Skip to content

Latest commit

 

History

History
136 lines (97 loc) · 5.74 KB

README.md

File metadata and controls

136 lines (97 loc) · 5.74 KB
Raw

LoadBalanced-WebSite

Provision a load-balanced static HTML page whose metrics are collected by Prometheus & illustrated by Grafana on AWS using Terraform and Ansible.

Requirements

Name Version Notes
terraform >= 0.12
ansible >= 2.9
ansible community.grafana collection >= 1.2.1 To install it use: ansible-galaxy collection install community.grafana

Tested on

The role was tested in the following distributions on AWS.

Distribution Tested
Ubuntu 20.04 ✅

Usage

Quickstart

  • You need to have access to your AWS Account
  • Edit variables.tf file if you need to set/edit terraform variables. All variables have default value except aws_region and aws_profile.
  • Edit grafana variable file and set Garafana admin password
  • Optionally all configuration related variables can be changed from the according role's default folder

Here is an example to provision a load-balanced static HTML page whose metrics are collected by Prometheus & illustrated by Grafana on AWS using Terraform and Ansible:

$ terraform init
$ terraform plan
$ terraform apply
...
Apply complete! Resources: 18 added, 0 changed, 0 destroyed.
  • After provisioning infrastructure with terraform a new inventory file will be created in ansible folder
  • Use created inventory file to configure infrastructure with ansible:
$ ansible-playbook -i ansible/inventory ansible/site.yml

...
PLAY RECAP ***********************************************************************************************************************************************************
grafana1                   : ok=17   changed=11   unreachable=0    failed=0    skipped=0    rescued=0    ignored=0
loadbalancer1              : ok=31   changed=19   unreachable=0    failed=0    skipped=5    rescued=0    ignored=0
web_server1                : ok=19   changed=16   unreachable=0    failed=0    skipped=1    rescued=0    ignored=0
web_server2                : ok=19   changed=16   unreachable=0    failed=0    skipped=1    rescued=0    ignored=0

Now we can get the IP address of loadbalancer and grafana UI from inventory file. We can just open the ./ansible/inventory file or use bash tools like awk to get the required IP addresses:

$ awk '{print $1 "\t" $2}'  ./ansible/inventory

[webservers]
web_server1     ansible_host={WEB_SERVER1_IP_ADDRESS}
web_server2     ansible_host={WEB_SERVER2_IP_ADDRESS}

[loadbalancers]
loadbalancer1   ansible_host={LOAD_BALANCER_IP_ADDRESS}

[grafanaservers]
grafana1        ansible_host={GRAFANA_PROMETHEUS_STACK_IP_ADDRESS}

Try to access web site:

$ curl {LOAD_BALANCER_IP_ADDRESS}   

<html>
    <head>
        <title>Welcome to {WEB_SERVER1_IP_ADDRESS} !</title>
    </head>
    <body>
        <h1>The apache virtual is running! IP Address: {WEB_SERVER1_IP_ADDRESS}! </h1>
    </body>
</html>%


$ curl {LOAD_BALANCER_IP_ADDRESS}

<html>
    <head>
        <title>Welcome to {WEB_SERVER2_IP_ADDRESS}!</title>
    </head>
    <body>
        <h1>The apache virtual is running! IP Address: {WEB_SERVER2_IP_ADDRESS}</h1>
    </body>
</html>%

As we can see the HAProxy roundrobin configuration works!

In order to access Grafana UI and check dashboards and collected metrics open the grafana1 host {GRAFANA_PROMETHEUS_STACK_IP_ADDRESS} in browser and access via password which set in grafana variable file.

There will be 1 datasource (prometheus) and 2 dashboards (HAProxy and Node Exporter Full) created.

  • The HAProxy dashboard contains metrics from loadbalancer. The Relative time ranges and Refresh dashboard settings can be configured to see data.

  • The Node Exporter Full dashboard contains metrics from these servers:

    • loadbalancer1 ansible_host={LOAD_BALANCER_IP_ADDRESS}
    • web_server1 ansible_host={WEB_SERVER1_IP_ADDRESS}
    • web_server2 ansible_host={WEB_SERVER2_IP_ADDRESS}

Note: A bash code can be run to create traffic on loadbalancer and monitor:

  • while true;do curl http://{LOAD_BALANCER_IP_ADDRESS};done

Architecture

Architecture

Issues & Ideas for improvement

  • There is an issue with HAProxy dashboard for grafana. When trying to import it via ansible module it returns Datasource named ${DS_PROMETHEUS} was not found error message. In order to fix that I customized the template and uploaded to this repository: haproxy_rev7.json
  • Currently there is no any check for AMI ID. Checking availability of the provided ami-id by terraform vars based on region needs to be implemented
  • There is no any tests configured. Tests for ansible roles needs to be implemented
  • Currently terraform exposes required ports to internet. Connections between services can be done privately. Multiple VPCs and VPC Peering needs to be implemented
  • Currently there is no SSL enabled for the services. Certificates from private certificate authorities or non-profit certificate authorities like Lets Encrypt can be added

Credits

I've been inspired by & used the following open source projects: