Category: Uncategorized

Syncing Content Between RackSpace Cloud Servers With Unison

June 2, 2011

We have two RackSpace Cloud Server running CentOS that need to have the web content kept in sync. Changes on either server need to be replicated to the other server. Easy way to get this accomplished is with the Unison File Syncronizer (http://www.cis.upenn.edu/~bcpierce/unison/). Here are the steps: 1. Install SSH keys for root on each server so SSH can run without passwords between the servers. 2. Add /etc/hosts file entries to map the private addresses of the servers to names.

192.168.10.1 web1-priv
192.168.10.2 web2-priv

192.168.10.1 web1-priv

192.168.10.2 web2-priv

3. Install Unison on each server:

yum install unison227

1	yum install unison227

4. Create a profile on each server at /root/.unison/sync_web.prf containing:


# Reasonable defaults
auto=true
confirmbigdeletes=true
contactquietly=true
group=true
maxthreads=20
numericids=true
owner=true
times=true

# Skip confirmation
batch=true

# Log all sync operations
log=true
logfile=/var/log/unison.log

# Backup deleted files
backup=Name *
backuplocation=central
backupdir=/var/www/vhosts/mydomain.com/backup/
maxbackups=5

# Local root
root=/var/www/vhosts/mydomain.com/httpdocs/

# Remote root (the double forward-slash between IP and remote path is correct)
root=ssh://<font style="color:red">web2-priv</font>///var/www/vhosts/mydomain.com/httpdocs/

# Resolve conflicts in favor of local root
prefer=newer

# Don't sync (can specify multiple ignore lines)
#ignore=Path */var/cache

# Reasonable defaults

auto=true

confirmbigdeletes=true

contactquietly=true

group=true

maxthreads=20

numericids=true

owner=true

times=true

# Skip confirmation

batch=true

# Log all sync operations

log=true

logfile=/var/log/unison.log

# Backup deleted files

backup=Name *

backuplocation=central

backupdir=/var/www/vhosts/mydomain.com/backup/

maxbackups=5

# Local root

root=/var/www/vhosts/mydomain.com/httpdocs/

# Remote root (the double forward-slash between IP and remote path is correct)

root=ssh://<font style="color:red">web2-priv</font>///var/www/vhosts/mydomain.com/httpdocs/

# Resolve conflicts in favor of local root

prefer=newer

# Don't sync (can specify multiple ignore lines)

#ignore=Path */var/cache

Notice the hostname “web2-priv” in red above. The file on web1 should reference web2-priv. The file on web2 should reference web1-priv. You’ll need to adjust the path listed in the file to match your environment. 5. Next create a script to run Unison with the profile created in the previous step.

MySQL Load Balancing With HAProxy

March 31, 2011

The goal of this series of articles has been to construct a high availability and load balanced MySQL cluster with CentOS on the RackSpace Cloud. You should begin with this article to setup the HA Linux cluster: HA Linux Cluster On RackSpace Cloud Servers Then follow this article to add the multi-master MySQL replication: MySQL Multi Master HA Cluster The last part of the process is to add load balancing. 1. Start by installing HAProxy on both servers:

yum install haproxy

1	yum install haproxy

Unfortunately yum will provide an outdated version so we need to upgrade from source as follows:

cd /root/archive/
wget http://haproxy.1wt.eu/download/1.4/src/haproxy-1.4.14.tar.gz
tar -xvzf haproxy-1.4.14.tar.gz
cd haproxy-1.4.14
make TARGET=linux26
make install

cd /root/archive/

wget http://haproxy.1wt.eu/download/1.4/src/haproxy-1.4.14.tar.gz

tar -xvzf haproxy-1.4.14.tar.gz

cd haproxy-1.4.14

make TARGET=linux26

make install

Now edit /etc/init.d/haproxy and change this line:

exec="/usr/sbin/haproxy"

1	exec="/usr/sbin/haproxy"

to:

exec="/usr/local/sbin/haproxy"

1	exec="/usr/local/sbin/haproxy"

2. Now edit /etc/haproxy/haproxy.cfg a. Remove all existing “listen”, “frontend” and “backend” sections. b. Find this line if it exists:

option      httplog

1	option httplog

and change it to:

option      tcplog

1	option tcplog

c. Add this listen section:

listen mysql-cluster 0.0.0.0:3307
    mode tcp
    balance roundrobin
    option mysql-check user root

    server db01 [HOST1_IP]:3306 check
    server db02 [HOST2_IP]:3306 check

listen mysql-cluster 0.0.0.0:3307

mode tcp

balance roundrobin

option mysql-check user root

server db01 [HOST1_IP]:3306 check

server db02 [HOST2_IP]:3306 check

The [PASSWORD] controls access to the HAProxy web interface. And [HOST1_IP] and

MySQL Multi Master HA Cluster

March 27, 2011

The goal of this article is to setup an HA Linux cluster on RackSpace Cloud Servers for MySQL with multi master replication and HAProxy for load balancing. Begin by creating the HA Linux client as described here: HA Linux Cluster On RackSpace Cloud Servers Once this is complete then return to these instructions. 1. Begin by logging into each of the servers and creating a replication slave user with the following SQL commands:

GRANT REPLICATION SLAVE ON *.* TO 'repl'@'%' IDENTIFIED BY '[PASSWORD]';
FLUSH PRIVILEGES;

1 2	GRANT REPLICATION SLAVE ON . TO 'repl'@'%' IDENTIFIED BY '[PASSWORD]'; FLUSH PRIVILEGES;

Where [PASSWORD] is a strong random string. Use the same password on each of the servers. 2. Stop mysqld on both servers:

/sbin/service mysqld stop

1	/sbin/service mysqld stop

3. On each server, edit the /etc/my.cnf file. Look through the my.cnf file for an existing section of replication options. If none is found then just add to the end of the file. a. Find the “server-id” line and make sure that each server has a unique server-id. It does not matter what

HA Linux Cluster On RackSpace Cloud Servers

March 27, 2011

Our goal is to setup a pair of RackSpace Cloud Servers in a redundant cluster using a shared IP address. We’ll use the “heartbeat” package from Linux-HA (http://www.linux-ha.org) for the cluster messaging layer and “pacemaker” package from ClusterLabs (http://clusterlabs.org) for the cluster resource manager. Before starting this procedure you’ll need to: a. Create the two cloud servers. These instructions are specific to CentOS for the operating system. b. Open a ticket with RackSpace Cloud support and request a public IP address to be shared between the servers. You can use the instructions for other situations but you’ll need to make the appropriate adjustments. 1. Setup hosts file entries. On each server, edit /etc/hosts and add entries for each servers public and private interfaces. You’ll also find it convenient to setup ssh keys between the servers for easy access. 2. Now use yum to install some prerequisite packages:

yum install net-snmp perl-libwww-perl libesmtp perl-Net-SSLeay perl-MailTools ipvsadm OpenIPMI libibverbs librdmacm openhpi PyXML

1	yum install net-snmp perl-libwww-perl libesmtp perl-Net-SSLeay perl-MailTools ipvsadm OpenIPMI libibverbs librdmacm openhpi PyXML

Repeat this

Reverse Proxy With Content Rewrites

February 1, 2011

One of our clients is a school district that wanted to make content on http://www.khanacademy.org/ (KA) available to users without Youtube. The KA site uses a bit of javascript to test if the user’s browser can get to youtube and if not it servers videos from an different source. Unfortunately this was not working for the client. To solve the problem we setup mod_proxy as a reverse proxy. Then we used mod_headers, mod_filter and mod_substitute to rewrite the javascript going to client and force the use of the youtube alternative. Here’s the apache config that does the content rewrite. It’s located in the reverse proxy virtual host:

<pre class="code">
# do not accept gzip - gzip bypasses filters
RequestHeader unset Accept-Encoding

# declare that content should be filtered through mod_substitute
FilterDeclare NOTUBE
FilterProvider NOTUBE SUBSTITUTE Content-Type $text/html
FilterChain +NOTUBE

# change the javacript test so that it always failes
Substitute "s|youtube.com/favicon.ico|youtube.com/nfavicon.ico|in"