HAProxy(High Availability Proxy) is an open source load balancer which can load balance any TCP service. It is particularly suited for HTTP load balancing as it supports session persistence and layer 7 processing.
With DigitalOcean Private Networking, HAProxy can be configured as a front-end to load balance two VPS through private network connectivity.
We will be using three VPS (droplets) here:
Droplet 1 - Load Balancer Hostname: haproxy OS: Ubuntu Public IP: 1.1.1.1 Private IP: 10.0.0.100
Droplet 2 - Node 1 Hostname: lamp1 OS: LAMP on Ubuntu Private IP: 10.0.0.1
Droplet 2 - Node 2 Hostname: lamp2 OS: LAMP on Ubuntu Private IP: 10.0.0.2
Use the apt-get
command to install HAProxy.
apt-get install haproxy
We need to enable HAProxy to be started by the init script.
nano /etc/default/haproxy
Set the ENABLED
option to 1
ENABLED=1
To check if this change is done properly execute the init script of HAProxy without any parameters. You should see the following.
root@haproxy:~# service haproxy
Usage: /etc/init.d/haproxy {start|stop|reload|restart|status}
We’ll move the default configuration file and create our own one.
mv /etc/haproxy/haproxy.cfg{,.original}
Create and edit a new configuration file:
nano /etc/haproxy/haproxy.cfg
Let us begin by adding configuration block by block to this file:
global
log 127.0.0.1 local0 notice
maxconn 2000
user haproxy
group haproxy
The log
directive mentions a syslog server to which log messages will be sent. On Ubuntu rsyslog is already installed and running but it doesn’t listen on any IP address. We’ll modify the config files of rsyslog later.
The maxconn
directive specifies the number of concurrent connections on the frontend. The default value is 2000
and should be tuned according to your VPS’ configuration.
The user
and group
directives changes the HAProxy process to the specified user/group. These shouldn’t be changed.
defaults
log global
mode http
option httplog
option dontlognull
retries 3
option redispatch
timeout connect 5000
timeout client 10000
timeout server 10000
We’re specifying default values in this section. The values to be modified are the various timeout
directives. The connect
option specifies the maximum time to wait for a connection attempt to a VPS to succeed.
The client
and server
timeouts apply when the client or server is expected to acknowledge or send data during the TCP process. HAProxy recommends setting the client
and server
timeouts to the same value.
The retries
directive sets the number of retries to perform on a VPS after a connection failure.
The option redispatch
enables session redistribution in case of connection failures. So session stickness is overriden if a VPS goes down.
listen appname 0.0.0.0:80
mode http
stats enable
stats uri /haproxy?stats
stats realm Strictly\ Private
stats auth A_Username:YourPassword
stats auth Another_User:passwd
balance roundrobin
option httpclose
option forwardfor
server lamp1 10.0.0.1:80 check
server lamp2 10.0.0.2:80 check
This contains configuration for both the frontend and backend. We are configuring HAProxy to listen on port 80 for appname
which is just a name for identifying an application. The stats
directives enable the connection statistics page and protects it with HTTP Basic authentication using the credentials specified by the stats auth
directive.
This page can viewed with the URL mentioned in stats uri
so in this case, it is http://1.1.1.1/haproxy?stats
; a demo of this page can be viewed here.
The balance
directive specifies the load balancing algorithm to use. Options available are Round Robin (roundrobin
), Static Round Robin (static-rr
), Least Connections (leastconn
), Source (source
), URI (uri
) and URL parameter (url_param
).
Information about each algorithm can be obtained from the official documentation.
The server
directive declares a backend server, the syntax is:
server <name> <address>[:port] [param*]
The name we mention here will appear in logs and alerts. There are many paratmeters supported by this directive and we’ll be using the check
and cookie
parameters in this article. The check
option enables health checks on the VPS otherwise, the VPS is
always considered available.
Once you’re done configuring start the HAProxy service:
service haproxy start
To test this setup, create a PHP script on all your web servers (backend servers - LAMP1 and LAMP2 here).
/var/www/file.php
<?php
header('Content-Type: text/plain');
echo "Server IP: ".$_SERVER['SERVER_ADDR'];
echo "\nClient IP: ".$_SERVER['REMOTE_ADDR'];
echo "\nX-Forwarded-for: ".$_SERVER['HTTP_X_FORWARDED_FOR'];
?>
Now we will use curl and request this file multiple times.
> curl http://1.1.1.1/file.php
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.X.X
> curl http://1.1.1.1/file.php
Server IP: 10.0.0.2
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.X.X
> curl http://1.1.1.1/file.php
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.X.X
Notice here how HAProxy alternatively toggled the connection between LAMP1 and LAMP2, this is how Round Robin works. The client IP we see here is the Private IP address of the load balancer and the X-Forwarded-For
header is your IP.
To see how failover works, go to a web server and stop the service:
lamp1@haproxy:~#service apache2 stop
Send requests with curl
again to see how things work.
If your web application serves dynamic content based on users’ login sessions (which application doesn’t), visitors will experience odd things due to continuous switching between VPS. Session stickiness ensures that a visitor sticks on to the VPS which served their first request. This is possible by tagging each backend server with a cookie.
We’ll use the following PHP code to demonstrate how session stickiness works.
/var/www/session.php
<?php
header('Content-Type: text/plain');
session_start();
if(!isset($_SESSION['visit']))
{
echo "This is the first time you're visiting this server";
$_SESSION['visit'] = 0;
}
else
echo "Your number of visits: ".$_SESSION['visit'];
$_SESSION['visit']++;
echo "\nServer IP: ".$_SERVER['SERVER_ADDR'];
echo "\nClient IP: ".$_SERVER['REMOTE_ADDR'];
echo "\nX-Forwarded-for: ".$_SERVER['HTTP_X_FORWARDED_FOR']."\n";
print_r($_COOKIE);
?>
This code creates a PHP session and displays the number of page views in a single session.
In this method, all responses from HAProxy to the client will contain a Set-Cookie:
header with the name of a backend server as its cookie value. So going forward the client (web browser) will include this cookie with all its requests and HAProxy will forward the request to the right backend server based on the cookie value.
For this method, you will need to add the cookie
directive and modify the server
directives under listen
cookie SRVNAME insert
server lamp1 10.0.0.1:80 cookie S1 check
server lamp2 10.0.0.2:80 cookie S2 check
This causes HAProxy to add a Set-Cookie:
header with a cookie named SRVNAME
having its value as S1
or S2
based on which backend answered the request. Once this is added restart the service:
service haproxy restart
and use curl
to check how this works.
> curl -i http://1.1.1.1/session.php
HTTP/1.1 200 OK
Date: Tue, 24 Sep 2013 13:11:22 GMT
Server: Apache/2.2.22 (Ubuntu)
X-Powered-By: PHP/5.3.10-1ubuntu3.8
Set-Cookie: PHPSESSID=l9haakejnvnat7jtju64hmuab5; path=/
Expires: Thu, 19 Nov 1981 08:52:00 GMT
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Vary: Accept-Encoding
Content-Length: 143
Connection: close
Content-Type: text/plain
Set-Cookie: SRVNAME=S1; path=/
This is the first time you're visiting this server
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.X.X
Array
(
)
This is the first request we made and it was answered by LAMP1 as we can see from Set-Cookie: SRVNAME=S1; path=/
. Now, to emulate what a web browser would do for the next request, we add these cookies to the request using the --cookie
parameter of curl.
> curl -i http://1.1.1.1/session.php --cookie "PHPSESSID=l9haakejnvnat7jtju64hmuab5;SRVNAME=S1;"
HTTP/1.1 200 OK
Date: Tue, 24 Sep 2013 13:11:45 GMT
Server: Apache/2.2.22 (Ubuntu)
X-Powered-By: PHP/5.3.10-1ubuntu3.8
Expires: Thu, 19 Nov 1981 08:52:00 GMT
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Vary: Accept-Encoding
Content-Length: 183
Connection: close
Content-Type: text/plain
Your number of visits: 1
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.87.127
Array
(
[PHPSESSID] => l9haakejnvnat7jtju64hmuab5
[SRVNAME] => S1
)
> curl -i http://1.1.1.1/session.php --cookie "PHPSESSID=l9haakejnvnat7jtju64hmuab5;SRVNAME=S1;"
HTTP/1.1 200 OK
Date: Tue, 24 Sep 2013 13:11:45 GMT
Server: Apache/2.2.22 (Ubuntu)
X-Powered-By: PHP/5.3.10-1ubuntu3.8
Expires: Thu, 19 Nov 1981 08:52:00 GMT
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Vary: Accept-Encoding
Content-Length: 183
Connection: close
Content-Type: text/plain
Your number of visits: 2
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.87.127
Array
(
[PHPSESSID] => l9haakejnvnat7jtju64hmuab5
[SRVNAME] => S1
)
Both of these requests were served by LAMP1 and the session was properly maintained. This method is useful if you want stickiness for all files on the web server.
On the other hand, if you want stickiness only for specific cookies or don’t want to have a separate cookie for session stickiness, the prefix
option is for you.
To use this method, use the following cookie
directive:
cookie PHPSESSID prefix
The PHPSESSID
can be replaced with your own cookie name. The server
directive remains the same as the previous configuration.
Now let’s see how this works.
> curl -i http://1.1.1.1/session.php
HTTP/1.1 200 OK
Date: Tue, 24 Sep 2013 13:36:27 GMT
Server: Apache/2.2.22 (Ubuntu)
X-Powered-By: PHP/5.3.10-1ubuntu3.8
Set-Cookie: PHPSESSID=S1~6l2pou1iqea4mnhenhkm787o56; path=/
Expires: Thu, 19 Nov 1981 08:52:00 GMT
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Vary: Accept-Encoding
Content-Length: 143
Content-Type: text/plain
This is the first time you're visiting this server
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.X.X
Array
(
)
Notice how the server
cookie S1
is prefixed to the session cookie. Now, let’s send two more requests with this cookie.
> curl -i http://1.1.1.1/session.php --cookie "PHPSESSID=S1~6l2pou1iqea4mnhenhkm787o56;"
HTTP/1.1 200 OK
Date: Tue, 24 Sep 2013 13:36:45 GMT
Server: Apache/2.2.22 (Ubuntu)
X-Powered-By: PHP/5.3.10-1ubuntu3.8
Expires: Thu, 19 Nov 1981 08:52:00 GMT
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Vary: Accept-Encoding
Content-Length: 163
Content-Type: text/plain
Your number of visits: 1
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.X.X
Array
(
[PHPSESSID] => 6l2pou1iqea4mnhenhkm787o56
)
> curl -i http://1.1.1.1/session.php --cookie "PHPSESSID=S1~6l2pou1iqea4mnhenhkm787o56;"
HTTP/1.1 200 OK
Date: Tue, 24 Sep 2013 13:36:54 GMT
Server: Apache/2.2.22 (Ubuntu)
X-Powered-By: PHP/5.3.10-1ubuntu3.8
Expires: Thu, 19 Nov 1981 08:52:00 GMT
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Vary: Accept-Encoding
Content-Length: 163
Content-Type: text/plain
Your number of visits: 2
Server IP: 10.0.0.1
Client IP: 10.0.0.100
X-Forwarded-for: 117.213.X.X
Array
(
[PHPSESSID] => 6l2pou1iqea4mnhenhkm787o56
)
We can clearly see that both the requests were served by LAMP1 and the session is perfectly working.
When we began configuring HAProxy, we added a line: log 127.0.0.1 local0 notice
which sends syslog messages to the localhost IP address. But by default, rsyslog on Ubuntu doesn’t listen on any address. So we have to make it do so.
Edit the config file of rsyslog.
nano /etc/rsyslog.conf
Add/Edit/Uncomment the following lines:
$ModLoad imudp
$UDPServerAddress 127.0.0.1
$UDPServerRun 514
Now rsyslog will work on UDP port 514 on address 127.0.0.1 but all HAProxy messages will go to /var/log/syslog
so we have to separate them.
Create a rule for HAProxy logs.
nano /etc/rsyslog.d/haproxy.conf
Add the following line to it.
if ($programname == 'haproxy') then -/var/log/haproxy.log
Now restart the rsyslog service:
service rsyslog restart
This writes all HAProxy messages and access logs to /var/log/haproxy.log
.
Under the listen
directive, we used option httpclose
which adds a Connection: close
header. This tells the client (web browser) to close a connection after a response is received.
If you want to enable keep-alives on HAProxy, replace the option httpclose
line with:
option http-server-close
timeout http-keep-alive 3000
Set the keep-alive timeout wisely so that a few connections don’t drain all the resources of the load balancer.
Keepalives can be tested using curl
by sending multiple requests at the same time. Unnecessary output will be omitted in the following example:
> curl -v http://1.1.1.1/index.html http://1.1.1.1/index.html
* About to connect() to 1.1.1.1 port 80 (#0)
* Trying 1.1.1.1... connected
> GET /index.html HTTP/1.1
> User-Agent: curl/7.23.1 (x86_64-pc-win32) libcurl/7.23.1 OpenSSL/0.9.8r zlib/1.2.5
> Host: 1.1.1.1
> Accept: */*
>
......[Output omitted].........
* Connection #0 to host 1.1.1.1 left intact
* Re-using existing connection! (#0) with host 1.1.1.1
* Connected to 1.1.1.1 (1.1.1.1) port 80 (#0)
> GET /index.html HTTP/1.1
> User-Agent: curl/7.23.1 (x86_64-pc-win32) libcurl/7.23.1 OpenSSL/0.9.8r zlib/1.2.5
> Host: 1.1.1.1
> Accept: */*
>
.......[Output Omitted].........
* Connection #0 to host 1.1.1.1 left intact
* Closing connection #0
In this output, we have to look for the line: Re-using existing connection! (#0) with host 1.1.1.1
, which indicates that curl used the same connection to make subsequent requests.
Thanks for learning with the DigitalOcean Community. Check out our offerings for compute, storage, networking, and managed databases.
This textbox defaults to using Markdown to format your answer.
You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!
Muchas gracias!
It would be great to see this article expanded to include SSL setup, but I suppose existing articles can be referenced to figure it out.
This configuration from the tutorial is throwing the following errors that prevent haproxy from starting:
539 (12110) : parsing [/etc/haproxy/haproxy.cfg:19] : ‘listen’ cannot handle unexpected argument ‘0.0.0.0:80’. 539 (12110) : parsing [/etc/haproxy/haproxy.cfg:19] : please use the ‘bind’ keyword for listening addresses.
Has the configuration for later versions of HAProxy changed, perhaps?
These instructions need to be updated. The
listen appname 0.0.0.0:80
directive is no longer valid as of the latest haproxy releases.It should be changed to:
Great…All your articles are really helpful
Hi, First of all thank you for your guide! I’ve set up HAproxy as a NLB for 2 Exchange 2013 Multi-role servers.
I’m now trying to get redirection working when a user types in the flat hostname “webmail” into a browser. I would like it redirected to https://webmail.domain.com/owa
Could you please help with that?
Thanks
Thanks.
Hi, Appreciate your good work.
I have a question. Please let me know the way to change/update the configuration of HAproxy with out any downtime?
Think this needs an update as of HAProxy 1.5
I have a server which host serveral websites (different domain names).
How to setup haproxy in this case?