tload draws a scrolling graph of system load straight in your terminal, and one of its options has a bug you need to know about before you touch it. Get the graph running, tune its delay and scale, then stop it cleanly. About ten minutes, no elevated privileges.
This guide uses tload from procps-ng 4.0.4, provided here by procps package version 2:4.0.4-4ubuntu3.3. You need a shell and a terminal that can stay open while the graph updates. Everything below is an ordinary user command.
Confirm the binary and package before trusting the rest of this guide. A different version can mean a different bug list:
$ command -v tload
/usr/bin/tload
$ dpkg-query -W -f='${Package} ${Version}\n' procps
procps 2:4.0.4-4ubuntu3.3
$ tload --version
tload from procps-ng 4.0.4
The installed manual gives this command shape:
tload [options] [tty]
Different path or version reported? Read that machine's own manual with man tload before copying anything else here.
Run tload with no arguments in the terminal you want the graph in. It reads /proc/loadavg and redraws that terminal directly:
$ tload
The first line shows three changing numbers followed by the graph itself, something like 5.52, 4.54, 3.81 on this machine; yours will reflect your host's own activity. This is a view of current system load, not a process list and not a CPU percentage.
Checkpoint: Let it run a few seconds. If it does not fit or update the way you expect, stop it with Ctrl-C before touching any options. That drops you back to the shell without changing any system configuration.
Use -d or --delay to set the seconds between updates. For a quieter display, try two seconds:
$ tload --delay 2
For a faster view during a short investigation, one second is a reasonable start:
$ tload -d 1
-d 0 entirely. The installed procps-ng manual documents a specific bug: -d 0 sets an alarm of zero, so no signal ever arrives to redraw the display. It looks like a request for continuous updates and instead freezes it.-d 0? Press Ctrl-C and restart with -d 1 or another positive value.Use -s or --scale to set how many character rows sit between tick marks:
$ tload --scale 10 --delay 2
The scale works backwards from what the name suggests: a smaller number gives a larger graph, a larger number compresses it. Trace too compressed, go smaller. Trace running off the top of the terminal, go larger.
There is no universally correct value, only what fits your terminal height and the load range you care about. Change one setting at a time so you can tell what actually helped:
$ tload -s 5 -d 2
$ tload -s 20 -d 2
Each of those starts a fresh process once you stop the last one with Ctrl-C. There is no setting saved anywhere to undo.
The optional final argument is a TTY. With none given, tload draws in its own terminal. If you have a known writable terminal device and a real reason to display there, pass its path:
$ tload --delay 2 /dev/tty2
Replace /dev/tty2 with the actual target. This is not a file for saving a report: tload expects a terminal and writes the graph to it repeatedly. Check the device first:
$ test -c /dev/tty2 && echo 'character device exists'
character device exists
That only confirms the device type, not that you can write to it or that nobody else is using it. Do not reach for sudo to force it through: running as root can make an unintended terminal write possible without answering the actual access question. If access is denied, pick a terminal you own, or ask whoever administers the box which display is appropriate.
Confirm option spelling with built-in help:
$ tload --help
Usage: tload [options] [tty]
Options:
-d, --delay <secs> update delay in seconds
-s, --scale <num> vertical scale
-h, --help display this help and exit
-V, --version output version information and exit
"Unrecognised option" usually means it belongs to a different version than the one installed here, or the value is not attached to the flag correctly. A graph that looks frozen is almost always -d 0: stop it and restart with a positive delay.
For an independent, point-in-time comparison, pause or stop the graph and run uptime:
$ uptime
13:47:02 up 3 days, 4:12, 2 users, load average: 5.52, 4.54, 3.81
Time, uptime, users and the numbers are all host-specific. What matters is that the three load-average values appear in the same order as the ones at the top of the tload display. A mismatch usually just means the two commands ran a moment apart.
-d 0 breaks refresh on this version.-s if needed, remembering smaller means bigger.Ctrl-C with nothing left to undo.