The Library

Related Documents

Other Resources

Training Video Exercises

Making Back-Ups with Galera Cluster

These exercises are part of the training video, Making Back-Ups with Galera Cluster. There are exercises related to what was taught in each section, except for the section entitled Back-Up & Restoration Plan. That particular section is in a sense another set of exercises.

Before starting these exercises, make sure you have all of the requirements and preparations in place.

Requirements & Preparation

  • Test Servers: 4
  • Operating System: Linux
  • Open Ports: TCP 22, TCP 3306. TCP 4444, TCP & UDP 4567, TCP 4568
  • Database Software: MySQL or MariaDB (vs. 10.4 or higher)
  • Other Software: * Galera (Only 3)

Student Materials

Before starting an exercise, read it fully and carefully. The headings for each set of exercises before correspond to the section with the same name in training video.

Back-Up Basics

Do these exercises after viewing the first two sections of the training video: Galera Cluster Overview, and Installing Software on Nodes. Don’t configure the nodes until the next section.

  1. Download the sample database called, company from the Coderhsip site (see margin) onto one of the nodes. It’s a dump file made with mysqldump. With MySQL or MariaDB and Galera running, use the mysql client to load the data on one node. When finished, check the other nodes to ensure they have replicated the company database. Delete the dump file.
  2. Make a directory called, backups, and a sub-directory in it called, temp. Use the SET statement to desync a different node. Execute a SHOW STATUS statement to check that the node has a status of Desync. Execute SHOW VARIABLES to determine the data directory. Next, use rsync to back-up of all of the database to the temp sub-directory, in the backups directory. When it’s finished, use tar to create a zipped archive file (e.g., db-backup.tgz). Disable desync on the node.
  3. On one of the Galera nodes, use the DROP DATABASE statement to drop the company database. Check to see that it’s replicated on the other two nodes. Now shutdown all of the nodes. When they’ve all stopped, use the back-up file to restore the data directory on the node where it’s located. Then start that node and check if the company database is back. After that, start the other two nodes. Give them a couple of minutes before checking that the company database has been restored on them.

Using Standard Replication

This set of exercises require all four servers mentioned in the requirements above: three with MySQL or MariaDB, and Galera installed and running; the fourth server with only MySQL or MariaDB.

  1. On one of the Galera nodes, configure it also to use standard replication, to be a master. See the link in at the top under Student Materials for an example configuration file. On the fourth server, the one without Galera Cluster, configure it to use standard replication, to be a slave to the master. If you’re unfamiliar with standard replication, you might watch our training video on Standard Replication and Galera Cluster—or read the related article on the same topic.
  2. Create a user with REPLICATION CLIENT on the Galera node you’ve designated to be a master. Use mysqldump to make a copy of all of the databases on the master. Be sure to use the --flush-logs and --master-data options.
  3. Use the scp command to copy the dump file you created on the master, to the slave. Load the data on the slave, using the mysql client. Execute the CHANGE MASTER statement to provide the IP address and port 3306, as well as the replication user name and password. Then start replication by executing the START SLAVE statement on the slave. Check that replication is running fine with SHOW SLAVE STATUS on the slave. Try changing some data on the master and see if it replicates to the slave.
  4. Make a directory called backups, with a sub-directory within it called temp. Stop the slave and use mysqldump to make a back-up on it to a the backups, temp sub-directory. Copy the binary log files, as well as the MySQL configuration file to the temporary back-up sub-directory. Then tar and zip the back-up files. Start the slave again and make sure replication is running.

Using Galera Arbitrator

For the exercises in this section and the remaining sections, you won’t need the replication slave server. You can shut it down. You’ll only use the three servers with Galera installed on them.

  1. Create a configuration file for Galera Arbitrator (see link at the top under Student Materials for an example). Write a simple back-up script that will use rsync—or use the one linked above. Execute it using garbd from the command-line. As soon as it starts, check the status of the donor node for comments to see if it’s desynced during the back-up.
  2. Write another simple back-up script that will use mysqldump and execute it. Make sure it gives a different name to the back-up file than in the previous exercise.

Restoring Nodes and a Cluster

  1. Drop the company database on one of the nodes. Check that the deletion of the database has occurred on all of the Galera nodes and the slave. Choose one Galera node for restoration. Shutdown the mysqld daemon on the other two nodes.
  2. Using the back-up you made with rsync, in the previous set of exercises, restore the data to one of the nodes. Start the other Galera nodes, one at a time. Verify that they’ve been restored.
  3. Drop the company database again. This time restore the data with the dump file, using the mysql client. After a reasonable amount of time, verify that they’ve all been restored.

Note

If you’re unfamiliar with how to install Galera Cluster, watch our training video on installing Galera with MySQL, or MariaDB (see the margin for links) or read the related articles. You’ll need to configure MySQL or MariaDB and Galera Cluster and start them. There’s a link in the margin for the recommended configuration for certain sections here.

Related Documents

Other Resources