[Ericsson AB]

cpu_sup

MODULE

cpu_sup

MODULE SUMMARY

A CPU Load and CPU Utilization Supervisor Process

DESCRIPTION

cpu_sup is part of the os_mon application and all configuration parameters are defined in the reference documentation for the os_mon application.

cpu_sup is a process which supervises the CPU load and CPU utilization.

The load values are proportional to how long time a runnable UNIX process has to spend in the run queue before it is scheduled. Accordingly, higher values mean more system load. The returned value divided by 256 produces the figure displayed by rup and top. What is displayed as 2.00 in rup, is displayed as as load up to the second mark in xload.

For example, rup displays a load of 128 as 0.50, and 512 as 2.00.

If the user wants to view load values as percentages of machine capacity, then this way of measuring presents a problem, because the load values are not restricted to a fixed interval. In this case, the following simple mathematical transformation can produce the load value as a percentage:

PercentLoad = 100 * (1 - D/(D + Load))

D determines which load value should be associated with which percentage. Choosing D = 50 means that 128 is 60% load, 256 is 80%, 512 is 90%, and so on.

Another way of measuring system load is to divide the number of busy CPU cycles by the total number of CPU cycles. This produces values in the 0-100 range immediately. However, this method hides the fact that a machine can be more or less saturated. CPU utilization is therefore a better name than system load for this measure.

A server which receives just enough requests to never become idle will score a CPU utilization of 100%. If the server receives 50% more requests, it will still scores 100%. When the system load is calculated with the percentage formula shown previously, the load will increase from 80% to 87%.

The avg1/0, avg5/0, and avg15/0 functions can be used for retrieving system load values, and the util/0, and util/1 functions can be used for retrieving CPU utilization values. System load values can currently be retrieved on Solaris, Linux, FreeBSD, and OpenBSD. CPU utilization values can currently be retrieved on Solaris and Linux.

When run on Linux, cpu_sup assumes that the /proc file system is present and accessible by cpu_sup. If it is not, cpu_sup will fail.

EXPORTS

nprocs() -> UnixProcesses | {error, Reason}

Types:

UnixProcesses = integer()
Reason = term()

Returns the number of UNIX processes running on this machine. This is a crude way of measuring the system load, but it may be of interest in some cases.

avg1() -> SystemLoad | {error, Reason}

Types:

SystemLoad = integer()
Reason = term()

Returns the average system load in the last 60 seconds, as described above. 0 represents no load, 256 represents the load reported as 1.00 by rup.

avg5() -> SystemLoad | {error, Reason}

Types:

SystemLoad = integer()
Reason = term()

Returns the average system load from the last 300 seconds, as described above. 0 represents no load, 256 represents the load reported as 1.00 by rup.

avg15() -> SystemLoad | {error, Reason}

Types:

SystemLoad = integer()
Reason = term()

Returns the average system load from the last 900 seconds, as described above. 0 represents no load, 256 represents the load reported as 1.00 by rup.

util(ArgList) -> UtilSpec | {error, Reason}

Types:

ArgList = [Arg]
Arg = atom()
UtilSpec = UtilDesc | [UtilDesc]
UtilDesc = {Cpus, Busy, NonBusy, Misc}
Cpus = integer() | [integer()] | atom()
Busy = CpuStateDesc
NonBusy = CpuStateDesc
CpuStateDesc = float() | [{atom(), float()}]
Misc = [{atom(), term()}]
Reason = term()

Returns a CPU utilization specification of the CPU utilization since the last call to util/0 or util/1 by the calling process.

Note!

The returned value of the first call to util/0 or util/1 by a process will on most systems be the CPU utilization since system boot, but this is not guaranteed and the value should therefore be regarded as garbage. This also applies to the first call to util/0 or util/1 by a process after a restart of cpu_sup.

Currently recognized Arguments:

detailed
The returned UtilDesc(s) will be more detailed.
per_cpu
Each CPU will be specified separately (assuming this information can be retrieved from the operating system), i.e. one UtilDesc per CPU will be returned.

If the per_cpu has been passed as Argument, a list of UtilDescs will be returned; otherwise, only one UtilDesc will be returned.

Description of the UtilDesc:

Cpus
If detailed and/or per_cpu has been passed as arguments, this element will contain the CPU number, or a list of CPU numbers of the CPU or CPUs that the UtilDesc contains information about.
If neither detailed nor per_cpu has been passed as arguments, this field will contain the atom all which implies that the UtilDesc contains information about all CPUs.
Busy
If detailed has been passed as argument, this element will contain a list of {atom(), float()} tuples. Each tuple in the list contains information about a processor state that has been identified as a busy processor state. The first element is the name of the state, and the second element contains a float representing the percentage share of the CPU cycles spent in this state since the last call to util/0 or util/1.
If detailed hasn't been passed as argument, this element will contain the sum of the percentage shares of the CPU cycles spent in all states identified as busy.
If per_cpu hasn't been passed, the value(s) presented are the average of all CPUs.
NonBusy
The same as for the Busy element, but for processor states that has been identified as non-busy.
Misc
Miscellaneous information. Currently unused; reserved for future use.

Currently these processor states are identified as busy:

user
Executing code in user mode.
nice_user
Executing code in low priority (nice) user mode. This state is currently only identified on Linux.
kernel
Executing code in kernel mode.

Currently these processor states are identified as non-busy:

wait
Waiting. This state is currently only identified on Solaris.
idle
Idle.

Note!

Identified processor states may be different on different operation systems and may change between different versions of cpu_sup on the same operating system. The sum of the percentage shares of the CPU cycles spent in all busy and all non-busy processor states will always add up to 100%, though.

Failure: badarg if the ArgList is not a list of recognized Arguments.

util() -> CpuUtil | {error, Reason}

Types:

CpuUtil = float()
Reason = term()

Returns CPU utilization since the last call to util/0 or util/1 by the calling process.

Note!

The returned value of the first call to util/0 or util/1 by a process will on most systems be the CPU utilization since system boot, but this is not guaranteed and the value should therefore be regarded as garbage. This also applies to the first call to util/0 or util/1 by a process after a restart of cpu_sup.

The CPU utilization is defined as the sum of the percentage shares of the CPU cycles spent in all busy processor states (see util/1) in average on all CPUs.

See Also

os_mon(3)

AUTHORS

Janne Lindblad - support@erlang.ericsson.se
Rickard Green - support@erlang.ericsson.se

os_mon 1.8
Copyright © 1991-2005 Ericsson AB