Explore

ASG + Load balancer

Implementing auto scaling group on memzo's production server to ensure the high availability, reliability and optimized billing.

Coda templates

💡 Purpose

Implementation of Auto Scaling Group with Load Balancer can ensure the reliability of system. This implementation will target to auto scale the production server of Memzo with Staging Database on some other domain/sub-domain.

🛤 Steps

Memzo Instance AMI Creation

Test/Staging DB Setup

Launch Instance and Change DB config

Ensure if DB is functioning correctly using Dashboard

Create New AMI for Launch

Configured Launch Template for ASG (with custom user data)

Configure Auto Scaling Group

Add Network Load Balancer & Target Group

Adding ASG Policies and Ensuring execution

⁠

Testing Scaling instances using 'stress'

Setting up Elastic IP with Network Load Balancer

Creating SSL for some domain (Ex:

memzo.in⁠

)

Configuring SSL in Load Balancer

Accessing using Domain Name and testing ASG Activities

❓ Difficulties Identified

1. When instance is added by ASG, Apache httpd service is stopped by default so need to configured launch template again to add User Data (Startup Script) when instance is being launched. Ref : instance Config -> Advance options -> user data

https://awstip.com/troubleshooting-ec2-startup-scripts-passed-as-user-data-618ed1e59edf⁠

⁠

2. Application load balancer doesn't support Elastic IP. So tried with Network Load Balancer but It created a loop of https request redirections. First solution is to use application load balancer - Ref :

https://serverfault.com/questions/983955/apache-redirect-loop-when-rewriting-http-to-https-behind-aws-network-load-balanc⁠

⁠

Another solution is - to change .htaccess configurations.

https://stackoverflow.com/questions/42715718/aws-load-balancer-err-too-many-redirects⁠

3. If there are crons running on server then Auto Scaling can lead to a state of Clashing Crons that can create challenges. First solution is to Keep crons on other Server. Another solution is to Allow only first instance of ASg to run crons. Ref: https://serverfault.com/questions/727466/running-crontab-on-aws-auto-scaling-group

https://gist.github.com/kixorz/5209217⁠

4. If Server is Using EBS volumes then ASG will replicate the EBS volumes for each instance and will clear all storage when the instance is terminated. To attach an EBS Volume to the instance is tricky in Auto Scaling Group. Solution - Decoupling Compute from Storage. Ref : https://portworx.com/wp-content/uploads/2017/05/asg-slides.pdf Another solution is to keep only one EBS volume and attach the same volume to instance while the instance is launching using the Advanced user-data in Launch Template configurations. Ex:

https://stackoverflow.com/questions/60898996/attach-ebs-volume-to-autoscalinggroup⁠

⁠

5. Practically, when the instance will be terminated the volume will stay attached/in-use until the instance terminated successfully.

Meanwhile, New instance is being launched by ASG and user-data will find volume is attached/in-use by another instance, and it’ll create a new volume.

To solve this. Analyse Auto Scaling Group lifecycle hooks.

6. EBS Attchment is possible for the instances of same availability zone.

EBS Scaling Ref :

https://paytm.com/blog/engineering/aws-ebs-auto-scaling/⁠

⁠

Create AMI of Current Instance

Launch Instance using AMI -

Consider following changes in launch server:

Database configuration with Staging

Remove Crons

Change .htaccess redirections. Related to force http → https

Create New AMI

Modify Current Volumes attached with Live Memzo Instance

Memzo Instance id -

i-0907f030bbdac357a⁠

⁠

Volume id -

vol-0a4a3c25a0ad65ea8⁠

(100 GB)

Volume id -

vol-078cd20aa401269f7⁠

(50 GB)

Keep only one volume with enough required storage.

Also, check availability zone of EBS volume. As per the AWS policy volume available in speific zone can be atttached to multiple instances of same zone only.

Launch Template

Use above AMI image in Launch template and keep desired instance type (t2.small in our case), security group, etc.

Dont add any EBS volume in launch template. We’ll attach it using advance user-data of instance.

User-data to Add in Launch Template Configuration

#!/bin/bash

sudo -i

service httpd start

$instanceId = Invoke-WebRequest -Uri http://169.254.169.254/latest/meta-data/instance-id

aws ec2 attach-volume --volume-id vol-0966b056db6a9e3f5 --instance-id $instanceId --device /dev/xvda

EBS Volume can be attach to instances available in same availability zone.

⁠

https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ebs-volumes-multi.html⁠