Aug 5, 2017
How to Deploy Your Node.JS Server Using MySQL Database on AWS (Mac)
After you have your Node.JS web server tested locally, you can follow this tutorial to deploy it on AWS.
Side note: Here is a really good course on Lynda talking about deploying services on AWS and the 9th section is a complete setup guide for this whole process we will be doing.
I. Set up Security Groups
In the EC2 Console, “Security Groups”:
1. For Load Balancer — Load Balancer
- Type: Custom TCP, Protocol: TCP, Port Range: 3000, Source: Anywhere
2. For App Server Instances — Web Tier
- Type: Custom TCP, Protocol: TCP, Port Range: 3000, Source: Load Balancer Security Group ID
- Type: Custom TCP, Protocol: TCP, Port Range: 3000, Source: My IP
- Type: SSH, Protocol: TCP, Port Range: 22, Source: My IP
3. For Database — DB Tier
- Type: MYSQL/Aurora, Protocol: TCP, Port Range: 3306, Source: My IP
- Type: MYSQL/Aurora, Protocol: TCP, Port Range: 3306, Source: Web Tier Security Group ID
II. Set up an app server in the web tier
- Create an EC2 instance and bootstrap it
use web tier security group, specify the region
Follow this:
Enter the following lines of bash script in the User Data to
- Install Node.js and MySQL server on the EC2 instance
- clone the code for my server
- install all the node dependencies
- start the server app.js
“sudo” in first 2 lines may not be needed
#!/bin/bash
yum update -y
curl --silent --location https://rpm.nodesource.com/setup_6.x | sudo bash -
sudo yum -y install nodejs mysql-server git -yAlso, add tags to help yourself manage your instances.
- name: AppServer
- project: p3
- env: test/prod
2. Connect to it via SSH
Follow this and after typing the following line, hit “yes”, then the instance is connected successful
cd Documents/S3/EDISS/PROJ
ssh -i "./Tina-EDISS.pem" ec2-user@[public DNS/public IP]
You will see something like this but it’s correct. Just ignore it.
The authenticity of host 'ec2-198-51-100-1.compute-1.amazonaws.com (10.254.142.33)' can't be established.
3. Deploy the code and Launch the server
git clone https://github.com/TinaHongBu/PROJ.git
cd PROJ
npm install
node app4. When you have to use a bigger instance, you have to copy file from local repo to EC2
scp -i /Users/Tina/Documents/S3/EDISS/PROJ_local/Tina-EDISS.pem /Users/Tina/Documents/S3/EDISS/PROJ/test/ ec2-user@34.230.41.147:~/inputDataIII. Create more servers in the web tier using an AMI
1. Create an AMI of the current instance
Select the current running instance, create AMI.
2. Create multiple instances with the AMI
Select the AMI at the left panel, create a new instance. Specify the availability zone to be different from the first instance, but within the AZ of your load balancer.
3. Launch the Server in all the instances
cd PROJ
npm install
node appIV. Set up an Classic Load Balancer
1. Create a Load Balancer
In the EC2 console, “load balancer”:
Create Load Balancer →
- Classic Load Balancer →
- Listeners: HTTP port 3000, Availability Zones: at least chose 2 →
- Security Groups: Group for Load Balancer →
- Ping Port: 3000, Ping Path: /health, Response Timeout:4, Interval: 5 →
- Add EC2 Instances later
- Routing: Create a new target group →
- Register instances to the target group
Don’t set it to be sticky, because
- it sends requests from the same client to the same server the whole time so when the instance got killed, all following requests from its clients will still be routed to it and fail. And,
- it will cause unbalance load among instances as the load balancer balances requests based on the traffic but not server load. So it could happen that one server gets a lot of big requests.
Connection draining
2. Set up the ping target response in your code
The health ping target will be: http:3000/health
app.get('/health', function(req, res){
res.send({"message": "Instance is alive"});
}); // just return anythin with 2003. Check if Load Balancer is attached to all instances
Everything should be in service if set up correctly.
Debug using this.
curl -s -k -o /dev/null -v http://private-IP-address-of-the-instance:3000/health
V. Set up an Auto Scaling Group (ASG)
1. Create a Launch Configuration
In the EC2 console, choose Launch Configuration
- Create from existing AMI
- User Data
#!/bin/sh
yum update -y
git clone https://github.com/TinaHongBu/PROJ.git
- Use app server security group
2. Create a Scaling Group
In the EC2 console, choose Auto Scaling Group
- Add subnets →
- Use scaling policies to adjust the capacity of this group: choose “Scale the Auto Scaling group using step or simple scaling policies”, Add new alarm for both the increase group size and decrease group size section.
3. Using Application Load Balancer with Auto Scaling Group
Create a new target group, in the “target” tag choose all instances launched by the auto scaling group. Add to register, then save.
Go to load balancer, add rule.
V(1) Setup MySQL Database using the RDS service
1. Setup a MySQL Database Instance
In the RDS dashboard,
Step 1: Select Engine
- MySQL →
Step 2: Production?
- Dev/Test (Production version has 2 database instances in 2 different availability zones with one being primary and one being the backup, and synchronization is handled by Amazon) →
Step 3: Specify DB Details
- DB Instance Class chose db.t2.micro — 1 vCPU, 1 GiB RAM,
- Multi-AZ Deployment: No,
- DB Instance Identifier: EDISS-P3,
- Master Username: admin,
- Master Password: adminadmin →
Step 4: Configure Advanced Settings
- Availability Zone: Make sure same as your primary instances to decrease latency
- VPC Security Groups: DB tier
2. Connect App Servers to the DB
Copy your DB endpoint: ediss-p3.c6xzjopfczba.us-east-1.rds.amazonaws.com:3306
Delete the port number, do
mysql -h ediss-p3.c6xzjopfczba.us-east-1.rds.amazonaws.com -P 3306 -u admin -p
adminadmin (the password)
mysql -h ediss-p4.c6xzjopfczba.us-east-1.rds.amazonaws.com -P 3306 -u admin -p
to enter the database.
mysql> show databases;
+--------------------+
| Database |
+--------------------+
| information_schema |
| EDISS |
| innodb |
| mysql |
| performance_schema |
| sys |
+--------------------+
6 rows in set (0.00 sec)
mysql> use EDISS;
Database changed
show tables;
select * from p2_prdct;
select * from p2_user;
select * from p2_usertype;
shows that the EDISS database is successfully created already.
V(2) Or, Hold MySQL Server on Ec2
The steps are taken from here.
1. Install MySQL
Install the MySQL database through the CentOS package manager (yum):
sudo yum install mysql-server
2. Start mysql
sudo /sbin/service mysqld start
3. Enable chkconfig on MySQL
sudo chkconfig mysqld on
4. Launch the mysql shell and enter it as the root user:
/usr/bin/mysql -u root -p
Hit enter because no password is set for root user so far.
Now set root password
SET PASSWORD FOR 'root'@'localhost' = PASSWORD('root');
5. Launch MySQL
brew services restart mysql
sudo mysql -u root -p
// enter password for root user
6. Create database, table and insert default users in the table
Hopefully last line will generate duplicate entry error
7. Exit mysql from the terminal
exit
8. Or stop mysql when you need to
sudo /sbin/service mysqld stop
VI. System Functionality Tests
If you use Postman to test requests, type in http://“your DNS address:”“Port number your system is listening to/” “end points”
For example: http://ec2-34-204-68-2.compute-1.amazonaws.com:3000/login
and hit send to see if you get your results back.
Or to use Artillery, in the test file config session modify to “target”: “http://ec2-34-204-68-2.compute-1.amazonaws.com:3000", and run this in terminal: artillery run test.json.
VI. System Availability Tests
Kill instance and see if the traffic gets routed to other nodes.
Add elastic load balancing.
VII. System Performance Tests
Hold Elasticsearch locally
- Install Elasticsearch
brew install elasticsearch
- Start the service
brew services start elasticsearch
- Set Elasticsearch to be accessible from the network
Move to Directory:
elasticsearch-5.5.1/config
vim elasticsearch.yml
## put
network.host: 0.0.0.0
Restart elasticsearch
brew services restart elasticsearch
Hold Elasticsearch on AWS EC2 instance
https://www.elastic.co/blog/running-elasticsearch-on-aws
sudo chkconfig --add elasticsearch
cd /usr/share/elasticsearch/
sudo bin/plugin install cloud-aws -y
sudo service elasticsearch start
curl localhost:9200/_cluster/health?pretty
