Can I split a large HAProxy config file into multiple smaller files? - haproxy

Question

I'm building an haproxy config file that has multiple front and backends. It's going to be several hundred lines long and I'd rather split it up into separate files for each of the different websites that I want to loadbalance.
Does HAProxy offer the ability to link to partial config files from the main haproxy.cfg file?

Answer 1

Configuration files can't be linked together from a configuration directive.
However HAProxy can load multiple configuration files from its command line, using the -f switch multiple times:
haproxy -f conf/http-defaults -f conf/http-listeners -f conf/tcp-defaults -f conf/tcp-listeners
If you want to be flexible with the amount of config files you can even specify a directory like this: -f /etc/haproxy. The files will then be used in their lexical order, newer files overriding older files.
See the mailing list for an example, if provides links to the documentation. This information can be found in the management guide, not the regular docs.

Answer 2

Stumbled on this answer where the author created scripts to imitate nginx disable enable sites functionality. In the haproxy init.d startup he uses script loop to build the haproxy -f commands concatenation.
/etc/init.d/haproxy:
EXTRAOPTS=`for FILE in \`find /etc/haproxy/sites-enabled -type l | sort
-n\`; do CONFIGS="$CONFIGS -f $FILE"; done; echo $CONFIGS`
haensite script:
#!/bin/bash
if [[ $EUID -ne 0 ]]; then
echo "You must be a root user" 2>&1
exit 1
fi
if [ $# -lt 1 ]; then
echo "Invalid number of arguments"
exit 1
fi
echo "Enabling $1..."
cd /etc/haproxy/sites-enabled
ln -s ../sites-available/$1 ./
echo "To activate the new configuration, you need to run:"
echo " /etc/init.d/haproxy restart"
hadissite script:
#!/bin/bash
if [[ $EUID -ne 0 ]]; then
echo "You must be a root user" 2>&1
exit 1
fi
if [ $# -lt 1 ]; then
echo "Invalid number of arguments"
exit 1
fi
echo "Disabling $1..."
rm -f /etc/haproxy/sites-enabled/$1
echo "To activate the new configuration, you need to run:"
echo " /etc/init.d/haproxy restart"

Answer 3

This was a solution building off of #stephenmurdoch's answer which involved the use of multiple -f <conf file> arguments to the haproxy executable.
Using the stock CentOS 6.x RPM's included /etc/init.d/haproxy script you can amend it like so:
start() {
$exec -c -q -f $cfgfile $OPTIONS
if [ $? -ne 0 ]; then
echo "Errors in configuration file, check with $prog check."
return 1
fi
echo -n $"Starting $prog: "
# start it up here, usually something like "daemon $exec"
#daemon $exec -D -f $cfgfile -f /etc/haproxy/haproxy_ds.cfg -f /etc/haproxy/haproxy_es.cfg -f /etc/haproxy/haproxy_stats.cfg -p $pidfile $OPTIONS
daemon $exec -D -f $cfgfile $(for i in /etc/haproxy/haproxy_*.cfg;do echo -n "-f $i ";done) -p $pidfile $OPTIONS
retval=$?
echo
[ $retval -eq 0 ] && touch $lockfile
return $retval
}
With the above in place you can then create files such as haproxy_<X>.cfg and haproxy_<Y>.cfg using whatever names you want. The above for loop will include these files in an augmented daemon haproxy ... line if these files are present, otherwise the stock haproxy.cfg file will be used solely.
Within the haproxy_<...>.cfg files you need to make sure that your global and defaults are defined in the "toplevel" haproxy.cfg file. The rest of the files simply need to have frontend/backends and nothing more.

Answer 4

You can follow this simple step.
Insert one line script (cat /etc/$BASENAME/conf.d/*.cfg > $CFG) in /etc/init.d/haproxy
Here is position where you must insert line
CFG=/etc/$BASENAME/$BASENAME.cfg
cat /etc/$BASENAME/conf.d/*.cfg > $CFG
[ -f $CFG ] || exit 1
Reload daemon config with systemctl daemon-reload
Make directory mkdir /etc/haproxy/conf.d
Move default haproxy.cfg to conf.d as global.cfg mv /etc/haproxy/haproxy.cfg /etc/haproxy/conf.d/global.cfg
Create your other .cfg file in conf.d directory
Just restart your haproxy service systemctl restart haproxy
NOTE: /etc/haproxy/haproxy.cfg will be automaticly created from all files in conf.d/

Answer 5

answer of #Bapstie memtioned that, a directory can be passed to haproxy as config file, and files inside will be loaded in alphabet order. It's correct.
But problem is, the package haproxy in CentOS 'base/7/x86_64' repository is so old that it does not support that.
So either do you need to write a wrapper to append -f <individual config file>to the command, or you need to install latest version of haproxy:
for package in centos-release-scl-rh rh-haproxy18-haproxy; do
yum install -y $package
done
and create a drop-in config for haproxy service:
[Service]
ExecStart=
ExecStart=/opt/rh/rh-haproxy18/root/sbin/haproxy -f /etc/haproxy-nutstore/ -p /run/haproxy.pid $OPTIONS

Answer 6

If you use Ansible you can do a trick like this:
- name: haproxy configuration
copy:
content: >
{{ lookup('template', haproxy_cfg_src_top) +
lookup('template', haproxy_cfg_src_edge) +
lookup('template', haproxy_cfg_src_bottom) }}
dest: "{{ haproxy_cfg }}"
owner: "{{ docker_user }}"
group: "docker"
mode: 0664
register: haproxy_cfg_change