Usage: ./monitord [OPTION]... CONFFILE

Options:
  --pid-file=<file>  store process ID to <file>
                     (default is to run without creating the pid file)

  --poll-interval=<int>  check the system after <int> second intervals
                         (default is 60 seconds)

  --keepalive-interval=<int>  send keepalives after <int> second intervals;
                              if <int> is not a multiple of the value given
                              with the --poll-interval option, it is rounded
                              down to the nearest multiple
                              (default is 300 seconds)

  --keepalive-message=<text>  use <text> for keepalive notification text
                              (default is "keepalive message")

  --notify-program=<prog>  all notifications are sent with program <prog>
                           (default is to write "application=%a object=%o 
                            severity=%s text=%t" to standard output)

  --severity-for-normal=<sev>  notifications for normal system conditions
                               (e.g., keepalives) have the severity <sev> 
                               (default is "normal")

  --max-stdout=<int>  take at most <int> lines from the standard output
                      of the monitoring script when forming a notification
                      text (default is 10)

  --program-name=<name>  use <name> for program name when logging to syslog
                         and sending notifications 
                         (default is ./monitord without the leading path)

  --syslog=<pri>  log program internal errors with syslog priority <pri>
                  (default is "user.err")

  --daemon | --nodaemon  run as a daemon | don't run as a daemon
                         (default is --nodaemon)

  --help, -?  print the usage information

  --version  print the version information

Notify program (see the --notify-program option) may contain the following 
variables which will be substituted with the real values when sending 
the notification:

%a - application that sent the notification (see the --program-name option)
%o - object that the notification is about, e.g., 'sendmail' or '/usr'
     (for keepalive messages, %o is set to keepalive interval)
%s - severity of the notification, e.g., 'minor' or 'major'
%t - notification text, e.g., 'Too few sendmail running', 
                              'Filesystem /usr: less than 10% free', etc. 

Note that before substitution all values will be put inside apostrophes,
and all apostrophes inside the values will be masked.


An example command line for starting monitord:

/usr/bin/monitord --daemon --poll-interval=60 --severity-for-normal=notice \
  --notify-prog='logger -i -t %a -p daemon.%s %t' /etc/monitord.conf


Configuration file directives (fields are separated by whitespace):

proc <regexp>  <min> <max>  <severity>  <resend time>
disk <mountpoint>  <disklimit> <diskreset>  <severity>  <resend time>
load <loadtype>  <loadlimit> <loadreset>  <severity>  <resend time>
exec <label> <progname> <option1> ... <optionN>  <severity>  <resend time>

	where <severity> = severity string (e.g., 'minor', 'major', etc.)
              <resend time> = if set to non-zero, notifications are resent 
                after every <resend time> seconds during the error condition.
                If set to 0, every notification is sent only once when
                the error condition is discovered.
              <regexp> = regular expression for matching process command lines
	      <disklimit> = integer followed by the optional % symbol
	      <diskreset> = integer followed by the optional % symbol. Reset
                value must be specified in the same units as the limit value.
	      <loadtype> = 1, 5, 15
	      <loadlimit> = float or integer  
	      <loadreset> = float or integer
	      <label> = unique string describing the nature of the script


An example configuration file:

# Issue a critical notification if there is less than 1 or more than 2 "ls -l"
# processes in the process table. If the number of "ls -l" processes becomes 
# 1 or 2, an "error clear" notification is issued (its severity can be set 
# with the --severity-for-normal option).

proc ls\s+-l 1 2 crit 0

# Issue a critical notification if there is no sshd process running. Continue 
# issuing critical notifications after every 30 seconds until sshd appears in
# the process table again. When sshd is running again, issue an "error clear"
# notification.

proc sshd 1 - crit 30

# Issue an error notification if there is less than 10% (10Mb) free space 
# in /tmp. If more than 30% (40Mb) of /tmp becomes free, an "error clear" 
# notification is issued. 

disk /tmp 10% 30% err 0
disk /tmp 10240 40960 err 0

# Issue a warning notification if 15 minute load exceeds 2.0. Repeat warning
# notifications after every 60 seconds as long as the load stays above 2.0. 
# If 15 minute load drops below 1.0, issue an "error clear" notification.

load 15 2.0 1.0 warning 60

# Issue a warning notification if the disk_monitor.sh script returns non-zero. 
# First N lines of the script's standard output (see the --max-stdout option)
# will be concatenated and the result is used as a notification text. After 
# the first non-zero is returned every different non-zero value returned by 
# disk_monitor.sh is considered a different error condition and a new warning 
# notification is produced. If the script finally returns 0, an "error clear"
# notification is issued.

exec diskmon disk_monitor.sh warning 0