4.2 KiB

Raw Blame History

Install ELK with beats to gather metrics

tags: openstack, ansible

About this repository

This set of playbooks will deploy elk cluster (Elasticsearch, Logstash, Kibana) with topbeat to gather metrics from hosts metrics to the ELK cluster.

These playbooks require Ansible 2.4+.

OpenStack-Ansible Integration

These playbooks can be used as standalone inventory or as an integrated part of an OpenStack-Ansible deployment. For a simple example of standalone inventory, see inventory.example.yml.

Optional | Load balancer VIP address

In order to use multi-node elasticsearch a loadbalancer is required. Haproxy can provide the load balancer functionality needed. The option internal_lb_vip_address is used as the endpoint (virtual IP address) services like Kibana will use when connecting to elasticsearch. If this option is omitted, the first node in the elasticsearch cluster will be used.

Optional | configure haproxy endpoints

Edit the /etc/openstack_deploy/user_variables.yml file and add fiel following lines

haproxy_extra_services:
 - service:
      haproxy_service_name: kibana
      haproxy_ssl: False
      haproxy_backend_nodes: "{{ groups['kibana'] | default([]) }}"
      haproxy_port: 81  # This is set using the "kibana_nginx_port" variable
      haproxy_balance_type: tcp
  - service:
      haproxy_service_name: elastic-logstash
      haproxy_ssl: False
      haproxy_backend_nodes: "{{ groups['elastic-logstash'] | default([]) }}"
      haproxy_port: 5044  # This is set using the "logstash_beat_input_port" variable
      haproxy_balance_type: tcp
  - service:
      haproxy_service_name: elastic-logstash
      haproxy_ssl: False
      haproxy_backend_nodes: "{{ groups['elastic-logstash'] | default([]) }}"
      haproxy_port: 9201  # This is set using the "elastic_hap_port" variable
      haproxy_check_port: 9200  # This is set using the "elastic_port" variable
      haproxy_backend_port: 9200  # This is set using the "elastic_port" variable
      haproxy_balance_type: tcp

Optional | run the haproxy-install playbook

cd /opt/openstack-ansible/playbooks/
openstack-ansible haproxy-install.yml --tags=haproxy-service-config

Deployment Process

Clone the elk-osa repo

cd /opt
git clone https://github.com/openstack/openstack-ansible-ops

Copy the env.d file into place

cd /opt/openstack-ansible-ops/elk_metrics_6x
cp env.d/elk.yml /etc/openstack_deploy/env.d/

Copy the conf.d file into place

cp conf.d/elk.yml /etc/openstack_deploy/conf.d/

In elk.yml, list your logging hosts under elastic-logstash_hosts to create the elasticsearch cluster in multiple containers and one logging host under kibana_hosts to create the kibana container

vi /etc/openstack_deploy/conf.d/elk.yml

Create the containers

cd /opt/openstack-ansible-playbooks
openstack-ansible lxc-containers-create.yml -e 'container_group=elastic-logstash:kibana'

install master/data elasticsearch nodes on the elastic-logstash containers

cd /opt/openstack-ansible-ops/elk_metrics_6x
openstack-ansible installElastic.yml

Install Logstash on all the elastic containers

cd /opt/openstack-ansible-ops/elk_metrics_6x
openstack-ansible installLogstash.yml

Install Kibana, nginx reverse proxy and metricbeat on the kibana container

cd /opt/openstack-ansible-ops/elk_metrics_6x
openstack-ansible installKibana.yml

Install Metricbeat everywhere to start shipping metrics to our logstash instances

cd /opt/openstack-ansible-ops/elk_metrics_6x
openstack-ansible installMetricbeat.yml

Trouble shooting

If everything goes bad, you can clean up with the following command

openstack-ansible lxc-containers-destroy.yml --limit=kibana:elastic-logstash_all

4.2 KiB Raw Blame History