Floating Point Math in Bash, Part 2 (Wait for System Load)

If you run scripts that require a lot of execution time it's a good idea to try to avoid letting them overload your system. You can run them via nice, but if for example your script is sending a bunch of emails your email daemon isn't running via nice and it may itself get out of control. One way to deal with this is by using the values in /proc/loadavg to pause when your system load gets too high.

The contents of /proc/loadavg look something like this:

  $ cat /proc/loadavg
  0.05 0.14 0.09 1/243 12667
The first three numbers are the load average over the last 1, 5, and 15 minutes. Load average being a measure of how busy the system is, the higher the load average the busier the system is. Specifically, load average is the number of processes in the system's run queue averaged over some time period, here 1, 5, or 15 minutes.

One slight problem that you'll notice is that the load average values are floating point values. This is where the floating point functions for bash that I wrote about come into play.

To wait for the load then we only need to extract the relevant value from /proc/loadavg then use float_cond described in the post I mentioned earlier to check if the load is above a certain point and pause if it is. We'll package this all up into its own file wait-for-load.sh so that it can be called from other scripts.

The command expects three command line arguments:

  • The load average to pause at, as a floating point value
  • The load average to restart at, as a floating point value
  • The time, in seconds, to pause between load checks
As it pauses it prints out a period each time it wakes up from a pause and prints the load every 10 times.

The source for the script follows:

#!/bin/bash

source float.sh

if [[ $# -ne 3 ]]; then
    echo "Usage: $0 STOP-AT-LOAD RESTART-AT-LOAD PAUSE-SECONDS"
    exit 1
fi

stop_at=$1
restart_at=$2
pause_seconds=$3


##############################################################
function get_load()
{
    local loadavg=$(cat /proc/loadavg | cut -d ' ' -f 1)
    echo $loadavg
}


##############################################################
load=$(get_load)

if float_cond "$load >= $stop_at"; then
    echo -n "Load is $load, pausing"
    n=0
    while float_cond "$load >= $restart_at"
    do
        sleep $pause_seconds
        echo -n .
        load=$(get_load)
        let n++
        if [[ $n -eq 10 ]]; then
            echo -n $load
            n=0
        fi
    done
    echo
fi

# vim: tabstop=4: shiftwidth=4: noexpandtab:
# kate: tab-width 4; indent-width 4; replace-tabs false;

A sample run of the script follows:

  $ sh wait-for-load.sh 2.0 1.0 10
  Load is 5.06, pausing..........1.49...
  $

Mitch Frazier is an embedded systems programmer at Emerson Electric Co. Mitch has been a contributor to and a friend of Linux Journal since the early 2000s.

Load Disqus comments