深入理解Linux的CPU

版权声明：本文为博主原创文章，未经博主允许不得转载。 https://blog.csdn.net/u012679583/article/details/82458247

前言

在分析linux机器上运行的负载特性时，常常需要分析很多的metrics。影响负载运行效率的关键因素中CPU是非常重要的一个。这里针对promethues中提取的有关cpu的metrics进行详细分析。同时也是对linux CPU的深入理解。

---

1.CPU利用率的理解

在linux系统中，CPU执行时分为用户态、系统态和空闲态。分别表示cpu处于用户态执行的时间、系统态执行的时间和空闲系统进程执行的时间。cpu利用率即是cpu进程非空闲进程的时间/cpu总执行时间
也可以表达成：
1-cpu空闲运行时间/cpu总运行时间

2.如何计算CPU利用率

可以在/proc/stat文件来计算cpu利用率。该文件包含了所有cpu活动的信息，所有值都是从系统启动开始到当前时刻

root@ccrfox111:/home/wychen# cat /proc/stat 
cpu  19482732 658887 17810123 4863711023 2257224 0 2502875 0 0 0
cpu0 1080572 36868 1022322 150931201 24236 0 320696 0 0 0
cpu1 970443 39283 896713 151127028 76293 0 139106 0 0 0
cpu2 885327 38633 924001 151120244 194046 0 95816 0 0 0
cpu3 809150 35663 824706 151368080 95386 0 85577 0 0 0
cpu4 888204 18108 853836 151196320 213321 0 88810 0 0 0
cpu5 813351 17197 760230 151469481 80663 0 83352 0 0 0
cpu6 903539 15295 839289 151242659 158490 0 87598 0 0 0
cpu7 810276 15995 752474 151496058 67115 0 81945 0 0 0
cpu8 881509 15523 827310 151344019 108021 0 86076 0 0 0
cpu9 801434 15916 745196 151520620 62655 0 80216 0 0 0
cpu10 900545 15076 811303 151365504 89333 0 83428 0 0 0
cpu11 801766 15720 741800 151525518 63670 0 79729 0 0 0
cpu12 870335 15164 801125 151402061 98221 0 80436 0 0 0
cpu13 799814 18229 741643 151529017 61832 0 78541 0 0 0
cpu14 883848 15634 796733 151428426 60969 0 81238 0 0 0
cpu15 783835 16051 737709 151551629 61513 0 78285 0 0 0
cpu16 367533 15365 284016 152582422 88945 0 54388 0 0 0
cpu17 324896 23586 309820 152657977 20958 0 60294 0 0 0
cpu18 416391 15450 280204 152504237 123288 0 50465 0 0 0
cpu19 320234 22779 300225 152682350 15508 0 60266 0 0 0
cpu20 406168 15375 285095 152503376 135212 0 45795 0 0 0
cpu21 323230 23777 313614 152663375 18832 0 58529 0 0 0
cpu22 385040 15693 291949 152546759 108834 0 46004 0 0 0
cpu23 327446 23554 322367 152650941 17094 0 59102 0 0 0
cpu24 364650 16359 278396 152637007 50923 0 45874 0 0 0
cpu25 316525 24195 307784 152673564 17915 0 60456 0 0 0
cpu26 362492 15379 273206 152670445 25924 0 45247 0 0 0
cpu27 323518 23521 307272 152671054 14859 0 61765 0 0 0
cpu28 351997 15438 277955 152659331 40648 0 50222 0 0 0
cpu29 322048 24629 312710 152665213 16160 0 61441 0 0 0
cpu30 366290 15614 279102 152653788 30606 0 51069 0 0 0
cpu31 320314 23805 310003 152671306 15738 0 61098 0 0 0
intr 3489535069 43 0 0 0 0 0 0 0 1 3 246 0 0 0 0 0 0 0 97 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 110547 320210 1365104 362686 1432413 352945 1118622 294995 682894 274136 648019 296554 580118 284050 372673 268098 792928 41528 1089936 44986 1060673 51181 1079085 39617 657784 62386 257702 39265 457135 80618 0 189382 36987 0 0 0 1498493 0 0 0 11027489 3486077 6199271 3816255 2364228 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
ctxt 10149973761
btime 1534667648
processes 8987149
procs_running 1
procs_blocked 0
softirq 4946136337 6 583266925 313944 3037525697 15491661 0 3221168 531837767 0 774479169

对参数的解释如下：
每一列参数值的含义
user 从系统启动开始累积到当前时刻，用户态的cpu时间
nice 从系统启动开始累积到当前时刻，nice值为负的进程所占用的cpu时间
system 从系统启动开始累积到当前时刻， 核心时间
idle 从系统启动开始累积到当前时刻，除硬盘IO等待时间以外的其他等待时间
iowait 从系统启动开始累积到当前时刻，硬盘IO等待时间
irq 从系统启动开始累积到当前时刻，硬中断时间
softirq 从系统启动开始累积到当前时刻，软中断时间
注：最后三列000不知道什么意思 有待研究
CPU时间=user+system+nice+idle+iowait+irq+softirq

“intr”这行给出中断的信息，第一个为自系统启动以来，发生的所有的中断的次数；然后每个数对应一个特定的中断自系统启动以来所发生的次数。
“ctxt”给出了自系统启动以来CPU发生的上下文交换的次数。
“btime”给出了从系统启动到现在为止的时间，单位为秒。
“processes (total_forks) 自系统启动以来所创建的任务的个数目。
“procs_running”：当前运行队列的任务的数目。
“procs_blocked”：当前被阻塞的任务的数目。

那么CPU利用率可以使用以下两个方法。先取两个采样点，然后计算其差值：
cpu usage=(idle2-idle1)/(total_2 - total_1)*100
cpu usage=[(user_2 +sys_2+nice_2) - (user_1 + sys_1+nice_1)]/(total_2 - total_1)*100

3.CPU每个核的测量

我锁在的课题组分配的服务器是有32个CPU核心的，因此在监控的时候显示的是32个cpu的统计数目。

参考

https://blog.csdn.net/Alisa_xf/article/details/71430406
https://blog.csdn.net/baidu_35692628/article/details/71773886