The cron daemon automatically starts when a system is booted and keeps running in the background. Daemon process is known as crond and is started by systemd or the init process, depending upon what your system has. Its task is to check configuration files regularly at one minute intervals and check whether any tasks are to be completed.

Cron configuration contains files and directories where the Cron jobs to be scheduled are written. They are available in the /etc/ directory. The most important file associated with cron configuration is crontab. In a Linux system, configuration files related to cron are as follows:

Scripts can be directly added into any of the directories such as cron.hourly/, cron.daily/, cron.weekly/, or cron.monthly/, in order to run them at an hourly, daily, weekly, or monthly basis respectively.

The following is a simple shell script firefox_memcheck.sh, which checks whether a Firefox process is running or not. If Firefox is running and its memory usage is greater than 30 percent, then restart Firefox:

#!/bin/sh
# Filename: firefox_memcheck.sh
# Desription: Resatrts application firefix if memory usage is more than 30%

pid='pidof firefox' # Get pid of firefox
if [ $pid -gt 1 ]
then
  # Get current memory usage of firefox
  current_mem_usage='ps -u --pid $pid| tail -n1 | tr -s ' ' | cut -d ' ' -f 4'
  # Check if firefox memory  usage is more than 30% or not
  if [ $(echo "$current_mem_usage > 30" | bc) -eq 1 ]
  then
    kill $pid   # Kill firefox if memory usage is > 30%
    firefox &   # Launch firefox
  fi
fi

We can add this script into the /etc/cron.hourly/ directory of the system and it will keep checking our Firefox memory usage. This script can be modified to monitor the memory usage for other processes too.

By putting scripts into cron.{hourly, daily, weekly, monthly}, we can only set tasks at an interval of an hour, day, week, and month. What if a task has to run at 2-day intervals, 10-day intervals, 90 minute intervals, and so on? To achieve this, we can add tasks into the /etc/crontab file or the /etc/cron.d/ directory. Each user may have their own crontab entry and files related to each users are available in /var/spool/.

A crontab entry looks as follows:

We can see from the preceding screenshot that a crontab entry has five asterisks. Each asterisk defines a specific duration. We can replace * with a value suggested against each of them or leave it as it is. If * is mentioned in a field, then it means consider all the instances of that field.

The timing syntax can also be described as follows:

All five fields are separated by blank spaces. It is followed by a username that specifies by which user the command will be executed. Specifying the username is optional and by default it is run as a root. The last field is command that is scheduled for execution.

An example demonstrating how to write the crontab entry is as follows:

20 7 * * 0 foo command

Each field can be explained as follows:

So, the command will run as root at 7:20 AM every Sunday.

We can specify multiple instances of a filed using a comma (,):

30 20,22 * * * command

Here, command will run at 8:30 PM and 10:30 PM every day.

We can also specify a range of time in a field using a hyphen (-) as follows:

35 7-11 * * 0-3 command

This means, the run command is at 7:35, 8:35, 9:35, 10:35, and 11:35 on Sunday, Monday, Tuesday, and Wednesday.

To run a script at a specific interval, we can specify the forward slash (/) as follows:

20-45/4 8 9 4 * command

The command will run on 9th April between 8:20 AM to 8:45 AM at an interval of 4 minutes.