The script works by pulling the latest data and database backup from the production server.  The script can be placed in a cron. The DR server user should be [...]]]> Here is a script I wrote to keep 2 or more web sites in sync.  The data files can be anything under the data tree.
The database is MySQL.

The script works by pulling the latest data and database backup from the production server.  The script can be placed in a cron.
The DR server user should be able to ssh without a password to the production server.  This is accomplished using RSA keys under SSH.

Once both the web site and database are setup, this script pulls the newest files, and latest mysql database backup via rsync.

Change the First set of VARIABLES, and <usernames> and <pathing> where needed.

I should probably have more information on the script here, but I wanted to save a copy somewhere before I forgot about it.

Anyway, my thought is if I setup a primary web server somewhere and want n+1 backend systems I might be able to use this poor-man’s approach.  If the backed systems need to have database updates then you are looking at database replication.  But, the rsync file copy and all of the rest could be used to make web server production updates to a cluster of systems.  You just update the main instance and all of the slave servers pull the latest information, including a database update script if needed., unix, script

#!/bin/ksh -x

#######################################
#
# Script to sync Prod web with DR server.
# The script is designed to pull from the
# production server.
#
# 12/2008:  Initial setup.
#        :  Mike Byers
#######################################

PRIME_HOST=’prod.edmz.com’                  #Production host
DR_HOST=’dr-orprod2.edmz.com’              #DR server
DATABASES=”mysql DB1 DB2″                  #database to transfer space sepa
rated
#MYSQLUSER=backup                               #User for database backup
MYSQLUSER=restore                                  #User for database restore
#MYSQLPWD=’ABC123′                             #Backup user Password
MYSQLPWD=’ABC123′                               #Restore user Password (Restore user needs more DB permissions)
DB_PATH=’/opt/mysql5/backups’                 #Path to database backup files
#YESTERDAY=`date +%Y%m%d -d”1 day ago”`         #Linux Version
YESTERDAY=`TZ=GMT+24 date +%Y%m%d`              #Solaris Version
TODAY=`date +%Y%m%d`
PATH=$PATH:/opt/mysql5/bin/;export PATH  #Path to mysql binaries

/usr/bin/rsync –exclude ‘configuration.*’ \
–exclude ‘**/logs’ \
–exclude ‘**/modules’ \
–exclude ‘**/conf’ \
–exclude ‘**/bin’ \
-vaue ’ssh -x -C -l <username>’ $PRIME_HOST:apache/htdocs/* /<path>/apache/htdocs/

for DB in $DATABASES
do
/usr/bin/rsync -vaue ’ssh -x -C -l <username>’ $PRIME_HOST:$DB_PATH/$DB-$TODAY.sql.
gz $DB_PATH
#Compare sizes of database backup files to see if anything changed
S1=`du -sk $DB_PATH/$DB-$YESTERDAY.sql.gz | awk ‘{ print $1 }’`
S2=`du -sk $DB_PATH/$DB-$TODAY.sql.gz | awk ‘{ print $1 }’`
if [[ S1 -eq S2 ]]; then
echo “No update for database $DB needed”
else
/bin/gunzip < $DB_PATH/$DB-$TODAY.sql.gz | mysql -u $MYSQLUSER -p$MYSQLPWD
$DB
fi
done

I need a really tricky find command the other day so I came up with the following.  I wanted to save the command some where as it really shows the power of find

Find all files that are not in the [...]]]> A thread dedicated to those useful Unix commands we all know and love.  Ya right.

I need a really tricky find command the other day so I came up with the following.  I wanted to save the command some where as it really shows the power of find

Find all files that are not in the images or languages directory that are older than seven days that end with .php or .doc and delete them also leave a log file of what was deleted.

find ./  \( -name images -o -name languages \) -prune -o \( -name ‘*.php’ -o -name ‘*.doc’ \) -mtime +7 -exec rm {} \; -print > /var/tmp/deleted_files

Sorry the command is all on one line.  Because the find statement is utilizing regular expressions and complex statements, ie “\(“, so you can’t use the “\” and take the statement to the next line.

To read the command better take it one step at a time.
Find all files or directories that have the name “images” OR the name “languages” and prune/remove those items found from the find results.   Using the -prune option is a great way to setup exclusions.  The next “-o” OR says if the files found are not named “images” or “languages” find the files that end in “.php” or “.doc” that are older than 7 days.  If anything is found that meets the conditions then remove those files and “print” the file name out to the log file.

This set of commands was to be used to change directory names inside a samba share, hense the directory names of \ and the unix escape needed making a bunch of \\ [...]]]> I just helped a friend out with placing a shell variable into a sed command and thought it would be prudent to document it.

This set of commands was to be used to change directory names inside a samba share, hense the directory names of \ and the unix escape needed making a bunch of \\ appear.

server_name#  DIR=”blahblah”
server_name#  echo “bob” | sed “s/^/${DIR}\\\ADMIND__${DIR}\\\/”
blahblah\ADMIND__blahblah\bob

Set the variable DIR to “blahblah”, I should have just used “foo” it’s less to type.  Now echo the word “bob” but first add our variable and a prefix of “ADMIND__” to the front of bob.

I wanted to show another example, but don’t have time at the momment.

The important things to remember is to have the variables enclosed in double quotes.