AWS UniLink deployment guide

The following is a reference HA architecture guide for deploying UniLink on AWS. For a cloud-native deployment, use our helm charts with AWS Elastic Kubernetes Service(EKS).

Introduction

We will use the Linux installation script to get a chatwoot instance running. Also instead of relying on Redis, Postgres and Nginx installed in the same ec2; we will proceed to make use of managed AWS services for the same viz Elasticache, RDS, and ALB.

Prerequisites

1. AWS account
2. Domain to use with UniLink

Architecture

This guide will follow a standard 3-tier architecture on AWS.

Network

Create VPC

1. Sign in to the AWS console and pick the region you will deploy.
2. Navigate to the VPC console and create a new vpc for chatwoot. At the name tag, enter chatwoot-vpc and use the CIDR block 10.0.0.0/16.
3. Leave the rest of the options as default and click on Create VPC.

Subnets

Create two public and private subnets in the vpc we created. Make sure to have them in different AZ's and have non-overlapping CIDR ranges.

1. Navigate to VPC > Subnets.

2. Click on Create Subnet. Select the chatwoot-vpc we created before, name it as chatwoot-public-1, select an availability zone (for example, ap-south-1a), and the CIDR block as 10.0.0.0/24.

3. Follow the same to create the remaining subnets.

Name	Type	Availability Zone	CIDR Block
chatwoot-public-1	public	ap-south-1a	10.0.0.0/24
chatwoot-public-2	public	ap-south-1b	10.0.1.0/24
chatwoot-private-1	private	ap-south-1a	10.0.2.0/24
chatwoot-private-2	private	ap-south-1b	10.0.3.0/24

4. After creating all subnets, enable auto assign public ipv4 address for public subnets under Actions > Subnet Settings.

Internet Gateway

1. Select Create Internet Gateway , name it chatwoot-igw, and click create.
2. Select it from the internet gateways list, choose actions and then select Attach to VPC.
3. Choose chatwoot-vpc and click attach.

NAT Gateway

UniLink app servers need to be deployed in the private subnet. For them to access the internet, we need to add NAT gateways to our public subnet and add a route from the private subnets.

1. Navigate the VPC dashboard and select NAT gateways.

2. Click Create NAT Gateway.

   1. Name it chatwoot-nat-1.
   2. Select the chatwoot-public-1 subnet.
   3. Click on Allocate Elastic IP.
   4. Add additional tags as per your need.
   5. Click Create NAT gateway.

3. Follow the same to create a second NAT gateway (chatwoot-nat-2) and choose the chatwoot-public-2 subnet.

Route tables

The route table controls the inbound and outbound access for a subnet.

Public Route table

We will create route tables so that our public subnets can reach the internet via the Internet gateway.

Navigate to the VPC dashboard and select Route Tables.

1. Click Create route table.
2. Use the name chatwoot-public-rt and choose chatwoot-vpc under VPC.
3. Click Create.

Next, we need to add a route to the internet gateway we created earlier(chatwoot-igw).

1. Select the chatwoot-public-rt route table from the list and click on Edit routes > Add Route.
2. Set the destination as 0.0.0.0/0 and choose the target as chatwoot-igw. Click on Save Changes.

Also,

1. Select the chatwoot-public-rt route table from the list and click on Subnet Associations > Edit subnet associations.
2. Select both the public subnets(chatwoot-public-1,chatwoot-public-2) and click save.

Private Route table

We will also create private route tables so that our private subnets can reach the internet via the NAT gateways.

1. Follow the above guide and create two private route tables, namely, chatwoot-private-a and chatwoot-private-b.
2. Select the route tables and add a route to the NAT gateway in their respective availability zone.
   1. For chatwoot-private-a, add a route to 0.0.0.0/0 and target as chatwoot-nat-1.
   2. For chatwoot-private-b, add a route to 0.0.0.0/0 and target as chatwoot-nat-2.

Also,

1. Associate the private route tables with corresponding private subnets.
   1. For chatwoot-private-a, associate chatwoot-private-1 subnet.
   2. For chatwoot-private-b, associate chatwoot-private-2 subnet.

Application Load Balancer (ALB)

Create an application load balancer to receive traffic on port 80 and 443 and distribute it across UniLink instances.

1. Navigate to the EC2 section and choose the Load Balancer section.
2. Click Create Load Balancer.
   1. Choose Application Load Balancer.
   2. For the load balancer name, use chatwoot-loadbalancer.
   3. Select the scheme as internet-facing and IP address type as IPv4.
   4. For the network mapping section,
      1. Select chatwoot-vpc.
      2. Select the public subnets chatwoot-public-1 and chatwoot-public-2 under the mapping section.
   5. For the Security group section,
      1. Create a new security group, chatwoot-loadbalancer-sg.
      2. Add rules to allow HTTP and HTTPS traffic from anywhere(0.0.0.0/0, ::/0).
      3. Also, add a rule to allow TCP on port 3000. This rule allows the load balancer health checks to pass since UniLink runs on port 3000.
      4. Add a rule to allow SSH traffic from the bastion security group we will create at the latter stage of the guide. Revisit this section after completing the bastion step.
   6. For the Listeners and routing section, create two listeners for 80 and 443. Set the forwarding rule on listener 80 to redirect http to https.
      1. Also, create a target group, chatwoot-tg, that will forward the requests to port 3000(UniLink listens on this port).
      2. Add a health check to the endpoint /api. This endpoint is not authenticated and should return the application version.

{
  "version": "1.22.1",
  "timestamp": "2021-12-06 16:07:39"
}

7. Add any necessary tags and click create.

Also, add if you have your domain on Route53 and use ACM to generate a certificate to use with ALB.

Postgresql using AWS RDS

UniLink uses Postgres as a DB layer, and we will use Amazon RDS with a multi-AZ option for reliability.

RDS security group

1. Navigate to EC2 > Security groups and create a new sg.
2. Name it chatwoot-rds-sg.
3. Select the chatwoot-vpc and add an inbound rule for postgres port with source chatwoot-loadbalancer-sg.

RDS subnet group

1. Navigate to the RDS section and select subnet groups.
2. Create chatwoot-rds-group and choose chatwoot-vpc.
3. Select both az's and the private subnets.

RDS

1. Select create a database.
2. Use standard create and choose the Postgres engine.
3. Use the production template, and create a Postgres master username and password.
4. Enable Multi-AZ deployment.
5. Select chatwoot-vpc and select the rds security group we created earlier.
6. Enable password authentication.
7. Click create.
8. After completing the creation, note down the hostname, username, and password. We will need this to configure UniLink.

Redis using AWS Elasticache

1. Follow similar steps like the rds to create Redis security and subnet groups.
2. Create the Redis cluster with a multi-AZ option.

Creating Bastion servers

Create bastion servers in both public subnets. These servers will be used to ssh into UniLink servers in private subnets.

1. Navigate to the EC2 dashboard and click launch instance.
2. Use an Ubuntu 20.04 image with a t3.micro type.
3. Choose chatwoot-vpc and subnet chatwoot-public-1.
4. Name it chatwoot-bastion-a.
5. Add a new sg, chatwoot-bastion-sg and enable ssh access from anywhere.
6. Leave the rest as defaults and click launch.
7. Once the instance is up, try to SSH into the instance.

Repeat and create another bastion, chatwoot-bastion-b in the other AZ.

Install UniLink

1. Navigate to the EC2 section, and click on launch instance.
2. Use an Ubuntu 20.04 image with a c5.xlarge instance type.
3. Choose the chatwoot-vpc and select the private subnet chatwoot-private-1.
4. Disable auto-assign public IP and increase the storage of root volume to 60 GB.
5. Add necessary tags. Set the Name tag to chatwoot.
6. Select the load balancer security group and click launch.
7. SSH into the bastion server and, from there, ssh to the chatwoot instance we created.
8. Switch to the root user.
9. Download the chatwoot Linux installation script.

wget https://get.chatwoot.app/linux/install.sh
chmod 755 install.sh

10. Run the script.

./install.sh --install

Configure UniLink

11. Once the installation is complete, switch to the chatwoot user and navigate to the chatwoot folder. Edit the .env file and replace the Postgres and Redis credentials with RDS and elasticache values.

sudo -i -u chatwoot
cd chatwoot
vi .env

12. Run the db migration.

RAILS_ENV=production bundle exec rake db:prepare

13. Also modify the other necessary environment variable for your chatwoot setup. Refer to https://www.chatwoot.com/docs/self-hosted/deployment/linux-vm#configure-the-required-environment-variables

14. Restart the chatwoot service.

sudo cwctl --restart

Verify login

15. Add this instance to the target group attached to the alb.
16. Navigate to your chatwoot domain to see if everything is working.

Create a custom AMI

1. If you are getting the onboarding page, complete the signup and verify the installation.
2. Voila !! Your chatwoot instance is up.
3. If everything looks good, proceed to create an ami from this instance and name it chatwoot-base-ami.

Auto Scaling Groups (ASG)

1. Create a launch configuration using the above base image.
2. Proceed to create an auto-scaling group from this launch config.
3. Set the minimum and desired capacity to 2 and the maximum to 4. Modify this as per your requirement.
4. Create a scaling policy based on CPU utilization.
5. At this point, we are good to terminate the instance we created earlier.
6. Check the load balancer or target group to verify if two new chatwoot instances have come up.
7. That's it.

Monitoring

1. Refer to https://www.chatwoot.com/docs/self-hosted/monitoring/apm-and-error-monitoring

Updating UniLink

1. log in to one of the application servers and complete the update instructions. Run migrations if needed. Refer to https://www.chatwoot.com/docs/self-hosted/deployment/linux-vm#upgrading-to-a-newer-version-of-chatwoot
2. Create a new ami and update the launch config.

Conclusion

This document is a reference guideline for an HA chatwoot architecture on AWS. Modify or build upon this to suit your requirements.