Getting Started with MariaDB Galera Cluster

Galera Cluster Support in MariaDB Server

MariaDB Galera Cluster is powered by:

• MariaDB Server.
• The MySQL-wsrep patch for MySQL Server and MariaDB Server developed by Codership. The patch currently supports only Unix-like operating systems.
• The Galera wsrep provider library.

In MariaDB 10.1 and later, the MySQL-wsrep patch has been merged into MariaDB Server. This means that the functionality of MariaDB Galera Cluster can be obtained by installing the standard MariaDB Server packages and the Galera wsrep provider library package. The following Galera version corresponds to each MariaDB Server version:

• In MariaDB 10.4 and later, MariaDB Galera Cluster uses Galera 4. This means that the MySQL-wsrep patch is version 26 and the Galera wsrep provider library is version 4.
• In MariaDB 10.3 and before, MariaDB Galera Cluster uses Galera 3. This means that the MySQL-wsrep patch is version 25 and the Galera wsrep provider library is version 3.

See Deciphering Galera Version Numbers for more information about how to interpret these version numbers.

See What is MariaDB Galera Cluster?: Galera Versions for more information about which specific Galera version is included in each release of MariaDB Server.

In supported builds, Galera Cluster functionality can be enabled by setting some configuration options that are mentioned below. Galera Cluster functionality is not enabled in a standard MariaDB Server installation unless explicitly enabled with these configuration options.

Prerequisites

Swap Size Requirements

During normal operation a MariaDB Galera node does not consume much more memory than a regular MariaDB server. Additional memory is consumed for the certification index and uncommitted writesets, but normally this should not be noticeable in a typical application. There is one exception though:

1. Writeset caching during state transfer. When a node is receiving a state transfer it cannot process and apply incoming writesets because it has no state to apply them to yet. Depending on a state transfer mechanism (e.g. mysqldump) the node that sends the state transfer may not be able to apply writesets as well. Thus they need to cache those writesets for a catch-up phase. Currently the writesets are cached in memory and, if the system runs out of memory either the state transfer will fail or the cluster will block waiting for the state transfer to end.

To control memory usage for writeset caching, check the Galera parameters: gcs.recv_q_hard_limit, gcs.recv_q_soft_limit, and gcs.max_throttle.

Limitations

Before using MariaDB Galera Cluster, we would recommend reading through the known limitations, so you can be sure that it is appropriate for your application.

Installing MariaDB Galera Cluster

To use MariaDB Galera Cluster, there are two primary packages that you need to install:

1. A MariaDB Server version that supports Galera Cluster.

2. The Galera wsrep provider library.

As mentioned in the previous section, in MariaDB 10.1 and above, Galera Cluster support is actually included in the standard MariaDB Server packages. That means that installing MariaDB Galera Cluster package is the same as installing standard MariaDB Server package in those versions. However, you will also have to install an additional package to obtain the Galera wsrep provider library.

Some SST methods may also require additional packages to be installed. The mariabackup SST method is generally the best option for large clusters that expect a lot of load.

Installing MariaDB Galera Cluster with a Package Manager

MariaDB Galera Cluster can be installed via a package manager on Linux. In order to do so, your system needs to be configured to install from one of the MariaDB repositories.

You can configure your package manager to install it from MariaDB Corporation's MariaDB Package Repository by using the MariaDB Package Repository setup script.

You can also configure your package manager to install it from MariaDB Foundation's MariaDB Repository by using the MariaDB Repository Configuration Tool.

Installing MariaDB Galera Cluster with yum/dnf

On RHEL, CentOS, Fedora, and other similar Linux distributions, it is highly recommended to install the relevant RPM packages from MariaDB's repository using yum or dnf. Starting with RHEL 8 and Fedora 22, yum has been replaced by dnf, which is the next major version of yum. However, yum commands still work on many systems that use dnf.

To install MariaDB Galera Cluster with yum or dnf, follow the instructions at Installing MariaDB Galera Cluster with yum.

Installing MariaDB Galera Cluster with apt-get

On Debian, Ubuntu, and other similar Linux distributions, it is highly recommended to install the relevant DEB packages from MariaDB's repository using apt-get.

To install MariaDB Galera Cluster with apt-get, follow the instructions at Installing MariaDB Galera Cluster with apt-get.

Installing MariaDB Galera Cluster with zypper

On SLES, OpenSUSE, and other similar Linux distributions, it is highly recommended to install the relevant RPM packages from MariaDB's repository using zypper.

To install MariaDB Galera Cluster with zypper, follow the instructions at Installing MariaDB Galera Cluster with ZYpp.

Installing MariaDB Galera Cluster with a Binary Tarball

To install MariaDB Galera Cluster with a binary tarball, follow the instructions at Installing MariaDB Binary Tarballs.

Installing MariaDB Galera Cluster from Source

To install MariaDB Galera Cluster by compiling it from source, you will have to compile both MariaDB Server and the Galera wsrep provider library. For some information on how to do this, see the pages at Installing Galera From Source. The pages at Compiling MariaDB From Source and Galera Cluster Documentation: Building Galera Cluster for MySQL may also be helpful.

Configuring MariaDB Galera Cluster

A number of options need to be set in order for Galera Cluster to work when using MariaDB. See Configuring MariaDB Galera Cluster for more information.

Bootstrapping a New Cluster

To first node of a new cluster needs to be bootstrapped by starting mariadbd on that node with the option --wsrep-new-cluster option. This option tells the node that there is no existing cluster to connect to. The node will create a new UUID to identify the new cluster.

Do not use the --wsrep-new-cluster option when connecting to an existing cluster. Restarting the node with this option set will cause the node to create new UUID to identify the cluster again, and the node won't reconnect to the old cluster. See the next section about how to reconnect to an existing cluster.

For example, if you were manually starting mariadbd on a node, then you could bootstrap it by executing the following:

$ mariadbd --wsrep-new-cluster

However, keep in mind that most users are not going to be starting mariadbd manually. Instead, most users will use a service manager to start mariadbd. See the following sections on how to bootstrap a node with the most common service managers.

Systemd and Bootstrapping

On operating systems that use systemd, a node can be bootstrapped in the following way:

$ galera_new_cluster

This wrapper uses systemd to run mariadbd with the --wsrep-new-cluster option.

If you are using the systemd service that supports the systemd service's method for interacting with multiple MariaDB Server processes, then you can bootstrap a specific instance by specifying the instance name as a suffix. For example:

$ galera_new_cluster mariadb@node1

Systemd support and the galera_new_cluster script were added in MariaDB 10.1.

SysVinit and Bootstrapping

On operating systems that use sysVinit, a node can be bootstrapped in the following way:

$ service mysql bootstrap

This runs mariadbd with the --wsrep-new-cluster option.

Adding Another Node to a Cluster

Once you have a cluster running and you want to add/reconnect another node to it, you must supply an address of one or more of the existing cluster members in the wsrep_cluster_address option. For example, if the first node of the cluster has the address 192.168.0.1, then you could add a second node to the cluster by setting the following option in a server option group in an option file:

[mariadb]
...
wsrep_cluster_address=gcomm://192.168.0.1  # DNS names work as well, IP is preferred for performance

The new node only needs to connect to one of the existing cluster nodes. Once it connects to one of the existing cluster nodes, it will be able to see all of the nodes in the cluster. However, it is generally better to list all nodes of the cluster in wsrep_cluster_address, so that any node can join a cluster by connecting to any of the other cluster nodes, even if one or more of the cluster nodes are down. It is even OK to list a node's own IP address in wsrep_cluster_address, since Galera Cluster is smart enough to ignore it.

Once all members agree on the membership, the cluster's state will be exchanged. If the new node's state is different from that of the cluster, then it will request an IST or SST to make itself consistent with the other nodes.

Restarting the Cluster

If you shut down all nodes at the same time, then you have effectively terminated the cluster. Of course, the cluster's data still exists, but the running cluster no longer exists. When this happens, you'll need to bootstrap the cluster again.

If the cluster is not bootstrapped and mariadbd on the first node is just started normally, then the node willl try to connect to at least one of the nodes listed in the wsrep_cluster_address option. If no nodes are currently running, then this will fail. Bootstrapping the first node solves this problem.

Determining the Most Advanced Node

In some cases Galera will refuse to bootstrap a node if it detects that it might not be the most advanced node in the cluster. Galera makes this determination if the node was not the last one in the cluster to be shut down or if the node crashed. In those cases, manual intervention is needed.

If you know for sure which node is the most advanced you can edit the grastate.dat file in the datadir. You can set safe_to_bootstrap=1 on the most advanced node.

You can determine which node is the most advanced by checking grastate.dat on each node and looking for the node with the highest seqno. If the node crashed and seqno=-1, then you can find the most advanced node by recovering the seqno on each node with the wsrep_recover option. For example:

$ mariadbd --wsrep_recover

Systemd and Galera Recovery

On operating systems that use systemd, the position of a node can be recovered by running the galera_recovery script. For example:

$ galera_recovery

If you are using the systemd service that supports the systemd service's method for interacting with multiple MariaDB Server processes, then you can recover the position of a specific instance by specifying the instance name as a suffix. For example:

$ galera_recovery mariadb@node1

The galera_recovery script recovers the position of a node by running mariadbd with the wsrep_recover option.

When the galera_recovery script runs mariadbd, it does not write to the error log. Instead, it redirects mariadbd log output to a file named with the format /tmp/wsrep_recovery.XXXXXX, where XXXXXX is replaced with random characters.

When Galera is enabled, MariaDB's systemd service automatically runs the galera_recovery script prior to starting MariaDB, so that MariaDB starts with the proper Galera position.

Support for systemd and the galera_recovery script were added in MariaDB 10.1.

State Snapshot Transfers (SSTs)

In a State Snapshot Transfer (SST), the cluster provisions nodes by transferring a full data copy from one node to another. When a new node joins the cluster, the new node initiates a State Snapshot Transfer to synchronize its data with a node that is already part of the cluster.

See Introduction to State Snapshot Transfers (SSTs) for more information.

Incremental State Transfers (ISTs)

In an Incremental State Transfer (SST), the cluster provisions nodes by transferring a node's missing writesets from one node to another. When a new node joins the cluster, the new node initiates a Incremental State Transfer to synchronize its data with a node that is already part of the cluster.

If a node has only been out of a cluster for a little while, then an IST is generally faster than an SST.

Data at Rest Encryption

In MariaDB 10.1 and above, MariaDB Galera Cluster supports Data at Rest Encryption. See SSTs and Data at Rest Encryption for some disclaimers on how SSTs are affected when encryption is configured.

Some data still cannot be encrypted:

• The disk-based Galera gcache is not encrypted (MDEV-8072).

Monitoring

Status Variables

Galera Cluster's status variables can be queried with the standard SHOW STATUS command. For example:

SHOW GLOBAL STATUS LIKE 'wsrep_%';

Cluster Change Notifications

The cluster nodes can be configured to invoke a command when cluster membership or node status changes. This mechanism can also be used to communicate the event to some external monitoring agent. This is configured by setting wsrep_notify_cmd. See Galera Cluster documentation: Notification Command for more information.

See Also

• What is MariaDB Galera Cluster?
• About Galera Replication
• Galera Use Cases
• Codership on Google Groups
• Galera Cluster documentation
• Galera Cluster documentation: Notification Command
• Introducing the “Safe-To-Bootstrap” feature in Galera Cluster
• Github - galera
• Github - mysql-wsrep

Footnotes