Creating a Load Balancer
Course Development Director in Content
In this learning activity, you will install and configure a load balancer to be the front end for two pre-built apache nodes. The load balancer should be configured to run in the best-effort stickiness mode.
Creating a Load Balancer
A business unit is requesting our assistance in resolving some performance issues. They've had additional application nodes created, but need us to build a load balancer that acts as their front end.
The application nodes will listen on port 8080, and the load balancer should listen on port 80 for web traffic.
Get logged in
Use the credentials and server IP in the hands-on lab overview page to log into our lab server. Notice there are multiple machines we're working with. Pay attention in the lab guide, as the shell prompt will reveal which one we're working with at the moment.
Install HAProxy on
Before we do anything else, let's run a
sudo -i right off. Then we'll be
root and won't have to keep typing in a password.
Now, we'll start off by installing HAProxy:
[root@Server1]# yum -y install haproxy
Enable and start the service:
[root@Server1]# systemctl enable haproxy [root@Server1]# systemctl start haproxy
Then verify that incoming port 80 traffic is permitted through the firewall. Running
firewall-cmd --list-all should show
http in the list of allowed services.
Configure HAProxy on
We need to configure the HAProxy's frontend and backend. HAProxy should listen on port 80, and use
node2 as the backend nodes. Let's edit
/etc/haproxy/haproxy.cfg and add the configuration code:
timeout check 10s maxconn 3000 # Our code starts here: frontend app1 bind *:80 mode http default_backend apache_nodes backend apache_nodes mode http balance source server node1 10.0.1.20:8080 check server node2 10.0.1.30:8080 check # End of our code #------------------------------------------------------- # main frontend which proxys to the backends #-------------------------------------------------------
We'll have to restart the daemon so that our changes take effect, then check with
[root@Server1]# systemctl restart haproxy [root@Server1]# ss -lntp
We need to permit incoming port 8080/TCP traffic on
node2. On each node (after logging in, then becoming
root), perform the following:
[root@node]# firewall-cmd --permanent --add-port=8080/tcp
Then reload the firewall configuration:
[root@node]# firewall-cmd --reload
Client1, Validate That Settings Are Correct
Client1, we can run a
curl on the
Server1 private IP address:
[cloud_user@Client1]$ curl &ltServer1_IP_ADDRESS>
We should get a reply from either
node2. Something like
While this means we're up and running, if we really want to test, we'll go shut down the one we didn't hear from. In this case, we got a reply from
node1, so if we log into it (with SSH), then shut down the web server (with
sudo systemctl stop httpd), another
curl command from
Client1 will get us a reply. But this time it will be
node2 serving up the web page.
We have two web servers, and they're each serving up traffic sent to them by a load balancer that we've just set up and tested. This is exactly what we wanted to have working. Congratulations!