Bash Script for Nagios to Check Status of Docker Container

check_docker_container.sh

#!/bin/bash

# Author: Erik Kristensen
# Email: [email protected]
# License: MIT
# Nagios Usage: check_nrpe!check_docker_container!_container_id_
# Usage: ./check_docker_container.sh _container_id_
#
# Depending on your docker configuration, root might be required. If your nrpe user has rights
# to talk to the docker daemon, then root is not required. This is why root privileges are not
# checked.
#
# The script checks if a container is running.
#   OK - running
#   WARNING - restarting
#   CRITICAL - stopped
#   UNKNOWN - does not exist
#
# CHANGELOG - March 20, 2017
#  - Removes Ghost State Check, Checks for Restarting State, Properly finds the Networking IP addresses
#  - Returns unknown (exit code 3) if docker binary is missing, unable to talk to the daemon, or if container id is missing

CONTAINER=$1

if [ "x${CONTAINER}" == "x" ]; then
  echo "UNKNOWN - Container ID or Friendly Name Required"
  exit 3
fi

if [ "x$(which docker)" == "x" ]; then
  echo "UNKNOWN - Missing docker binary"
  exit 3
fi

docker info > /dev/null 2>&1
if [ $? -ne 0 ]; then
  echo "UNKNOWN - Unable to talk to the docker daemon"
  exit 3
fi

RUNNING=$(docker inspect --format="{{.State.Running}}" $CONTAINER 2> /dev/null)

if [ $? -eq 1 ]; then
  echo "UNKNOWN - $CONTAINER does not exist."
  exit 3
fi

if [ "$RUNNING" == "false" ]; then
  echo "CRITICAL - $CONTAINER is not running."
  exit 2
fi

RESTARTING=$(docker inspect --format="{{.State.Restarting}}" $CONTAINER)

if [ "$RESTARTING" == "true" ]; then
  echo "WARNING - $CONTAINER state is restarting."
  exit 1
fi

STARTED=$(docker inspect --format="{{.State.StartedAt}}" $CONTAINER)
NETWORK=$(docker inspect --format="{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}" $CONTAINER)

echo "OK - $CONTAINER is running. IP: $NETWORK, StartedAt: $STARTED"

I get the same error as michabb, (Docker version 1.9.1, build a34a1d5, on Ubuntu, Trusty Tahr 14.04.4 LTS), but when I issue the docker inspect command for the container I am using, the State.Ghost field does not exist. So I expect this is:
a. a State property that is just not in the container
b. a State property that is not included in the docker version
Neither give me any worries. But it would be nice to know which it is :)

I thought off option c: the property is only added in specific circumstances, that would be strange though, imho.

Ow the properties I do have are:
"State": {
"Status": "running",
"Running": true,
"Paused": false,
"Restarting": false,
"OOMKilled": false,
"Dead": false,
"Pid": 5667,
"ExitCode": 0,
"Error": "",
"StartedAt": "2016-02-23T11:44:17.338327578Z",
"FinishedAt": "0001-01-01T00:00:00Z"
},

A bit of a newbie in docker, but maybe Dead is a replacement for ghost, have to read more changelogs and manual pages.

Awesome, it helped me to write my own script. Good job.

since this comment, I've adopted this script with two modifications:

1. It returns status UNKNOWN if the docker command is missing (instead of OK)
2. I removed all the GHOST check code

hash docker 2>/dev/null || { echo "UNKNOWN - docker command not found"; exit 3; }

first of thanks for sharing.
Small comment:
Guys please be sure that when you run above script from nrpe i.e as nagios unprivileged user you have sudo rights for

docker inspect

command.
Lost like an hour to discover why the scriprt didn't work through nrpe.

Added something like to top
# permissions
if [ "$(whoami)" != "root" ]; then
    echo "Root privileges are required to run this, try running with sudo..."
    exit 2
fi

Thanks man, after 3 years it's still not outdated! 👍

If you are using Docker 1.12+ and using the IP driver (direct,macvlan, etc) -- the way to get the IP address is:

NETWORK=$(docker inspect --format="{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}" $CONTAINER)

That also works on the default networked containers, so it's probably a better way.

A slight improvement:

docker ps -q --filter "name=nginx" | awk '{print $1}' | xargs docker inspect --format="{{ .State.Status }}"

other ps filter options:
(https://docs.docker.com/v1.11/engine/reference/commandline/ps/)
and inspect additional features :
(https://docs.docker.com/v1.11/engine/reference/commandline/inspect/)

Thanks a lot!!

2> /dev/null is what I needed after hours of searching. 🙇‍♂️
Thank you!

I wasn't getting notifications on this! My apologies.

I've updated the script with most of the suggestions in the comments.

Please note, I'm not using this script anymore, but if needed I'll move this to a git repo so pull requests can be accepted.

Hello guys.
Added nagios user to docker group, so it has permissions to speak to docker daemon.
When I execute it as nagios:

nagios@nrpe-client-host:~$ sh /usr/lib/nagios/plugins/check_docker_container.sh redis
/usr/lib/nagios/plugins/check_docker_container.sh: 25: [: xredis: unexpected operator
/usr/lib/nagios/plugins/check_docker_container.sh: 30: [: x/usr/bin/docker: unexpected operator
/usr/lib/nagios/plugins/check_docker_container.sh: 48: [: true: unexpected operator
/usr/lib/nagios/plugins/check_docker_container.sh: 55: [: false: unexpected operator
OK - redis is running. IP: 172.17.0.2, StartedAt: 2018-03-01T08:07:42.857992735Z

So, it works.
But when I'm trying to execute it from nagios host:

user@nagios-host:~$ /usr/local/nagios/libexec/check_nrpe -H 123.45.67.89 -c check_docker_container redis
NRPE: Unable to read output

Nagios displays the same "UNKNOWN - NRPE: Unable to read output"

Here is nrpe.cfg
command[check_docker_container]=/usr/lib/nagios/plugins/check_docker_container.sh
and service definition

define service {
        use                             generic-service
        host_name                       host-example
        service_description             Redis Docker Container
        check_command                   check_nrpe!check_docker_container!redis
}

Am I missing something? What's wrong?
Thank you!

Just solved my issue.
It was about permissions, sorry.
Great script, btw. Thanks.

@wirtoo can you share the permission issue fix you used?

I'm running NRPE in a container. Do I need to add the nagios user to /etc/sudoers in the container itself?
From my Nagios host:

./check_nrpe -H 10.99.125.131 -c check_docker_container1
NRPE: Unable to read output

What I do wrong?

Remote
/usr/local/nagios/libexec/check_nrpe -H hostip -c check_docker -a asterisk
UNKNOWN - Missing docker binary

Local
/usr/lib64/nagios/plugins/check_docker asterisk
OK - asterisk is running. IP: 172.19.0.2, StartedAt: 2018-09-14T06:44:09.174409454Z

I understand this might be outdated, I mean this thread not the script. Take a look here and it may help some of you who are having permission problems. Nagios and Docker Monitoring

Awesome script, thank you, just what I needed!

HI,

i need docker stat output with mail alert shell script. Please help on this

Thank you for sharing this!

Thank you Erik Kristensen

Thanks a lot, this is also working for PRTG (with some small changes in the output).