Setup Load-Balancing Cluster with LVS and Piranha on Centos 6

This setup guide is about building web cluster with a pair of Linux loadbalancers as frontend. It uses Linux Virtual Server(LVS) and Piranha. Piranha is a web-based gui installed in LVS Routers primarily to generating a valid /etc/lvs.cf file. Start by installing LVS on LVS Router.

[root@lvsrouter ~]# yum groupinstall "Load Balancer"
================================================================================
 Package         Arch        Version                         Repository    Size
================================================================================
Installing:
 ipvsadm         x86_64      1.25-10.el6                     base          41 k
 piranha         x86_64      0.8.6-2.el6_4.1                 updates      623 k
Installing for dependencies:
 libedit         x86_64      2.11-4.20080712cvs.1.el6        base          74 k
 libnl           x86_64      1.1-14.el6                      base         121 k
 php             x86_64      5.3.3-22.el6                    base         1.1 M
 php-cli         x86_64      5.3.3-22.el6                    base         2.2 M
 php-common      x86_64      5.3.3-22.el6                    base         524 k

Transaction Summary
================================================================================
Install       7 Package(s)

Total download size: 4.7 M
Installed size: 18 M
Is this ok [y/N]: y

Start piranha and pulse services on reboot.

chkconfig piranha-gui on
chkconfig pulse on

Set a password for piranha web

/usr/sbin/piranha-passwd

Allow ports in iptables

vi /etc/sysconfig/iptables
-A INPUT -m state --state NEW -m tcp -p tcp --dport 3636 -j ACCEPT

Edit piranha config

vi  /etc/sysconfig/ha/conf/httpd.conf

Start the piranha gui service

service piranha-gui start

For LVS router to forward network packets properly to real servers, each LVS router node must have IP forwarding turned on. Turn on packet forwarding by editing sysctl.conf:

vi /etc/sysctl.conf
net.ipv4.ip_forward = 1

Reload sysctl

sysctl -p /etc/sysctl.conf

Start http services on the Real Servers

service httpd start

On the LVS Router, log in to the Piranha web ui to begin configuration. (eg http://(loadbalancer-ip):3636). Using "piranha" as user name, and key in the newly setup password. Enter Primary Router's IP address in "Global Settings"->"Primary server public IP:".

Enter Redundant Router's IP address in "Redundancy"->"Redundant server public IP:".

Enter Virtual Server settings in "Virtual Servers"->"Virtual Server".

Include Real Server settings in "Virtual Servers"->"Real Server".

On each time the Piranha Gui was changed, have to sync the settings to Backup Routers, and restart the pulse service on both routers.

service pulse restart

To see the virtual server's statistics, use "watch ipvsadm".

[root@lvsrouter ~]# watch ipvsadm 

Create arptables entry for each Virtual IP address on each Real Server(eg Webserver). You can add below command to /etc/rc.local to start on every reboot. If the network adaptor on Real Server is eth0, refer to below, if not change as according:

ip addr add (virtual ip) dev eth0:1

Direct Routing with arptables_jf. To configure each real server to ignore ARP requests for each of the virtual IP addresses the Piranha cluster services:

yum install arptables_jf
arptables -A IN -d (virtual_ip) -j DROP
arptables -A OUT -d (virtual_ip) -j mangle --mangle-ip-s (real_ip)
chkconfig arptables_jf on
service arptables_jf save
service arptables_jf restart

Create a loopback on each Real Server for monitoring Virtual IP.

vi /etc/sysconfig/network-scripts/ifcfg-lo:0

DEVICE=lo:0
IPADDR=(Virtual IP)
NETMASK=255.255.255.255
NETWORK=192.168.0.0
ONBOOT=yes
NAME=loopback

PHP: Enable zlib output compression

You can enable php compression as it have a lot of benefits. It reduces the web traffic from host provider to your web. It speeds up the loading of your pages on clients PC. And it is supported by all newer web browser. All you have to do is to set "zlib.output_compression=on" in php.ini:

zlib.output_compression=on

Bonnie Benchmark for ext3, ext4, xfs on CENTOS6

This benchmark was to compare the performance of ext3,ext4,xfs using CENTOS6. Bonnie++ was used to identify performance difference for the 3 filesystems on Linux.

Below were the steps and commands used to create the partitions.

For ext3 fs:

fdisk /dev/sda
mke2fs -t ext3 /dev/sda4
mount /dev/sda4 /bench -t ext3
bonnie++ -d /bench/ -c 5 -s 1G -n 32 -m ext3

For ext4 fs:

fdisk /dev/sda
mke2fs -t ext4 /dev/sda4
mount /dev/sda4 /bench -t ext4
bonnie++ -d /bench/ -c 5 -s 1G -n 32 -m ext4

For xfs fs:

fdisk /dev/sda
/sbin/mkfs.xfs /dev/sda4
mount /dev/sda4 /bench -t xfs
bonnie++ -d /bench/ -c 5 -s 1G -n 32 -m xfs

The results:

1. xfs is having highest SO_PC_KPS (Sequential Output,Per Char,k/sec).

2. ext4 is having highest SI_PC_KPS (Sequential Input,Per Char,k/sec).

3. ext4 is having highest RS_PS (Random Seek,/sec).

4. ext4 is having highest SCC_PS (Sequential Create Create, /sec).

5. ext4 is having highest SCD_PS (Sequential Create Delete, /sec).

6. ext4 is having highest RCC_PS (Random Create Create, /sec).

7. ext4 is having highest RCD_PS (Random Create Delete, /sec).

To create user account on Linux

To create a new user account, you have to login as root, and use "useradd". To change the password, use "passwd".

[root@localhost /]# useradd john
[root@localhost /]# passwd john

How to install Varnish cache on Fedora 18

Varnish is a web accelerator in front of your web server, and it significantly speeds up web content delivery. This guide will show you how to install Varnish cache using YUM on Fedora 18. First, install Varnish and its dependencies:

[root@localhost ~]# yum install varnish

 After above installation, there are 2 config files to setup. "/etc/varnish/default.vcl" and "/etc/varnish/varnish.params".

If you are installing Varnish with Apache Web Server on the same hardware(localhost). Insert below config to "/etc/varnish/default.vcl".

[root@localhost ~]# vi /etc/varnish/default.vcl

backend default {
 .host = "127.0.0.1";
 .port = "80";
}

 By default, Varnish is installed to listen to port 6081. The admin port is installed on port 6082.

[root@localhost ~]# vi /etc/varnish/varnish.params

 Here is my varnish.params:

# Varnish environment configuration description. This was derived from
# the old style sysconfig/defaults settings


# Set this to 1 to make systemd reload try to switch vcl without restart.
RELOAD_VCL=1


# Main configuration file. You probably want to change it.
VARNISH_VCL_CONF=/etc/varnish/default.vcl


# Default address and port to bind to. Blank address means all IPv4
# and IPv6 interfaces, otherwise specify a host name, an IPv4 dotted
# quad, or an IPv6 address in brackets.
# VARNISH_LISTEN_ADDRESS=192.168.1.5
VARNISH_LISTEN_PORT=6081


# Admin interface listen address and port
#VARNISH_ADMIN_LISTEN_ADDRESS=127.0.0.1
VARNISH_ADMIN_LISTEN_PORT=6082

# Shared secret file for admin interface
VARNISH_SECRET_FILE=/etc/varnish/secret

# The minimum and maximum number of worker threads to start
VARNISH_MIN_THREADS=5
VARNISH_MAX_THREADS=1000

# Idle timeout for worker threads
VARNISH_THREAD_TIMEOUT=120

# Backend storage specification, see Storage Types in the varnishd(5)
# man page for details.
VARNISH_STORAGE="file,/var/lib/varnish/varnish_storage.bin,1G"

# Default TTL used when the backend does not specify one
VARNISH_TTL=120

# User and group for the varnishd worker processes
VARNISH_USER=varnish
VARNISH_GROUP=varnish

After you have edited the config files, set the machine to start Varnish on reboot. And start the cache service.

[root@localhost ~]# systemctl enable varnish.service
[root@localhost ~]# systemctl start varnish.service

To access the cache, you can type the IP address of the server on the web browser. Assume the Varnish and Apache Web server is at 192.168.0.5.

http://192.168.0.5:6081

 To bypass varnish cache, you can still reach web server (Apache) at:

http://192.168.0.5

How to host multiple domains on the same http server using the same ip address

In Apache HTTP Server,we can use Name-based Virtual Hosts to host multiply domains on the same server using the same ip address. Using this method, you can have many different hosts sharing the same IP address. This method you need to configure your DNS server to map each hostname to the (same) IP address and set the Apache HTTP Server to recognize the different hostnames. This method of virtual hosting reduce the demand for IP addresses. Supposing you want to host 2 domains on 1 http server,you add the following to httpd.conf.

<virtualhost>
    # The first virtual host is also the default for *:80
    ServerName www.fc18.com
    ServerAlias fc18.com *.fc18.com
    DocumentRoot /var/www/html
</virtualhost>

<virtualhost>
    ServerName other.hosts.com
    ServerAlias hosts.com
    DocumentRoot /var/www/hosts
</virtualhost>

In the DNS server, you can add these to (bind) named.conf:

www.fc18.com.    A 192.168.0.16
fc18.com.        A 192.168.0.16
other.hosts.com. A 192.168.0.16
hosts.com.       A 192.168.0.16 

If you are using hosts file, you can add these to hosts file:

192.168.0.16 www.fc18.com
192.168.0.16 fc18.com
192.168.0.16 www1.fc18.com
192.168.0.16 other.hosts.com
192.168.0.16 hosts.com

Install APC (php) on Fedora 18

The Alternative PHP Cache (APC) is a bytecode optimization opcode cache for PHP. This free cache leads to faster runtime execution as it stores the final, compiled result in shared memory. To install APC, you can setup using YUM.

[root@localhost ~]# yum install php-pecl-apc

Using YUM method, it will create /etc/php.d/apc.ini , enabling apc extension module "extension = apc.so". If apc.ini is not found, you can insert "extension = apc.so" into php.ini. After apache restart(service httpd restart), you will see apc enabled in phpinfo.php.