Chapter 21 InnoDB Cluster

MySQL Shell includes the AdminAPI, which is accessed through the dba global variable and its associated methods. The dba variable's methods enable you to deploy, configure, and administer InnoDB clusters. For example, use the dba.createCluster() method to create an InnoDB cluster.

MySQL Shell provides online help for the AdminAPI. To list all available dba commands, use the dba.help() method. For online help on a specific method, use the general format object.help('methodname'). For example:

mysql-js> dba.help('getCluster')

Retrieves a cluster from the Metadata Store.

SYNTAX

  dba.getCluster([name][, options])

WHERE

  name: Parameter to specify the name of the cluster to be returned.
  options: Dictionary with additional options.

RETURNS

  The cluster object identified by the given name or the default cluster.

DESCRIPTION

If name is not specified or is null, the default cluster will be returned.

If name is specified, and no cluster with the indicated name is found, an error
will be raised.

The options dictionary accepts the connectToPrimary option,which defaults to
true and indicates the shell to automatically connect to the primary member of
the cluster.

EXCEPTIONS

  MetadataError in the following scenarios:

   - If the Metadata is inaccessible.
   - If the Metadata update operation failed.

  ArgumentError in the following scenarios:

   - If the Cluster name is empty.
   - If the Cluster name is invalid.
   - If the Cluster does not exist.

  RuntimeError in the following scenarios:

   - If the current connection cannot be used for Group Replication.

To test if high availability works, simulate an unexpected halt by killing an instance. The cluster detects the fact that the instance left the cluster and reconfigures itself. Exactly how the cluster reconfigures itself depends on whether you are using a single-primary or multi-primary cluster, and the role the instance serves within the cluster.

In single-primary mode:

For more information see Section 18.1.3.1, “Single-Primary Mode”.

In multi-primary mode:

For more information see Section 18.1.3.2, “Multi-Primary Mode”.

There are various ways to simulate an instance leaving a cluster, for example you can forcibly stop the MySQL server on an instance, or use the AdminAPI dba.killSandboxInstance() if testing a sandbox deployment. In this example assume there is a single-primary sandbox cluster deployment with three server instances and the instance listening at port 3310 is the current primary. Simulate the instance leaving the cluster unexpectedly:

mysql-js> dba.killSandboxInstance(3310)

The cluster detects the change and elects a new primary automatically. Assuming your session is connected to port 6446, the default read-write classic MySQL protocol port, MySQL Router should detect the change to the cluster's topology and redirect your session to the newly elected primary. To verify this, switch to SQL mode in MySQL Shell using the \sql command and select the instance's port variable to check which instance your session has been redirected to. Notice that the first SELECT statement fails as the connection to the original primary was lost. This means the current session has been closed, MySQL Shell automatically reconnects for you and when you issue the command again the new port is confirmed.

mysql-js> \sql
Switching to SQL mode... Commands end with ;
mysql-sql> SELECT @@port;
ERROR: 2013 (HY000): Lost connection to MySQL server during query
The global session got disconnected.
Attempting to reconnect to 'root@localhost:6446'...
The global session was successfully reconnected.
mysql-sql> SELECT @@port;
+--------+
| @@port |
+--------+
|   3330 |
+--------+
1 row in set (0.00 sec)

In this example, the instance at port 3330 has been elected as the new primary. This shows that the InnoDB cluster provided us with automatic failover, that MySQL Router has automatically reconnected us to the new primary instance, and that we have high availability.

When MySQL Router is bootstrapped against a cluster, it records the server instance's addresses in its configuration file. If any additional instances are added to the cluster after bootstrapping the MySQL Router, they are not automatically detected and therefore are not used for connection routing.

To ensure that newly added instances are routed to correctly you must bootstrap MySQL Router against the cluster to read the updated metadata. This means that you must restart MySQL Router and include the --bootstrap option.

Before creating a production deployment from server instances you need to check that MySQL on each instance is correctly configured. In addition to dba.configureInstance(), which checks the configuration as part of configuring an instance, you can use the dba.checkInstanceConfiguration() function. This ensures that the instance satisfies the Section 21.2.2, “InnoDB Cluster Requirements” without changing any configuration on the instance. This does not check any data that is on the instance, see Checking Instance State for more information. The following demonstrates issuing this in a running MySQL Shell:

mysql-js> dba.checkInstanceConfiguration('ic@ic-1:3306')
Please provide the password for 'ic@ic-1:3306': ***
Validating MySQL instance at ic-1:3306 for use in an InnoDB cluster...

This instance reports its own address as ic-1
Clients and other cluster members will communicate with it through this address by default.
If this is not correct, the report_host MySQL system variable should be changed.

Checking whether existing tables comply with Group Replication requirements...
No incompatible tables detected

Checking instance configuration...

Some configuration options need to be fixed:
+--------------------------+---------------+----------------+--------------------------------------------------+
| Variable                 | Current Value | Required Value | Note                                             |
+--------------------------+---------------+----------------+--------------------------------------------------+
| binlog_checksum          | CRC32         | NONE           | Update the server variable                       |
| enforce_gtid_consistency | OFF           | ON             | Update read-only variable and restart the server |
| gtid_mode                | OFF           | ON             | Update read-only variable and restart the server |
| server_id                | 1             |                | Update read-only variable and restart the server |
+--------------------------+---------------+----------------+--------------------------------------------------+

Please use the dba.configureInstance() command to repair these issues.

{
    "config_errors": [
        {
            "action": "server_update",
            "current": "CRC32",
            "option": "binlog_checksum",
            "required": "NONE"
        },
        {
            "action": "restart",
            "current": "OFF",
            "option": "enforce_gtid_consistency",
            "required": "ON"
        },
        {
            "action": "restart",
            "current": "OFF",
            "option": "gtid_mode",
            "required": "ON"
        },
        {
            "action": "restart",
            "current": "1",
            "option": "server_id",
            "required": ""
        }
    ],
    "status": "error"
}

Repeat this process for each server instance that you plan to use as part of your cluster. The report generated after running dba.checkInstanceConfiguration() provides information about any configuration changes required before you can proceed. The action field in the config_error section of the report tells you whether MySQL on the instance requires a restart to detect any change made to the configuration file.

Instances which do not support persisting configuration changes automatically (see Persisting Settings) require you to connect to the server, run MySQL Shell, connect to the instance locally and issue dba.configureLocalInstance(). This enables MySQL Shell to modify the instance's option file after running the following commands against a remote instance:

The recommended method is to log in to the remote machine, for example using SSH, run MySQL Shell as the root user and then connect to the local MySQL server. For example, use the --uri option to connect to the local instance:

shell> sudo -i mysqlsh --uri=instance

Alternatively use the \connect command to log in to the local instance. Then issue dba.configureInstance(instance), where instance is the connection information to the local instance, to persist any changes made to the local instance's option file.

mysql-js> dba.configureLocalInstance('ic@ic-2:3306')
	

Repeat this process for each instance in the cluster which does not support persisting configuration changes automatically. For example if you add 2 instances to a cluster which do not support persisting configuration changes automatically, you must connect to each server and persist the configuration changes required for InnoDB cluster before the instance restarts. Similarly if you modify the cluster structure, for example changing the number of instances, you need to repeat this process for each server instance to update the InnoDB cluster metadata accordingly for each instance in the cluster.

When you create a cluster using dba.createCluster(), the operation returns a Cluster object which can be assigned to a variable. You use this object to work with the cluster, for example to add instances or check the cluster's status. If you want to retrieve a cluster again at a later date, for example after restarting MySQL Shell, use the dba.getCluster([name],[options]) function. For example:

mysql-js> var cluster1 = dba.getCluster()

If you do not specify a cluster name then the default cluster is returned. By default MySQL Shell attempts to connect to the primary instance of the cluster when you use dba.getCluster(). Set the connectToPrimary option to configure this behavior. If connectToPrimary is true and the active global MySQL Shell session is not to a primary instance, the cluster is queried for the primary member and the cluster object connects to it. If there is no quorum in the cluster, the operation fails. If connectToPrimary is false, the cluster object uses the active session, in other words the same instance as the MySQL Shell's current global session. If connectToPrimary is not specified, MySQL Shell treats connectToPrimary as true, and falls back to connectToPrimary being false.

To force connecting to a secondary when getting a cluster, establish a connection to the secondary member of the cluster and use the connectToPrimary option by issuing:

mysql-js> shell.connect(secondary_member)
mysql-js> var cluster1 = dba.getCluster(testCluster, {connectToPrimary:false})

To get information about the structure of the InnoDB cluster itself, use the Cluster.describe() function:

mysql-js> cluster.describe();
{
    "clusterName": "testCluster",
    "defaultReplicaSet": {
        "name": "default",
        "topology": [
            {
                "address": "ic-1:3306",
                "label": "ic-1:3306",
                "role": "HA"
            },
            {
                "address": "ic-2:3306",
                "label": "ic-2:3306",
                "role": "HA"
            },
            {
                "address": "ic-3:3306",
                "label": "ic-3:3306",
                "role": "HA"
            }
        ]
    }
}

The output from this function shows the structure of the InnoDB cluster including all of its configuration information, and so on. The address, label and role values match those described at Checking a cluster's Status with Cluster.status() .

Cluster objects provide the status() method that enables you to check how a cluster is running. Before you can check the status of the InnoDB cluster, you need to get a reference to the InnoDB cluster object by connecting to any of its instances. However, if you want to make changes to the configuration of the cluster, you must connect to a "R/W" instance. Issuing status() retrieves the status of the cluster based on the view of the cluster which the server instance you are connected to is aware of and outputs a status report.

For information about how the InnoDB cluster is running, use the cluster's status() method:

mysql-js> var cluster = dba.getCluster()
mysql-js> cluster.status()
{
    "clusterName": "testcluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "primary": "ic-1:3306", 
        "ssl": "REQUIRED", 
        "status": "OK", 
        "statusText": "Cluster is ONLINE and can tolerate up to ONE failure.", 
        "topology": {
            "ic-1:3306": {
                "address": "ic-1:3306", 
                "mode": "R/W", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "ic-2:3306": {
                "address": "ic-2:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }, 
            "ic-3:3306": {
                "address": "ic-3:3306", 
                "mode": "R/O", 
                "readReplicas": {}, 
                "role": "HA", 
                "status": "ONLINE"
            }
        }
    }, 
    "groupInformationSourceMember": "mysql://ic@ic-1:3306"
}

The output of Cluster.status() provides the following information:

To display more information about the cluster use the extended option. From version 8.0.17, the extended option supports integer or Boolean values. To configure the additional information that Cluster.status({'extended':value}) provides, use the following values:

Setting extended using Boolean values is the equivalent of setting the integer values 0 and 1. In versions prior to 8.0.17, the extended option was only Boolean. Similarly prior versions used the queryMembers Boolean option to provide more information about the instances in the cluster, which is the equivalent of setting extended to 3. The queryMembers option is deprecated and scheduled to be removed in a future release.

When you issue Cluster.status({'extended':1}), or the extended option is set to true, the output includes:

To see information about recovery and regular transaction I/O, applier worker thread statistics and any lags; applier coordinator statistics, if parallel apply is enabled; error, and other information from I/O and applier threads issue use the values 2 and 3. A value of 3 is the equivalent of setting the deprecated queryMembers option to true. When you use these values, a connection to each instance in the cluster is opened so that additional instance specific statistics can be queried. The exact statistics that are included in the output depend on the state and configuration of the instance and the server version. This information matches that shown in the replication_group_member_stats table, see the descriptions of the matching columns for more information. Instances which are ONLINE have a transactions section included in the output. Instances which are RECOVERING have a recovery section included in the output. When you set extended to 2, in either case, these sections can contain the following:

When you set extended to 3, the connection section shows information from the replication_connection_status table. The connection section can contain the following:

The currentlyQueueing section has information about the transactions currently queued:

The lastQueued section has information about the most recently queued transaction:

Instances which are using a multithreaded slave have a workers section which contains information about the worker threads, and matches the information shown by the replication_applier_status_by_worker table.

The lastApplied section shows the following information about the last transaction applied by the worker:

The currentlyApplying section shows the following information about the transaction currently being applied by the worker:

The lastProcessed section has the following information about the last transaction processed by the worker:

If parallel applier workers are enabled, then the number of objects in the workers array in transactions or recovery matches the number of configured workers and an additional coordinator object is included. The information shown matches the information in the replication_applier_status_by_coordinator table. The object can contain:

The currentlyProcessing section has the following information about the transaction being processed by the worker:

worker objects have the following information if an error was detected in the replication_applier_status_by_worker table:

connection objects have the following information if an error was detected in the replication_connection_status table:

coordinator objects have the following information if an error was detected in the replication_applier_status_by_coordinator table:

The output of Cluster.status() shows information about the progress of recovery operations for instances in RECOVERING state. Information is shown for instances recovering using either MySQL Clone, or incremental recovery. Monitor these fields:

From MySQL 8.0.16, Group Replication has the concept of a communication protocol for the group, see Section 18.4.1.4, “Setting a Group's Communication Protocol Version” for background information. The Group Replication communication protocol version usually has to be managed explicitly, and set to accommodate the oldest MySQL Server version that you want the group to support. However, InnoDB cluster automatically and transparently manages the communication protocol versions of its members, whenever the cluster topology is changed using AdminAPI operations. A cluster always uses the most recent communication protocol version that is supported by all the instances that are currently part of the cluster or joining it.

To see the communication protocol version being used in a cluster, use the Cluster.status() function with the extended option enabled. The communication protocol version is returned in the GRProtocolVersion field, provided that the cluster has quorum and no cluster members are unreachable.

The following operations can report information about the MySQL Server version running on the instance:

The behavior varies depending on the MySQL Server version of the Cluster object session.

Whenever Group Replication stops, the super_read_only variable is set to ON to ensure no writes are made to the instance. When you try to use such an instance with the following AdminAPI commands you are given the choice to set super_read_only=OFF on the instance:

When AdminAPI encounters an instance which has super_read_only=ON, in interactive mode you are given the choice to set super_read_only=OFF. For example:

mysql-js> var myCluster = dba.dropMetadataSchema()
Are you sure you want to remove the Metadata? [y/N]: y
The MySQL instance at 'localhost:3310' currently has the super_read_only system
variable set to protect it from inadvertent updates from applications. You must
first unset it to be able to perform any changes to this instance.
For more information see:
https://dev.mysql.com/doc/refman/en/server-system-variables.html#sysvar_super_read_only.

Do you want to disable super_read_only and continue? [y/N]: y

Metadata Schema successfully removed. 

The number of current active sessions to the instance is shown. You must ensure that no applications can write to the instance inadvertently. By answering y you confirm that AdminAPI can write to the instance. If there is more than one open session to the instance listed, exercise caution before permitting AdminAPI to set super_read_only=OFF.

To force the function to set super_read_only=OFF in a script, pass the clearReadOnly option set to true. For example dba.configureInstance(instance, {clearReadOnly: true}).

The recommended way to create a user which can administer an InnoDB cluster is to use the clusterAdmin and clusterAdminPassword options with the dba.configureInstance() or dba.configureLocalInstance() operations. If you want to manually configure a user which can administer an InnoDB cluster that user requires the following privileges, all with GRANT OPTION:

If only read operations are needed, for example to create a user for monitoring purposes, an account with more restricted privileges can be used. To give the user your_user the privileges needed to monitor InnoDB cluster issue:

GRANT SELECT ON mysql_innodb_cluster_metadata.* TO your_user@'%';
GRANT SELECT ON performance_schema.global_status TO your_user@'%';
GRANT SELECT ON performance_schema.replication_applier_configuration TO your_user@'%';
GRANT SELECT ON performance_schema.replication_applier_status TO your_user@'%';
GRANT SELECT ON performance_schema.replication_applier_status_by_coordinator TO your_user@'%';
GRANT SELECT ON performance_schema.replication_applier_status_by_worker TO your_user@'%';
GRANT SELECT ON performance_schema.replication_connection_configuration TO your_user@'%';
GRANT SELECT ON performance_schema.replication_connection_status TO your_user@'%';
GRANT SELECT ON performance_schema.replication_group_member_stats TO your_user@'%';
GRANT SELECT ON performance_schema.replication_group_members TO your_user@'%';
GRANT SELECT ON performance_schema.threads TO your_user@'%' WITH GRANT OPTION; 

For more information, see Section 13.7.1, “Account Management Statements”.

Instances running MySQL 8.0.16 and later support the Group Replication automatic rejoin functionality, which enables you to configure instances to automatically rejoin the cluster after being expelled. See Section 18.6.6, “Responses to Failure Detection and Network Partitioning” for background information. AdminAPI provides the autoRejoinTries option to configure the number of tries instances make to rejoin the cluster after being expelled. By default instances do not automatically rejoin the cluster. You can configure the autoRejoinTries option at either the cluster level or for an individual instance using the following commands:

The autoRejoinTries option accepts positive integer values between 0 and 2016 and the default value is 0, which means that instances do not try to automatically rejoin. When you are using the automatic rejoin functionality, your cluster is more tolerant to faults, especially temporary ones such as unreliable networks. But if quorum has been lost, you should not expect members to automatically rejoin the cluster, because majority is required to rejoin instances.

Instances running MySQL version 8.0.12 and later have the group_replication_exit_state_action variable, which you can configure using the AdminAPI exitStateAction option. This controls what instances do in the event of leaving the cluster unexpectedly. By default the exitStateAction option is READ_ONLY, which means that instances which leave the cluster unexpectedly become read-only. If exitStateAction is set to OFFLINE_MODE (available from MySQL 8.0.18), instances which leave the cluster unexpectedly become read-only and also enter offline mode, where they disconnect existing clients and do not accept new connections (except from clients with administrator privileges). If exitStateAction is set to ABORT_SERVER then in the event of leaving the cluster unexpectedly, the instance shuts down MySQL, and it has to be started again before it can rejoin the cluster. Note that when you are using the automatic rejoin functionality, the action configured by the exitStateAction option only happens in the event that all attempts to rejoin the cluster fail.

There is a chance you might connect to an instance and try to configure it using the AdminAPI, but at that moment the instance could be rejoining the cluster. This could happen whenever you use any of these operations:

These operations might provide extra information while the instance is automatically rejoining the cluster. In addition, when you are using Cluster.removeInstance(), if the target instance is automatically rejoining the cluster the operation aborts unless you pass in force:true.

Once a sandbox instance is running, it is possible to change its status at any time using the following:

You can remove an instance from a cluster at any time should you wish to do so. This can be done with the Cluster.removeInstance(instance) method, as in the following example:

mysql-js> cluster.removeInstance('root@localhost:3310')

The instance will be removed from the InnoDB cluster. Depending on the instance
being the Seed or not, the Metadata session might become invalid. If so, please
start a new session to the Metadata Storage R/W instance.

Attempting to leave from the Group Replication group...

The instance 'localhost:3310' was successfully removed from the cluster.

You can optionally pass in the interactive option to control whether you are prompted to confirm the removal of the instance from the cluster. In interactive mode, you are prompted to continue with the removal of the instance (or not) in case it is not reachable. The cluster.removeInstance() operation ensures that the instance is removed from the metadata of all the cluster members which are ONLINE, and the instance itself.

When the instance being removed has transactions which still need to be applied, AdminAPI waits for up to the number of seconds configured by the MySQL Shell dba.gtidWaitTimeout option for transactions (GTIDs) to be applied. The MySQL Shell dba.gtidWaitTimeout option has a default value of 60 seconds, see Configuring MySQL Shell Options for information on changing the default. If the timeout value defined by dba.gtidWaitTimeout is reached when waiting for transactions to be applied and the force option is false (or not defined) then an error is issued and the remove operation is aborted. If the timeout value defined by dba.gtidWaitTimeout is reached when waiting for transactions to be applied and the force option is set to true then the operation continues without an error and removes the instance from the cluster.

When you create a cluster and add instances to it, values such as the group name, the local address, and the seed instances are configured automatically by AdminAPI. These default values are recommended for most deployments, but advanced users can override the defaults by passing the following options to the dba.createCluster() and Cluster.addInstance().

To customize the name of the replication group created by InnoDB cluster, pass the groupName option to the dba.createCluster() command. This sets the group_replication_group_name system variable. The name must be a valid UUID.

To customize the address which an instance provides for connections from other instances, pass the localAddress option to the dba.createCluster() and cluster.addInstance() commands. Specify the address in the format host:port. This sets the group_replication_local_address system variable on the instance. The address must be accessible to all instances in the cluster, and must be reserved for internal cluster communication only. In other words do not use this address for communication with the instance.

To customize the instances used as seeds when an instance joins the cluster, pass the groupSeeds option to the dba.createCluster() and cluster.addInstance() commands. Seed instances are contacted when a new instance joins a cluster and used to provide data to the new instance. The addresses are specified as a comma separated list such as host1:port1,host2:port2. This configures the group_replication_group_seeds system variable.

For more information see the documentation of the system variables configured by these AdminAPI options.

If an instance leaves the cluster, for example because it lost connection, and for some reason it could not automatically rejoin the cluster, it might be necessary to rejoin it to the cluster at a later stage. To rejoin an instance to a cluster issue Cluster.rejoinInstance(instance).

In the case where an instance has not had it's configuration persisted (see Persisting Settings), upon restart the instance does not rejoin the cluster automatically. The solution is to issue cluster.rejoinInstance() so that the instance is added to the cluster again and ensure the changes are persisted. Once the InnoDB cluster configuration is persisted to the instance's option file it rejoins the cluster automatically.

If you are rejoining an instance which has changed in some way then you might have to modify the instance to make the rejoin process work correctly. For example, when you restore a MySQL Enterprise Backup backup, the server_uuid changes. Attempting to rejoin such an instance fails because InnoDB cluster instances are identified by the server_uuid variable. In such a situation, information about the instance's old server_uuid must be removed from the InnoDB cluster metadata and then a Cluster.rescan() must be executed to add the instance to the metadata using it's new server_uuid. For example:

cluster.removeInstance("root@instanceWithOldUUID:3306", {force: true})

cluster.rescan()

In this case you must pass the force option to the Cluster.removeInstance() method because the instance is unreachable from the cluster's perspective and we want to remove it from the InnoDB cluster metadata anyway.

If an instance (or instances) fail, then a cluster can lose its quorum, which is the ability to vote in a new primary. This can happen when a there is a failure of enough instances that there is no longer a majority of the instances which make up the cluster to vote on Group Replication operations. When a cluster loses quorum you can no longer process write transactions with the cluster, or change the cluster's topology, for example by adding, rejoining, or removing instances. However if you have an instance online which contains the InnoDB cluster metadata, it is possible to restore a cluster with quorum. This assumes you can connect to an instance that contains the InnoDB cluster metadata, and that instance can contact the other instances you want to use to restore the cluster.

Connect to an instance which contains the cluster's metadata, then use the Cluster.forceQuorumUsingPartitionOf(instance) operation, which restores the cluster based on the metadata on instance, and then all the instances that are ONLINE from the point of view of the given instance definition are added to the restored cluster.

mysql-js> cluster.forceQuorumUsingPartitionOf("ic@ic-1:3306")

  Restoring replicaset 'default' from loss of quorum, by using the partition composed of [ic@ic-1:3306]

  Please provide the password for 'ic@ic-1:3306': ******
  Restoring the InnoDB cluster ...

  The InnoDB cluster was successfully restored using the partition from the instance 'ic@ic-1:3306'.

  WARNING: To avoid a split-brain scenario, ensure that all other members of the replicaset
  are removed or joined back to the group that was restored.

In the event that an instance is not automatically added to the cluster, for example if its settings were not persisted, use Cluster.rejoinInstance() to manually add the instance back to the cluster.

The restored cluster might not, and does not have to, consist of all of the original instances which made up the cluster. For example, if the original cluster consisted of the following five instances:

and the cluster experiences a split-brain scenario, with ic-1, ic-2, and ic-3 forming one partition while ic-4 and ic-5 form another partition. If you connect to ic-1 and issue Cluster.forceQuorumUsingPartitionOf('ic@ic-1:3306') to restore the cluster the reulting cluster would consist of these three instances:

because ic-1 sees ic-2 and ic-3 as ONLINE and does not see ic-4 and ic-5.

If your cluster suffers from a complete outage, you can ensure it is reconfigured correctly using dba.rebootClusterFromCompleteOutage(). This operation takes the instance which MySQL Shell is currently connected to and uses its metadata to recover the cluster. In the event that a cluster's instances have completely stopped, the instances must be started and only then can the cluster be started. For example if the machine a sandbox cluster was running on has been restarted, and the instances were at ports 3310, 3320 and 3330, issue:

mysql-js> dba.startSandboxInstance(3310)
mysql-js> dba.startSandboxInstance(3320)
mysql-js> dba.startSandboxInstance(3330)
    

This ensures the sandbox instances are running. In the case of a production deployment you would have to start the instances outside of MySQL Shell. Once the instances have started, you need to connect to an instance with the GTID superset, which means the instance which had applied the most transaction before the outage. If you are unsure which instance contains the GTID superset, connect to any instance and follow the interactive messages from the dba.rebootClusterFromCompleteOutage(), which detects if the instance you are connected to contains the GTID superset. Reboot the cluster by issuing:

mysql-js> var cluster = dba.rebootClusterFromCompleteOutage();

The dba.rebootClusterFromCompleteOutage() operation then follows these steps to ensure the cluster is correctly reconfigured:

If you are not using MySQL Shell's interactive mode, you can use the rejoinInstances and removeInstances options to manually configure instances which should be joined or removed during the reboot of the cluster.

If you encounter an error such as The active session instance isn't the most updated in comparison with the ONLINE instances of the Cluster's metadata. then the instance you are connected to does not have the GTID superset of transactions applied by the cluster. In this situation, connect MySQL Shell to the instance suggested in the error message and issue dba.rebootClusterFromCompleteOutage() from that instance.

mysql-sql> SHOW VARIABLES LIKE 'gtid_executed';

If this process fails, and the cluster metadata has become badly corrupted, you might need to drop the metadata and create the cluster again from scratch. You can drop the cluster metadata using dba.dropMetadataSchema().

If you make configuration changes to a cluster outside of the AdminAPI commands, for example by changing an instance's configuration manually to resolve configuration issues or after the loss of an instance, you need to update the InnoDB cluster metadata so that it matches the current configuration of instances. In these cases, use the Cluster.rescan() operation, which enables you to update the InnoDB cluster metadata either manually or using an interactive wizard. The Cluster.rescan() operation can detect new active instances that are not registered in the metadata and add them, or obsolete instances (no longer active) still registered in the metadata, and remove them. You can automatically update the metadata depending on the instances found by the command, or you can specify a list of instance addresses to either add to the metadata or remove from the metadata. You can also update the topology mode stored in the metadata, for example after changing from single-primary mode to multi-primary mode outside of AdminAPI.

The syntax of the command is Cluster.rescan([options]). The options dictionary supports the following:

The cluster.checkInstanceState() function can be used to verify the existing data on an instance does not prevent it from joining a cluster. This process works by validating the instance's global transaction identifier (GTID) state compared to the GTIDs already processed by the cluster. For more information on GTIDs see Section 17.1.3.1, “GTID Format and Storage”. This check enables you to determine if an instance which has processed transactions can be added to the cluster.

The following demonstrates issuing this in a running MySQL Shell:

mysql-js> cluster.checkInstanceState('ic@ic-4:3306')

The output of this function can be one of the following:

Instances with an OK status can be added to the cluster because any data on the instance is consistent with the cluster. In other words the instance being checked has not executed any transactions which conflict with the GTIDs executed by the cluster, and can be recovered to the same state as the rest of the cluster instances.

To dissolve an InnoDB cluster you connect to a read-write instance, for example the primary in a single-primary cluster, and use the Cluster.dissolve() command. This removes all metadata and configuration associated with the cluster, and disables Group Replication on the instances. Any data that was replicated between the instances is not removed.

The Cluster.dissolve() operation can only configure instances which are ONLINE or reachable. If members of a cluster cannot be reached by the member where you issued the Cluster.dissolve() command you have to decide how the dissolve operation should proceed. If there is any chance you want to rejoin any instances that are identified as missing from the cluster, it is strongly recommended to cancel the dissolve operation and first bring the missing instances back online, before proceeding with a dissolve operation. This ensures that all instances can have their metadata updated correctly, and that there is no chance of a split-brain situation. However, if the instances from the cluster which cannot be reached have permanently left the cluster there could be no choice but to force the dissolve operation, which means that the missing instances are ignored and only online instances are affected by the operation.

In interactive mode, if members of a cluster are not reachable during a dissolve operation then an interactive prompt is displayed, for example:

mysql-js> Cluster.dissolve()
The cluster still has the following registered ReplicaSets:
{
    "clusterName": "testCluster", 
    "defaultReplicaSet": {
        "name": "default", 
        "topology": [
            {
                "address": "ic-1:3306", 
                "label": "ic-1:3306", 
                "role": "HA"
            }, 
            {
                "address": "ic-2:3306", 
                "label": "ic-2:3306", 
                "role": "HA"
            }, 
            {
                "address": "ic-3:3306", 
                "label": "ic-3:3306", 
                "role": "HA"
            }
        ]
    }
}
WARNING: You are about to dissolve the whole cluster and lose the high
availability features provided by it. This operation cannot be reverted. All
members will be removed from their ReplicaSet and replication will be stopped,
internal recovery user accounts and the cluster metadata will be dropped. User
data will be maintained intact in all instances.

Are you sure you want to dissolve the cluster? [y/N]: y

ERROR: The instance 'ic-2:3306' cannot be removed because it is on a '(MISSING)'
state. Please bring the instance back ONLINE and try to dissolve the cluster
again. If the instance is permanently not reachable, then you can choose to
proceed with the operation and only remove the instance from the Cluster
Metadata.

Do you want to continue anyway (only the instance metadata will be removed)?
[y/N]: y

Instance 'ic-3:3306' is attempting to leave the cluster...  Instance 'ic-1:3306'
is attempting to leave the cluster...

WARNING: The cluster was successfully dissolved, but the following instance was
skipped: 'ic-2:3306'. Please make sure this instance is permanently unavailable
or take any necessary manual action to ensure the cluster is fully dissolved.

In this example, the cluster consisted of three instances, one of which was offline when dissolve was issued. The error is caught, and you are given the choice how to proceed. In this case the missing ic-2 instance is ignored and the reachable members have their metadata updated.

When MySQL Shell is running in non-interactive mode, for example when running a batch file, you can configure the behavior of the Cluster.dissolve() operation using the force option. To force the dissolve operation to ignore any instances which are unreachable, issue:

mysql-js> Cluster.dissolve({force: true})

Any instances which can be reached are removed from the cluster, and any unreachable instances are ignored. The warnings in this section about forcing the removal of missing instances from a cluster apply equally to this technique of forcing the dissolve operation.

You can also use the interactive option with the Cluster.dissolve() operation to override the mode which MySQL Shell is running in, for example to make the interactive prompt appear when running a batch script. For example:

mysql-js> Cluster.dissolve({interactive: true})

The dba.gtidWaitTimeout MySQL Shell option configures how long the Cluster.dissolve() operation waits for cluster transactions to be applied before removing a target instance from the cluster, but only if the target instance is ONLINE. An error is issued if the timeout is reached when waiting for cluster transactions to be applied on any of the instances being removed, except if force: true is used, which skips the error in that case.

Server instances can be configured to use secure connections. For general information on using SSL with MySQL see Section 6.3, “Using Encrypted Connections”. This section explains how to configure a cluster to use SSL. An additional security possibility is to configure which servers can access the cluster, see Creating a Whitelist of Servers.

When using dba.createCluster() to set up a cluster, if the server instance provides SSL encryption then it is automatically enabled on the seed instance. Pass the memberSslMode option to the dba.createCluster() method to specify a different SSL mode. The SSL mode of a cluster can only be set at the time of creation. The memberSslMode option is a string that configures the SSL mode to be used, it defaults to AUTO. The permitted values are DISABLED, REQUIRED, and AUTO. These modes are defined as:

When you issue the cluster.addInstance() and cluster.rejoinInstance() commands, SSL encryption on the instance is enabled or disabled based on the setting found for the seed instance.

When using createCluster() with the adoptFromGR option to adopt an existing Group Replication group, no SSL settings are changed on the adopted cluster:

MySQL Shell always enables or disables SSL for the cluster for both Group Replication recovery and Group Communication, see Section 18.5.2, “Group Replication Secure Socket Layer (SSL) Support”. A verification is performed and an error issued in case those settings are different for the seed instance (for example as the result of a dba.createCluster() using adoptFromGR) when adding a new instance to the cluster. SSL encryption must be enabled or disabled for all instances in the cluster. Verifications are performed to ensure that this invariant holds when adding a new instance to the cluster.

The deploySandboxInstance() command attempts to deploy sandbox instances with SSL encryption support by default. If it is not possible, the server instance is deployed without SSL support. Use the ignoreSslError option set to false to ensure that sandbox instances are deployed with SSL support, issuing an error if SSL support cannot be provided. When ignoreSslError is true, which is the default, no error is issued during the operation if the SSL support cannot be provided and the server instance is deployed without SSL support.

When using a cluster's createCluster(), addInstance(), and rejoinInstance() methods you can optionally specify a list of approved servers that belong to the cluster, referred to as a whitelist. By specifying the whitelist explicitly in this way you can increase the security of your cluster because only servers in the whitelist can connect to the cluster. Using the ipWhitelist option configures the group_replication_ip_whitelist system variable on the instance. By default, if not specified explicitly, the whitelist is automatically set to the private network addresses that the server has network interfaces on. To configure the whitelist, specify the servers to add with the ipWhitelist option when using the method. IP addresses must be specified in IPv4 format. Pass the servers as a comma separated list, surrounded by quotes. For example:

mysql-js> cluster.addInstance("ic@ic-3:3306", {ipWhitelist: "203.0.113.0/24, 198.51.100.110"})

This configures the instance to only accept connections from servers at addresses 203.0.113.0/24 and 198.51.100.110. The whitelist can also include host names, which are resolved only when a connection request is made by another server.

You can automate cluster configuration with scripts, which can be run using MySQL Shell. For example:

shell> mysqlsh -f setup-innodb-cluster.js

The contents of an example script file is shown here:

print('InnoDB cluster sandbox set up\n');
print('==================================\n');
print('Setting up a MySQL InnoDB cluster with 3 MySQL Server sandbox instances.\n');
print('The instances will be installed in ~/mysql-sandboxes.\n');
print('They will run on ports 3310, 3320 and 3330.\n\n');

var dbPass = shell.prompt('Please enter a password for the MySQL root account: ', {type:"password"});

try {
   print('\nDeploying the sandbox instances.');
   dba.deploySandboxInstance(3310, {password: dbPass});
   print('.');
   dba.deploySandboxInstance(3320, {password: dbPass});
   print('.');
   dba.deploySandboxInstance(3330, {password: dbPass});
   print('.\nSandbox instances deployed successfully.\n\n');

   print('Setting up InnoDB cluster...\n');
   shell.connect('root@localhost:3310', dbPass);

   var cluster = dba.createCluster("prodCluster");

   print('Adding instances to the cluster.');
   cluster.addInstance({user: "root", host: "localhost", port: 3320, password: dbPass});
   print('.');
   cluster.addInstance({user: "root", host: "localhost", port: 3330, password: dbPass});
   print('.\nInstances successfully added to the cluster.');

   print('\nInnoDB cluster deployed successfully.\n');
} catch(e) {
   print('\nThe InnoDB cluster could not be created.\n\nError: ' +
   + e.message + '\n');
}

You can optionally configure how a single-primary cluster elects a new primary, for example to prefer one instance as the new primary to fail over to. Use the memberWeight option and pass it to the dba.createCluster() and Cluster.addInstance() methods when creating your cluster. The memberWeight option accepts an integer value between 0 and 100, which is a percentage weight for automatic primary election on failover. When an instance has a higher precentage number set by memberWeight, it is more likely to be elected as primary in a single-primary cluster. When a primary election takes place, if multiple instances have the same memberWeight value, the instances are then prioritized based on their server UUID in lexicographical order (the lowest) and by picking the first one.

Setting the value of memberWeight configures the group_replication_member_weight system variable on the instance. Group Replication limits the value range from 0 to 100, automatically adjusting it if a higher or lower value is provided. Group Replication uses a default value of 50 if no value is provided. See Section 18.1.3.1, “Single-Primary Mode” for more information.

For example to configure a cluster where ic-3 is the preferred instance to fail over to in the event that ic-1, the current primary, leaves the cluster unexpectedly use memberWeight as follows:

dba.createCluster('cluster1', {memberWeight:35})
var mycluster = dba.getCluster()
mycluster.addInstance('ic@ic2', {memberWeight:25})
mycluster.addInstance('ic@ic3', {memberWeight:50})

Group Replication provides the ability to specify the failover guarantees (eventual or “read your writes”) if a primary failover happens in single-primary mode (see Section 18.4.2.2, “Configuring Transaction Consistency Guarantees”). You can configure the failover guarantees of an InnoDB cluster at creation by passing the consistency option (prior to version 8.0.16 this option was the failoverConsistency option, which is now deprecated) to the dba.createCluster() operation, which configures the group_replication_consistency system variable on the seed instance. This option defines the behavior of a new fencing mechanism used when a new primary is elected in a single-primary group. The fencing restricts connections from writing and reading from the new primary until it has applied any pending backlog of changes that came from the old primary (sometimes referred to as “read your writes”). While the fencing mechanism is in place, applications effectively do not see time going backward for a short period of time while any backlog is applied. This ensures that applications do not read stale information from the newly elected primary.

The consistency option is only supported if the target MySQL server version is 8.0.14 or later, and instances added to a cluster which has been configured with the consistency option are automatically configured to have group_replication_consistency the same on all cluster members that have support for the option. The variable default value is controlled by Group Replication and is EVENTUAL, change the consistency option to BEFORE_ON_PRIMARY_FAILOVER to enable the fencing mechanism. Alternatively use consistency=0 for EVENTUAL and consistency=1 for BEFORE_ON_PRIMARY_FAILOVER.

By default, an InnoDB cluster runs in single-primary mode, where the cluster has one primary server that accepts read and write queries (R/W), and all of the remaining instances in the cluster accept only read queries (R/O). When you configure a cluster to run in multi-primary mode, all of the instances in the cluster are primaries, which means that they accept both read and write queries (R/W). If a cluster has all of its instances running MySQL server version 8.0.15 or later, you can make changes to the topology of the cluster while the cluster is online. In previous versions it was necessary to completely dissolve and re-create the cluster to make the configuration changes. This uses the group action coordinator exposed through the UDFs described at Section 18.4.1, “Configuring an Online Group”, and as such you should observe the rules for configuring online groups.

Usually a single-primary cluster elects a new primary when the current primary leaves the cluster unexpectedly, for example due to an unexpected halt. The election process is normally used to choose which of the current secondaries becomes the new primary. To override the election process and force a specific server to become the new primary, use the Cluster.setPrimaryInstance(instance) function, where instance specifies the connection to the instance which should become the new primary. This enables you to configure the underlying Group Replication group to choose a specific instance as the new primary, bypassing the election process.

You can change the mode (sometimes described as the topology) which a cluster is running in between single-primary and multi-primary using the following operations:

You can check and modify the settings in place for an InnoDB cluster while the instances are online. To check the current settings of a cluster, use the following operation:

You can configure the options of an InnoDB cluster at a cluster level or instance level, while instances remain online. This avoids the need to remove, reconfigure and then again add the instance to change InnoDB cluster options. Use the following operations:

The way which you use InnoDB cluster options with the operations listed depends on whether the option can be changed to be the same on all instances or not. These options are changeable at both the cluster (all instances) and per instance level:

This option is changeable at the per instance level only:

These options are changeable at the cluster level only:

When you are using an instance as part of an InnoDB cluster, the auto_increment_increment and auto_increment_offset variables are configured to avoid the possibility of auto increment collisions for multi-primary clusters up to a size of 9 (the maximum supported size of a Group Replication group). The logic used to configure these variables can be summarized as: