writing perl script to take mysql incremental backup with mysqldump

Question

I am dealing with an incremental backup solution for a mysql database in centos. I need to write a perl script to take incremental backup. then i will run this script by using crontabs. I am a bit confused. There are solutions but not really helping. I did lots of research. there are so many ways to take full backup and incremental backup for files. I can easily understand them but I need to take an incremental backup of a mysql database. I do not know how to do it. Can anyone help me either advising a source or a piece of code.

Answer 1

The incremental backup method you've been looking at is documented by MySQL here:

http://dev.mysql.com/doc/refman/5.0/en/binary-log.html

What you are essentially going to want to do is set up your mysql instance to write any changes to your database to this binary log. What this means is any updates, deletes, inserts etc go in the binary log, but not select statements (which don't change the db, therefore don't go in the binary log).

Once you have your mysql instance running with binary logging turned on, you take a full backup and take note of the master position. Then later on, to take an incremental backup, you want to run mysqlbinlog from the master position and the output of that will be all the changes made to your database since you took the full backup . You'll want to take note of the master position again at this point, so you know the point that you want to take the next incremental backup from.

Clearly, if you then take multiple incremental backups over and over, you need to retain all those incremental backups. I'd recommend taking a full backup quite often.

Indeed, I'd recommend always doing a full backup, if you can. Taking incremental backups is just going to cause you pain, IMO, but if you need to do it, that's certainly one way to do it.

Answer 2

mysqldump is the ticket.

Example:

mysqldump -u [user_name] -p[password] --database [database_name] >/tmp/databasename.sql

-u = mysql database user name

-p = mysql database password

Note: there is no space after the -p option. And if you have to do this in perl, then you can use the system function to call it like so:

system("mysqldump -u [user_name] -p[password] --database [database_name] >/tmp/databasename.sql") or die "system call failed: $?";

Be aware though of the security risks involved in doing this. If someone happened to do a listing of the current processes running on a system as this was running, they'd be able to see the credentials that were being used for database access.