mohammed6688 / zbx-docker

Zabbix UserParams and scripts that support discovery and monitoring of docker swarm nodes and services.

Docker Zabbix Userparams

Simple scripts and userparameter conf to monitor Docker Swarm cluster status, Swarm service status and docker containers.

Introduction

Sometimes you may want to run an agent check that does not come predefined with Zabbix. This is where user parameters come to help.

Docker Userparams is a simple set of scripts that supports docker container discovery and attr introspection.

With this solution, you can start to monitor your docker containers on your Servers.

Aditionally, you can monitor your swarm cluster nodes and services status.

Requirements

For now, the Zabbix User should have access to the docker socket (in linux usually under /var/run/docekr.sock).

To use docker swarm mode monitoring the target where the userparameters are installed need to be a manager of the cluster.

Installation - Standard

To install this solution you'll need to do the following steps:

1. Install latest Zabbix agent (https://www.zabbix.com/documentation/3.2/manual/installation/install_from_packages/repository_installation)

2. Download the latest version of the repo into the "/usr/local/share/zabbix/swarm-monitoring" folder using the following command (or similar):

wget https://github.com/mohammed6688/zbx-docker.git -O /usr/local/share/zabbix/swarm-monitoring/zbx-docker.tar 

3. Extract the .tar file:

cd /usr/local/share/zabbix/swarm-monitoring/
tar xf /usr/local/share/zabbix/swarm-monitoring/zbx-docker.tar
cd ./zbx-docker

4. Install the UserParameters available at the 'userparameters'.

mv ./userparameters/ /etc/zabbix/zabbix_agentd.d/

5. change the permission of each script to be executable

chmod 755 ./scripts/*

6. move scripts into the following dir

mv ./scripts/ /usr/local/share/zabbix/

4. Install the templates available at the 'templates' folder.

5. Link the Templates 'Docker Server' against every machine that you want to monitor the docker Containers;

6. Link the Template 'Docker Swarm Manager' against a MANAGER of your cluster;

After some minutes (10 by default) new hosts will be created at the Docker Containers Group when using the 'Docker Server' Template.

For the 'Docker Swarm Manager' template the Nodes and Services will be created at the groups 'Docker Cluster Nodes' and 'Docker Nodes Services'.

LLD - Values returned by each Script

docker_discovery.py

This script can discover all containers created in the given machine. Returned LLD info:

• {#CONTAINER}: return the container ID, which will be mapped to hostname on zabbix;
• {#CONTNAME}: return the container name, which will be mapped to visible hostname on zabbix;
• {#IMAGE}: return the image used by container;
• {#LOGPATH}: return the log path (usually a json file if you arent using a logger like graylog or fluentd to receive logs);
• {#STATE}: the state of the container in the moment of discovery ('running', 'dead', 'exit')

docker_cluster_discovery.py

This script can discovery all nodes and services on a docker swarm mode cluster. Returned LLD info:

• When using --resource nodes:

  • {#NODE_NAME}: Hostname of the node in the swarm cluster;
  • {#NODE_ID}: The ID given to the node.

• When using --resource services:

  • {#SERVICE_ID}: The id of the service in the cluster (as you can see with docker service ls);
  • {#SERVICE_NAME}: The name given to the service;
  • {#SERVICE_HTTPSUPPORT}: If you service have a label called 'com.df.serviceDomain', the url's are returned here and can be used, for example, to populate macros and web scenarios.

Other Notes

• If you are using the Windows 2016 Server with docker support, you can still use this solution; however, you'll need to install python and docker-py package and use the scripts instead..

• You can change the address that will be used to query about containers and discover trough zabbix interface adding a parameter on the discovery or a 3rd parameter in the info, but per default zabbix will not allow unsecure caracters (like @, / and other needed on the URL). In this case, you can create a bash script that wraps the py modules and that passes the --addr parameter with the correct URL of your API Endpoint.