Automating implementation and reducing time to deploy complex environments is key. In this story, I am planning to get one of the environments that fairly used in the industry to map NFS FS over multiple subnets. This is a very basic configuration but complexity starts when you wanted to use the same template for deploying the entire application in one go.
I am using the Terraform template function to achieve this. I am certainly can use “Ansible” or “Chef” or any other tool but I wanted to make it relatively simple and have things done by just using a single input file.
I am creating a single EFS FS that will be part of a given region and will have a single mount target in that AZ. I am planning to use a maximum of 3 AZ in this document. AZ count can be increased in case needed for more redundancy.
Single instance started in each AZ and mounted newly created EFS using local IP. Internet gateway attached so that my local environment I could be able to access instances to check EFS is working fine.
Parameter store used to get a “keypair” name.
Download source code for this implementation from Github page —
Download main.tf, terraform.tfvars and user_data_import.tpl file
user_data_import.tpl is user_data template file. You can add or modify any commands you like to execute during boot time. Mainly I am using this file to mount newly created EFS FS automatically on EC2 instance.
New EFS name is part of the input and UNIX mountpoint is also part of the input. If VPC and subnet already created and wanted to use same subnet make sure to add the “data” block in main.tf accordingly and change “EFS” and “instance” block accordingly.
Please change localip parameter to your own domain subnet ip from where you need ssh access to each EC2 instance. Do not use default 0.0.0.0/0 which opens port 22 for all world.
Execute Terraform job
To execute terraform job please download terraform file and entier following commands.
Please review terraform documentation for more information. You can send your questions as well.
This job will create total of 32 resources. Const be very minimum if you will use the attached configuration and upon testing perform the cleanup task.
Output “efsip” are EFS IP for each Availability Zone. Since I am working on the first 3 availability zone, I did assign 3 IP for inter AZ communication. “instance_public_ip(typo)” is an IP address for each instance that I created in given AZ. I will use this public ip to connect to each EC2 instance.
Verify FS is mounted successfully. Each instance used its own EFS IP from AZ to connect. EFS is mounted successfully.
Perform Read/Write test from each instance. I am creating new file from one of the instance and the file is visible from other two instances.
Tags are added as per EFS FS in case needed for local scripting purposes.
Elastic Filesystem Configuration
EFS fs is created with 3 mount point
Access point to used mount FS as “/” this can be easily changed as per need.
FS is part of 3 Availability zone and each availability zone has a different IP address.
To cleanup enter following command