"TOP" command

[1raw] Statistics
 current time / working time / login user / working jobs before 1min(ave) / 5min / 15min
  >>It can read mean number (Load avarage) of the job awaiting practice as CPU utilization .


less than 1  -> comfortable(there is no waiting process)
more than 3  -> good
more than 5  -> some burden
more than 10 -> severe
more than 20 -> abnormal


*case 1 CPU


[2raw] number of process
 number of total processes / waiting processes / working processes / zombie processes / stoped processes


[3raw] Utilization
us -> utilization of user process
sy -> utilization of system process
ni -> utilization of priority process
id -> rate of system idol
wa -> waiting time of file IO
hi -> use time in the hardware IRQ
si -> use time in the software IRQ


[4raw] Utilization of memory
total -> total size
used  -> used size
free  -> free size
buffers -> buffer cache size


[5raw] Utilization fo swap
total -> total size
used  -> used size
free  -> free size
cached -> cache file size


[List]
PID  -> process id
USER -> user of process
PR   -> priority
NI   -> Nice level of task


 *The negative nice level means a high priority.


VIRT -> virtual memory size that a task uses.
RES  -> physical memory size which is not performed the swap that a task uses of
SHR  -> Shared memory size(kb) which a process uses
S    -> State of the process
  S: sleep
  T: stop
  D: impossible IRQ (sleep)
  N: nice value adds it
  < : nice value subtracks it
  R: executing
  Z: zombie
  W: swap out


%CPU  -> CPU utilization
%MEM  -> memory utilization
TIME+ -> execute time of the process(s)
COMMAND -> command name


*It can read Load avarage of the TOP command as CPU utilization , but am not suitable for graphs such as MRTG because numerical value may exceed 1.


*From a value of /proc/stat , It can calculate CPU utilization instead.


#cat /proc/stat
Result indication[user/nice/system/idle/iowait/irq/softirq/steal]


*I display time (Tick number ) that a CPU used since the OS starts according to a user when I execute the command mentioned above.
I carry it out repeatedly and can calculate CPU utilization when I divide the difference share by the total for the difference.


*CPU utilization = (user1 - user2) / (user1-user2 + nice1-nice2 + system1-system2 + idle1-idle2) which user used