit notification (total events = 1)
   
4. CPU3: Package power limit notification (total events = 1)
   
5. CPU19: Package power limit notification (total events = 1)
   
6. CPU5: Package power limit notification (total events = 1)
   
7. CPU21: Package power limit notification (total events = 1)
   
8. CPU7: Package power limit notification (total events = 1)
   
9. CPU23: Package power limit notification (total events = 1)
   
10. CPU9: Package power limit notification (total events = 1)
    
11. CPU25: Package power limit notification (total events = 1)
    
12. CPU11: Package power limit notification (total events = 1)
    
13. CPU27: Package power limit notification (total events = 1)
    
14. CPU31: Package power limit notification (total events = 1)
    
15. CPU15: Package power limit notification (total events = 1)
    
16. CPU29: Package power limit notification (total events = 1)
    
17. CPU17: Core power limit normal
    
18. CPU1: Core power limit normal
    
19. CPU3: Core power limit normal
    
20. CPU19: Core power limit normal
    
21. CPU21: Core power limit normal
    
22. CPU5: Core power limit normal
    
23. CPU23: Core power limit normal
    
24. CPU7: Core power limit normal

复制代码

明显很不正常。通过查资料发现，DELL R620 和R720都有这个毛病。

给出的信息是 CPU 电源限制通知，这些并不是什么错误，当CPU处于低负载的时候，BIOS会控制CPU进行核心频率调节，或关闭某些核心和模块来达到省电的目的，这些功能源自于Intel C-State Tech 技术，AMD也有类似的省电技术。
12G机型的默认电源管理配置是优化性能功耗比 (DAPC模式 DELL主动电源控制器) 来对性能和功耗进行一个均衡处理，并保持一个较低的功耗水平(单路E5 2620 16G内存 4*300G 10K 2.5 SAS 闲置约46W)，要知道在一个数据中心和机房使用此模式，每年节约的电力是相当客观的。


解决方法：
进入BIOS中，进入system profile settings，第一个选项默认是performance per watt(DAPC)，选择该选项后，默认是开启Turbo boost的，在system profile中选择最后一个custom手动模式，剩下的其他选项就可以设置了，只要修改Turbo boost为disable，其余选项均不变即可，保存后重启机器。