mirror of
https://github.com/PDP-10/its.git
synced 2026-01-11 23:53:12 +00:00
89 lines
3.5 KiB
Plaintext
89 lines
3.5 KiB
Plaintext
METER is a program to display a variety of system statistics.
|
||
It differs from PEEK in that it shows only system-wide
|
||
information, not per-job. It also gives a lot of statistics
|
||
that PEEK won't.
|
||
|
||
On a display terminal (%TOERS and %TOMVU) METER will periodically
|
||
update its display, changing only those numbers that have
|
||
accidentally changed. On other terminals, it just types the
|
||
whole thing out over and over again.
|
||
|
||
METER does not take any options. It stops when you hit space.
|
||
If it is invoked by :METERD it will display on the 340.
|
||
|
||
If it is invoked by :SMETER, it will repeat slower.
|
||
|
||
The first display shows figures which for the most part are averages
|
||
over the whole time that the system has been up. After that, all
|
||
figures are averaged over the last 15 seconds or so.
|
||
|
||
The statistics displayed are explained below. All times are in seconds.
|
||
All numbers are [of course] decimal. "ATB" means average time between.
|
||
|
||
% User Time - fraction of CPU time given to users.
|
||
|
||
% Idle - fraction of CPU time not used.
|
||
|
||
% Overhead - fraction of CPU time spent on system overhead - page
|
||
fault handling, TTY and Disk I/O, scheduling, etc.
|
||
|
||
# Processes - the total number of processes (jobs, procedures)
|
||
on the system.
|
||
|
||
% active last 15 seconds - fraction of processes which ran at least
|
||
for a moment sometime within the last 15 seconds.
|
||
|
||
% Processes running - fraction of processes currently running
|
||
user programs.
|
||
|
||
% Exec run - fraction of processes currently running in the system.
|
||
|
||
% Blocked - fraction of processes currently waiting for something,
|
||
such as type in, a page to be swapped in, etc.
|
||
|
||
% Stopped - fraction of processes currently unrunnable - usually
|
||
this is due to ^Z, er, ah, CALL. Swap-blocked processes
|
||
are also counted here.
|
||
|
||
Disk ATB read - average time between disk reads for file accesses.
|
||
|
||
write - average time between disk writes for file accesses.
|
||
|
||
PgIn - average time between disk reads for swapping in pages.
|
||
|
||
PgOut - average time between disk writes for swapping out pages.
|
||
|
||
TTY ch/s in - characters per second input from TTY's.
|
||
|
||
TTY ch/s out - characters per second output to TTY's. This does not
|
||
include TV's, and does not count cursor motion accurately.
|
||
Since ITS does not keep this meter, the METER program
|
||
estimates it by looking at the teletype buffers frequently.
|
||
|
||
ATB Sched - average time between reschedulings.
|
||
|
||
ATB PCLSR - average time between attempts to force a process into
|
||
user mode so it can be stopped or take an interrupt.
|
||
There may be several attempts for each successful
|
||
PCLSR.
|
||
|
||
ATB swap out - average time between attempts to swap out a process.
|
||
|
||
ATB page faults - average time between page faults.
|
||
|
||
% in - fraction of page faults on pages already in core.
|
||
|
||
% coming - fraction of page faults on pages already coming into
|
||
core due to a previous page fault.
|
||
|
||
% out - fraction of page faults on pages not in core. These
|
||
are the "real" page faults.
|
||
|
||
% going - fraction of page faults on pages on their way out of
|
||
core. The page will have to come right back in.
|
||
|
||
The above four numbers usually don't add up to 100%.
|
||
The rest of the page faults are those which occurred
|
||
while a page-control lock was locked, and therefore
|
||
had to be taken over again after a delay.
|
||
|