Help Center> Distributed Cache Service> Best Practices> Using Nginx for Public Access to Single-node, Master/Standby, or Proxy Cluster DCS Redis Instances

Updated on 2023-09-28 GMT+08:00

View PDF

Using Nginx for Public Access to Single-node, Master/Standby, or Proxy Cluster DCS Redis Instances

Currently, DCS Redis 4.0, 5.0, and 6.0 instances cannot be bound with elastic IP addresses (EIPs) and cannot be accessed over public networks directly.

This section describes how to access a single-node, master/standby, read/write splitting, or Proxy Cluster DCS Redis 4.0, 5.0, or 6.0 instance by using a jump server. This solution cannot be used to access a Redis Cluster instance over public networks.

As shown in Figure 1, the ECS where Nginx is installed is a jump server. The ECS is in the same VPC as the DCS Redis instances and can access the DCS Redis instances through the subnet IP addresses. After an EIP is bound to the ECS, the ECS can be accessed over the public network. Nginx can listen on multiple ports and forward requests to different DCS Redis instances.

Figure 1 Accessing DCS Redis instances in a VPC by using Nginx
Click to enlarge

Do not use public network access in the production environment. Client access exceptions caused by poor public network performance will not be included in the SLA.

Buying an ECS

Obtain the VPC where the DCS Redis instance is deployed.

As shown in the following figure, the master/standby instance is deployed in the vpc-demo VPC.
Figure 2 DCS Redis instance details
Buy an ECS. Configure the ECS with the vpc-demo VPC, bind an EIP to the ECS, and select the bandwidth as required.

Figure 3 ECS details

Installing Nginx

After buying an ECS, install Nginx on the ECS. The following uses CentOS 7.x as an example to describe how to install Nginx. The commands vary depending on the OS.

Run the following command to add Nginx to the Yum repository:
sudo rpm -Uvh http://nginx.org/packages/centos/7/noarch/RPMS/nginx-release-centos-7-0.el7.ngx.noarch.rpm
Run the following command to check whether Nginx has been added successfully:
yum search nginx
Run the following command to install Nginx:
sudo yum install -y nginx
Run the following command to install the stream module:
yum install nginx-mod-stream --skip-broken
Run the following commands to start Nginx and set it to run automatically upon system startup:
sudo systemctl start nginx.service

sudo systemctl enable nginx.service
In the address box of a browser, enter the server address (the EIP of the ECS) to check whether Nginx is installed successfully.
If the following page is displayed, Nginx has been installed successfully.

Setting Up Nginx

After installing Nginx, configure request forwarding rules to specify the ports that Nginx listens on and the DCS Redis instances that Nginx forwards requests to.

Open and modify the configuration file.
cd /etc/nginx

vi nginx.conf

The following is a configuration example. To access multiple DCS Redis instances over public networks, configure multiple server sections and configure the DCS Redis instance connection addresses for proxy_pass.
```
stream {
    server {
        listen 8080;
        proxy_pass 192.168.0.5:6379;
    }
    server {
        listen 8081;
        proxy_pass 192.168.0.6:6379;
    }
}
```
Set proxy_pass to the IP address of the DCS Redis instance in the same VPC. You can obtain the IP address from the Connection area on the DCS instance details page.

Figure 4 Adding Nginx configurations
Restart Nginx.
service nginx restart
Verify whether Nginx has been started.
netstat -an|grep 808

Figure 5 Starting Nginx and verifying the start

If Nginx is listening on ports 8080 and 8081, Nginx has been started successfully.

(Optional) Persistent Connections

If persistent connections ("pconnect" in Redis terminology) are required for public network access, add the following configuration in Configuring Nginx:

Timeout of a connection from Nginx to the server

stream {
    server {
        listen 8080;
        proxy_pass 192.168.0.5:6379;
        proxy_socket_keepalive on;
        proxy_timeout 60m;
        proxy_connect_timeout 60s;
    }
    server {
        listen 8081;
        proxy_pass 192.168.0.6:6379;
        proxy_socket_keepalive on;
        proxy_timeout 60m;
        proxy_connect_timeout 60s;
    }
}

The default value of proxy_timeout is 10m. You can set it to 60m or other values as required. For details about this parameter, see the Nginx official website.

Timeout of a connection from the client to Nginx
```
http {
    keepalive_timeout 3600s;
}
```
The default value of keepalive_timeout is 75s. You can set it to 3600s or other values as required. For details about this parameter, see the Nginx official website.

Accessing DCS Redis Instances Using Nginx

Log in to the ECS console and check the security group rules of the ECS that serves as the jump server. Ensure that access over ports 8080 and 8081 is allowed.
1. Click the ECS name to go to the details page.
2. On the Security Groups tab page, click Modify Security Group Rule. The security group configuration page is displayed.
Figure 6 Checking the ECS security group

Figure 7 Adding an inbound rule for the security group
In the public network environment, open the redis-cli and run the following command to check whether the login and query are successful:

Ensure that redis-cli has been installed in the public network environment by referring to redis-cli.

./redis-cli -h {myeip} -p {port} -a {mypassword}

In the preceding command, {myeip} indicates the host connection address, which should be replaced with the EIP of the ECS. Replace {port} with the listening port of Nginx.

As shown in the following figures, the two listening ports are 8080 and 8081, which correspond to two DCS Redis instances.

Figure 8 Accessing the first DCS Redis instance using Nginx

Figure 9 Accessing the second DCS Redis instance using Nginx

The jump server has now been set up. You can access Redis over public networks.