nanomon is an extremely small and simple monitoring system meant largely for reporting on a few systems such as reporting failures on RAID arrays, ZFS volums, or file-systems don't fill up. It's meant to be similar to a simple "check" script that one may write, but with the additional features:
- Checks several times, so a transient failure does not result in an alert.
- Remembers state so that a failure doesn't generate mail with every check.
- Can check multiple issues from a single instance.
It's meant to be a much simpler system than things like OpsView (which we use in our data center), Zenoss, or Zabbix, without the learning-curve of nagios (which most of the others listed above use under the hood). nanomon was written after I had spent nearly a week trying to get the other solutions working.
nanomon uses "cron" or it's own scheduler for the scheduling, and is a single program executable. nanomon runs external programs (as specified in the config file), which can be simple commands or shell scripts and determines success or failure by either exit code, string match, or Python function (including regular expression match).
Probably the simplest way to deploy nanomon is to run it from cron every minute. However, you can also run it in "daemon" mode, which allows you to run the checks on a different schedule than every minute. This would allow you to run checks every 15 seconds or every 90 seconds, for example.
Note, however, that nanomon does not run checks in parallel. If you have 10 checks that each could take 10 seconds before timing out, the check frequency could be extended to 100 seconds.
The scheduler tries to start a round of checks "--interval" seconds after
the start of the previous set of checks. However, if the previous set of
checks took longer than "--interval", the checks will start after
"--min-interval" seconds (default 0) after the last set of checks finished.
"--min-interval" allows you to set a minimum interval between checks.
Download from:
Please report any bugs on the github Tracker
Simple example which calls Nagios NRPE scripts (a good source of scripts to
perform various checks, see "yum search nrpe" or "apt-cache search nrpe"
for a list of available packages).
statusfile('/var/lib/nanomon.status')
mailto('user@example.com')
mailfrom('nanomon@example.com')
# this check has a description, otherwise the basename of the
# check command is used ("check_zfs.sh" in this example).
command('/usr/lib/nagios/plugins/check_zfs.sh -p data',
success = re.compile(r'^OK:').match,
description = 'ZFS')
command('/usr/lib/nagios/plugins/check_mdstat.sh',
success = re.compile(r'^OK:').match)
command('/usr/lib/nagios/plugins/check_disk -c 10% -p /',
success = re.compile(r'^DISK OK ').match)
command('/usr/lib/nagios/plugins/check_disk -c 10% -p /boot',
success = re.compile(r'^DISK OK ').match)
# uncomment the following line to simulate a failure.
# this takes 15 runs of nanomon before an alert is sent.
#command('/bin/false', success = 0)
The following steps will install nanomon:
- Copy "
nanomon" to some location on the system, such as/usr/local/sbin - Create a "
/usr/local/etc/nanomon.conf" file. (This location can be changed by editing the top of "nanomon") - Add "nanomon" to cron, for example with:
echo '* * * * * root /usr/local/sbin/nanomon' >/etc/cron.d/nanomon
If you would like nanomon to re-alert periodically if a failure persists,
you can set up a cron job to run "nanomon reset", which acts like the
service has recovered (without sending an alert). So if it is still down,
this would cause another alert to be sent. However, if the service
recovers within 15 (default) minutes of the reset, no recovery will be
sent.
Running "nanomon --help" will display the help message:
Usage: nanomon [options] [status|reset|daemon]
A small service checking program, which can run an series of commands, check
their results, and report when they repeatedly fail, and when they recover.
Options:
-h, --help show this help message and exit
-c FILE, --config=FILE
Override the configuration file path
-i INTERVAL, --interval=INTERVAL
Daemon mode: How frequently in seconds checks are run
-m INTERVAL, --min-interval=INTERVAL
Daemon mode: Minimum time in seconds between checks
-p FILE, --pidfile=FILE
Daemon mode: File to write process ID into.
If called with the "reset" command, the status of all services is reset to to
the default state. If called with the "status" command, the state file is
loaded and either "OK" is printed and nanomon exits with 0, or information on
the failed services is printed and nanomon exits with 1. If called with no
command-line commands, a check of all services is run. Typically this is how
it is called from cron. If called with the "daemon" command, nanomon loops
running its own scheduler rather than running from cron. This is useful if
you want to run checks more frequently than every minute.
nanomon is licensed under the GPL v3 or later.