In this lab we will install some monitoring. It's not a complete solution, we will make it better during next labs.
Install Prometheus node exporters on all your VMs. No need to configure them, default settings are good enough. Installation can be done in "init" role.
Install Prometheus on VM that doesn't have Agama running.
Configure Prometheus job "linux" with static_configs. Include all your VMs there, even future ones, don't delete job "prometheus" which comes with default config.
Hint: use Ansible variable "groups['all']".
It's not allowed to use IP addresses in Prometheus config, only names are allowed.
Install Nginx and configure reverse proxy on VM with prometheus:
/prometheus -> localhost:9090
Reuse existing nginx role.
Hint: use Ansible variable "groups['prometheus']" for condition in nginx config.
To make Prometheus reachable from outside, run it with
--web.external-url=http://<your_public_http_endpoint>/prometheus
Put required arguments in /etc/default/prometheus. Adjust metrics_path for job "prometheus" to fix prometheus monitoring after that.
Using docs write a query for memory consumption and average CPU load for each VM.
Save queries to prom_queries.txt, you will use them during next lab.
Your repository contains these files and directories:
ansible.cfg
group_vars/all.yaml
hosts
infra.yaml
prom_queries.txt
roles/bind/tasks/main.yaml
roles/prometheus/tasks/main.yaml
Your repository also contains all the required files from the previous labs.
Your repository does not contain Ansible Vault master password.
Prometheus and Node Exporters are installed and configured with this command:
ansible-playbook infra.yaml
Running the same command again does not make any changes to any of the managed hosts.
After playbook execution all targets at <your_VM_http_link>/prometheus/classic/targets should be UP: all VMs + prometheus itself.
After playbook execution you should be able to query historical data from Prometheus web interface <your_VM_http_link>/prometheus/graph.