How to Provision a Linux VM With Kibana Using Ansible

The corresponding source code for this post is available on GitHub.

Kibana is a fancy pants web application that tries to make data in Elasticsearch user-friendly. Rounding out the previous two posts on how to install an elasticsearch cluster and how to install multiple logstash hosts, I will now show you how to stack kibana on top of them.

Create the Ansible Role

Navigate to the directory you want the ansible role to reside and type:

$ molecule init role -d vagrant -r install-kibana

I'm using molecule to wrap vagrant and I'm calling this role install-kibana.

We will put kibana on a single host, 192.168.56.121. To make this happen, adjust the platforms section of your molecule/default/molecule.yml file to look like:

platforms:
  - name: kibana
    box: ubuntu/xenial64
    memory: 4096
    provider_raw_config_args:
    - "customize ['modifyvm', :id, '--uartmode1', 'disconnected']"
    interfaces:
    - auto_config: true
      network_name: private_network
      ip: 192.168.56.121
      type: static

We provision this local VM with 4GB of RAM. You can bring up this VM at this point with:

$ molecule create

First, we'll decide on the version of kibana we want to provision. To keep this compatible with the two posts mentioned above, we'll choose version 6.4.0, and update the vars/main.yml file to reflect the full name of the deb file we'll be grabbing in our playbook like so:

---
# vars file for install-kibana
kibana_deb_file: kibana-6.4.0-amd64.deb

Our tasks/main.yml file can now look like:

---
# tasks file for install-kibana
- name: download deb file
  get_url:
    dest: "/etc/{{ kibana_deb_file }}"
    url: "https://artifacts.elastic.co/downloads/kibana/{{ kibana_deb_file }}"
    checksum: "sha512:https://artifacts.elastic.co/downloads/kibana/{{ kibana_deb_file }}.sha512"
  become: yes

- name: install kibana from deb file
  apt:
    deb: "/etc/{{ kibana_deb_file }}"
    update_cache: yes
  become: yes

- name: send kibana config file
  template:
    dest: /etc/kibana/kibana.yml
    src: kibana.yml.j2
  become: yes
  notify: restart kibana

You can see we're using a handler that restarts kibana, which is in the handlers/main.yml file like so:

---
# handlers file for install-kibana
- name: restart kibana
  service:
    name: kibana
    state: restarted
  become: yes

Finally, we will have to create a templates/kibana.yml.j2 template, which as of right now is just a simple file:

elasticsearch.url: "http://192.168.56.102:9200"
server.host: 0.0.0.0

Keeping it as a template makes extending it and extracting variables as the role gets more involved possible.

You should now be able to run:

$ molecule converge

And the playbook will install kibana successfully. Once it comes up, you should navigate to http://192.168.56.121:5601, and you'll see kibana's home page. If you don't have any backing elasticsearch database, then you'll see an error, but bring on up at the port in the kibana.yml.j2 template and that error will go away.