This role installs and configures Fluentbit.

• This role is only tested with Ansible >= 2.9.

This role tries to keep the same default configuration as if you manually install Fluentbit. You can find more information about each option in the Fluentbit documentation.

Variables used for the installation:

Variables used for the general configuration:

For fluentbit_svc_http, each key is used as a configuration option name and values as values. But you don't need to add the prefix HTTP_, it will be added by the template. For example, you can define it like this:

fluentbit_svc_http:
  server: On
  listen: 0.0.0.0
  port: "{{ fluentbit_monitoring_port }}"

It's the same for fluentbit_svc_storage: you don't need to specify the prefix storage.. Example:

fluentbit_svc_storage:
  path: /var/log/flb-storage/
  sync: full
  checksum: "off"

Other variables:

Variables for log processing:

In most cases, you should not modify these variables. Templating is used to build these lists with other variables.

• _fluentbit_inputs will aggregate all the variables whose name matches this regex: ^fluentbit_.+_input(s)?$'.
• _fluentbit_filters will aggregate all the variables whose name matches this regex: ^fluentbit_.+_filter(s)?$'.
• _fluentbit_outputs will aggregate all the variables whose name matches this regex: ^fluentbit_.+_output(s)?$'.
• _fluentbit_parsers will aggregate all the variables whose name matches this regex: ^fluentbit_.+_parser(s)?$'.
• _fluentbit_outputs will aggregate all the variables whose name matches this regex: ^fluentbit_.+_mlparser(s)?$'.

Each variables matching these regexes must be:

• a dictionary defining one input/filter/output/parser or
• a list of dictionaries defining one or more inputs/filters/outputs/parser.

Each dictionary is used to define one [INPUT], [FILTER], [OUTPUT], [PARSER] or [MULTILINE_PARSER] section in the Fluentbit configuration file or in managed parser file. Each configuration section is configured with key/value couples, so the dictionary's keys are used as configuration keys and values as values.

For example:

fluentbit_nginx_input:
  - name: tail
    path: /var/log/nginx/access.log
  - name: tail
    path: /var/log/nginx/error.log

fluentbit_kernel_input:
  name: tail
  path: /var/log/kern.log

will create:

[INPUT]
  name tail
  path /var/log/nginx/access.log
[INPUT]
  name tail
  path /var/log/nginx/error.log
[INPUT]
  name tail
  path /var/log/kern.log

It allows you to define variables in multiple group_vars and cumulate them for hosts in multiples groups without the need to rewrite the complete list.

Ansible dictionaries can't have the same key multiple times. This can be a problem to use stuff like record. To overcome this issue, each key can be prefix with a number that will be remove in the template file. It needs to match this regex: '^[0-9]+__'.

Fir example:

fluentbit_add_context_filter:
  - name: record_modifier
    match: '*'
    0__record: hostname ${HOSTNAME}
    1__record: product Awesome_Tool

will create:

[FILTER]
  name record_modifier
  match *
  record hostname ${HOSTNAME}
  record product Awesome_Tool

None

in group_vars/all.yml:

fluentbit_kernel_input:
  name: tail
  path: /var/log/kern.log

fluentbit_env_filter
  name: record_modifier
  match: '*'
  record: "env {{ env }}"

fluentbit_central_output:
  name: forward
  match: '*'
  host: "{{ logs_server_address }}"
  port: "{{ log_forward_port }}"

in group_vars/web.yml:

fluentbit_nginx_input:
  - name: tail
    path: /var/log/nginx/access.log
  - name: tail
    path: /var/log/nginx/error.log

in playbook.yml:

- hosts: web
  gather_facts: True
  become: yes
  roles:
    - bimdata.fluentbit

BSD

BIMData.io