Média de Carga Aparentemente Inexplicada

Question

Eu tenho experimentado uma carga de servidor muito estranha, mas sem nenhuma razão óbvia. Alguém poderia explicar a causa / como depurar mais?

One Minute      - 22.9
Five Minutes    - 17.98
Fifteen Minutes - 10.02

top - 20:34:28 up 22 days, 7:51, 0 users, load average: 22.55, 22.49, 14.51
Tasks: 131 total,   3 running, 128 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.1%us,  0.0%sy,  0.0%ni, 98.6%id,  1.3%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   2097152k total,   596576k used,  1500576k free,        0k buffers
Swap:        0k total,        0k used,        0k free,        0k cached

 PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND            
11854 root      18   0  2444  980  720 R  2.0  0.0   0:00.01 top                
11856 root      18   0  2444  988  720 R  2.0  0.0   0:00.01 top                
   1 root      15   0  2156  592  564 S  0.0  0.0   0:10.26 init               
3393 apache    18   0 50276  33m 1888 S  0.0  1.6   0:00.00 httpd              
3445 pegpro    18   0 17872 3304 2368 D  0.0  0.2   0:00.03 php-cgi            
3446 root      18   0  5040 1056  852 S  0.0  0.1   0:00.00 crond              
3723 apache    15   0 50276  33m 1896 S  0.0  1.6   0:00.01 httpd              
3735 pegpro    18   0 17872 3308 2368 D  0.0  0.2   0:00.03 php-cgi            
3752 root      18   0  9152 2068 1740 S  0.0  0.1   0:00.01 dataskq            
3956 root      18   0  5040 1128  852 S  0.0  0.1   0:00.00 crond              
5138 root      18   0 20380  15m 1712 S  0.0  0.8   0:00.05 lfd                
5279 root      18   0  9152 2084 1752 S  0.0  0.1   0:00.05 dataskq            
5331 root      18   0  5040 1108  852 S  0.0  0.1   0:00.00 crond              
5496 admin     18   0 17872 3308 2368 D  0.0  0.2   0:00.01 php-cgi            
5637 root      18   0  9152 2080 1752 S  0.0  0.1   0:00.01 dataskq            
5641 apache    16   0 50276  33m 1896 S  0.0  1.6   0:00.03 httpd              
5648 root      18   0 49988  33m 2036 S  0.0  1.6   0:00.67 httpd              
5702 apache    18   0 50280  33m 1820 S  0.0  1.6   0:00.03 httpd              
5851 admin     18   0 17872 3304 2368 D  0.0  0.2   0:00.01 php-cgi            
7256 mail      16   0 10364 2700 2176 D  0.0  0.1   0:00.02 exim               
7287 apache    15   0 50276  33m 1876 S  0.0  1.6   0:00.00 httpd              
7379 root      18   0  5040 1128  860 S  0.0  0.1   0:00.02 crond              
7474 apache    16   0 50280  33m 1836 S  0.0  1.6   0:00.00 httpd 

One Minute      - 22.9
Five Minutes    - 17.98
Fifteen Minutes - 10.02

top - 20:34:28 up 22 days,  7:51,  0 users,  load average: 22.51, 22.49, 14.55
Tasks: 131 total,   3 running, 128 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.1%us,  0.0%sy,  0.0%ni, 98.6%id,  1.3%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   2097152k total,   596576k used,  1500576k free,        0k buffers
Swap:        0k total,        0k used,        0k free,        0k cached

 PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND            
11856 root      18   0  2444  988  720 R  2.0  0.0   0:00.01 top                
   1 root      15   0  2156  592  564 S  0.0  0.0   0:10.26 init               
3393 apache    18   0 50276  33m 1888 S  0.0  1.6   0:00.00 httpd              
3445 pegpro    18   0 17872 3304 2368 D  0.0  0.2   0:00.03 php-cgi            
3446 root      18   0  5040 1056  852 S  0.0  0.1   0:00.00 crond              
3723 apache    15   0 50276  33m 1896 S  0.0  1.6   0:00.01 httpd              
3735 pegpro    18   0 17872 3308 2368 D  0.0  0.2   0:00.03 php-cgi            
3752 root      18   0  9152 2068 1740 S  0.0  0.1   0:00.01 dataskq            
3956 root      18   0  5040 1128  852 S  0.0  0.1   0:00.00 crond              
5138 root      18   0 20380  15m 1712 S  0.0  0.8   0:00.05 lfd                
5279 root      18   0  9152 2084 1752 S  0.0  0.1   0:00.05 dataskq            
5331 root      18   0  5040 1108  852 S  0.0  0.1   0:00.00 crond              
5496 admin     18   0 17872 3308 2368 D  0.0  0.2   0:00.01 php-cgi            
5637 root      18   0  9152 2080 1752 S  0.0  0.1   0:00.01 dataskq            
5641 apache    16   0 50276  33m 1896 S  0.0  1.6   0:00.03 httpd              
5648 root      18   0 49988  33m 2036 S  0.0  1.6   0:00.67 httpd              
5702 apache    18   0 50280  33m 1820 S  0.0  1.6   0:00.03 httpd              
5851 admin     18   0 17872 3304 2368 D  0.0  0.2   0:00.01 php-cgi            
7256 mail      16   0 10364 2700 2176 D  0.0  0.1   0:00.02 exim               
7287 apache    15   0 50276  33m 1876 S  0.0  1.6   0:00.00 httpd              
7379 root      18   0  5040 1128  860 S  0.0  0.1   0:00.02 crond              
7474 apache    16   0 50280  33m 1836 S  0.0  1.6   0:00.00 httpd              
7550 apache    18   0 50276  33m 1924 S  0.0  1.6   0:00.00 httpd 

    

Answer 1

Se você observar os dois principais resultados, notará um número razoável de processos no estado (S) 'D'. Isso significa que eles estão esperando pelo disco io. Na maioria dos ambientes UNIX modernos, o número médio de carga é uma combinação de processos executáveis (aguardando CPU) e processos aguardando pelo disco io. Parece que o seu servidor provavelmente saturou o subsistema de IO disponível para a demanda colocada nele. Você pode verificar isso com ferramentas como iostat (tente adicionar os pacotes iostat ou sysstat em seu sistema). Então corra:

# iostat -x 1

e assista ao ocupado%