Load average in Linux的精确含义

Man 上的解释：

load average

System load averages is the average number of processes that are either in a runnable or uninterruptable state.  A process in a runnable state is either using the CPU or waiting to use the CPU.  A  process  in uninterruptable state is waiting for some I/O access, eg waiting for disk.  The averages are taken over the three time intervals.  Load averages are not normalized for the number of CPUs in a system, so a load average of 1 means a single CPU system is loaded all the time while on a 4 CPU system it means it was idle 75% of the time.

单从这点上可以看出有些资料说load average可以理解为cpu运行队列中的进程数或者解释为CPU的工作负载量是不靠谱的，只有在早期Unix上定义仅处于Running状态的平均进程数代表load average，这样的话才算是队列中的包含正在被服务和等待被服务的进程数总和（排队论）

我们还是通过下面两点深入和准确的了解下什么是load average：

1.为什么只统计进程状态是Running和Uninterruptible 

传统的OS进程有三种状态：Running、ready和block，而在Linux中把进程状态进行了细分扩展，ps man 上描述了如下状态及标记：

D    uninterruptible sleep (usually IO)

R    running or runnable (on run queue)

S    interruptible sleep (waiting for an event to complete)

T    stopped, either by a job control signal or because it is being traced.

W    paging (not valid since the 2.6.xx kernel)

X    dead (should never be seen)

Z    defunct ("zombie") process, terminated but not reaped by its parent.

R状态包括了传统划分的runnig跟ready两种状态，R状态的进程消耗CPU时间片，所以该状态的进程自然是系统的一种负载，那么D状态的Uninterruptible为什么也会被统计进来？

首先到底什么是interruptible和Uninterruptible进程，stackoverflow上有这样的解释：

• TASK_INTERRUPTIBLE, the interruptible sleep. If a task is marked with this flag, it is sleeping, but can be woken by signals. This means the code which marked the task as sleeping is expecting a possible signal, and after it wakes up will check for it and return from the system call. After the signal is handled, the system call can potentially be automatically restarted (and I won't go into details on how that works).

• TASK_UNINTERRUPTIBLE, the uninterruptible sleep. If a task is marked with this flag, it is not expecting to be woken up by anything other than whatever it is waiting for, either because it cannot easily be restarted, or because programs are expecting the system call to be atomic. This can also be used for sleeps known to be very short.

这两种状态的进程都是sleep状态的，区别在于可中断也就是S态的进程可以被其他系统信号中断唤醒，而不可中断D状态的进程不能被其他的任何信号中断唤醒除非等到它需要的资源被释放，或者说进程进行了一种被系统认为类似于原子操作的行为。

原子操作不可分割，所以进程处于D状态时不能被其他信号中断。例如进程在申请内存页，或者等待磁盘返回都是属于不可中断的状态。

也就是说D状态的进程通常都是申请使用系统中除CPU以外的资源，所以把D状态算进系统的load中也算是相当的合乎情理。所以说load average是针对CPU的负载情况体现是不正确，它代表的是系统中主要资源的整体负载情况。

2.load average的计算 

怎么来的？

$ cat /proc/loadavg 

0.14 0.05 0.06 1/122 13870

前面三个分别指的1/5/15分钟的load average，第四列数表示总共有多少线程以及当前运行着的线程数，最后一个是在该命令运行时最近或最后的线程ID

那么loadavg里面的数据又是怎么计算出来的呢？

wiki上说不同的系统load average计算有所不同，其中Linux上是这样计算的：

unsigned long avenrun[3]; 
 
static inline void calc_load(unsigned long ticks) 
{ 
   unsigned long active_tasks; /* fixed-point */ 
   static int count = LOAD_FREQ; 
 
   count -= ticks; 
   if (count < 0) { 
      count += LOAD_FREQ; 
      active_tasks = count_active_tasks(); 
      CALC_LOAD(avenrun[0], EXP_1, active_tasks); 
      CALC_LOAD(avenrun[1], EXP_5, active_tasks); 
      CALC_LOAD(avenrun[2], EXP_15, active_tasks); 
   } 
}

#define FSHIFT   11 /* nr of bits of precision */ 
#define FIXED_1  (1<<FSHIFT) /* 1.0 as fixed-point */ 
#define LOAD_FREQ (5*HZ) /* 5 sec intervals */ 
#define EXP_1  1884 /* 1/exp(5sec/1min) as fixed-point */ 
#define EXP_5  2014 /* 1/exp(5sec/5min) */ 
#define EXP_15 2037 /* 1/exp(5sec/15min) */ 
 
#define CALC_LOAD(load,exp,n) \ 
   load *= exp; \ 
   load += n*(FIXED_1-exp); \ 
   load >>= FSHIFT;

这一系列的计算代表着在前n分钟处于Running和Uninterruptible进程总数的指数移动平均数。

好像有点难理解，其实主要目的就是解决数值大幅抖动造成的平均数计算误差，得到一个更平滑的接近最多真实情况的平均数值。

我们可以从业务角度这样理解：

For example, one can interpret a load average of "1.73 0.60 7.98" on a single-CPU system as:

• during the last minute, the system was overloaded by 73% on average (1.73 runnable processes, so that 0.73 processes had to wait for a turn for a single CPU system on average).

• during the last 5 minutes, the CPU was idling 40% of the time on average.

• during the last 15 minutes, the system was overloaded 698% on average (7.98 runnable processes, so that 6.98 processes had to wait for a turn for a single CPU system on average).

This means that this system (CPU, disk, memory, etc.) could have handled all of the work scheduled for the last minute if it were 1.73 times as fast.

In a system with four CPUs, a load average of 3.73 would indicate that there were, on average, 3.73 processes ready to run, and each one could be scheduled into a CPU

下面使用高速公路收费排队的模型更能形象的解释load average的数值：

= load of 1.00

  = load of 0.50

 = load of 1.70

Reference ：

http://en.wikipedia.org/wiki/Load_(computing)

http://en.wikipedia.org/wiki/Exponential_smoothing