Cluster groups

Anton Vasilev

Overview

Starting with ADBM 2.10.0, you can use cluster groups. A cluster group is a mechanism for managing a shared backup configuration and for using a shared backup repository across multiple ADB clusters registered in ADBM.

When clusters are not part of a group, each maintains its own independent backup repository. Clusters within a group share a single repository managed by the group leader.

This approach reduces the time required to restore the original production cluster from a DR cluster: the original cluster can perform an incremental restore from the shared repository rather than take a full backup on the DR cluster and then restore from it. Cluster groups also reduce storage consumption because a shared backup chain in one storage repository is kept and reused for the entire group.

Key characteristics of cluster groups:

  • All clusters in a group share a single backup configuration (which includes storage, archiving, and compression settings) inherited from the current leader. While a cluster is in a group, its backup configuration cannot be edited.

  • The first cluster added to a group becomes the initial leader. Its storage is used as the shared repository for all other members. The initial leader cannot be removed from the group even if another cluster is the current leader. It can leave the group only when the entire group is deleted.

  • Only the group leader writes data to the shared repository. Running any backup action is not available to the non-leader members.

  • The leader schedule is used to run scheduled jobs (such as backups, restore points, and cleanups). Non-leader members have their scheduled jobs disabled. When a cluster is promoted to leader, the new leader’s schedule becomes active and is used for the group.

The following constraints apply to cluster groups:

  • Data flows are allowed between clusters of the same group or between clusters that are not part of any group.

  • In data flows between group members, only the group leader can be the source cluster.

  • Clusters in a group must meet the same requirements as clusters in a data flow: all clusters in the group must share the same ADB major version, the same number of primary segments, and the same operating system on their hosts.

  • A cluster can belong to only one cluster group at a time.

  • If POSIX storage is used, the mount point must be identical on all clusters in the group.

Group member states

Each cluster within a group can have one of the following states.

State Description

INIT

A temporary state indicating that the cluster is added to the group, but the leader configuration has not yet been applied to it

CONFIGURED

The leader backup configuration has been successfully applied to the cluster. The cluster is ready to begin restoring data from the shared repository. When another cluster is promoted, the previous leader also gets this state

JOINED

The cluster has been fully restored from the group repository. Clusters of this state are eligible for leader promotion

PROMOTING

A temporary state indicating that the promotion process is in progress for this cluster

LEADER

The cluster is the group leader
Cluster states in a cluster group
Cluster states in a cluster group
Cluster states in a cluster group
Cluster states in a cluster group

A typical workflow for cluster groups is as follows:

  1. Create a group (initially containing only one cluster).

  2. Add clusters to that group.

  3. Restore data to synchronize new clusters with the rest of the group. This is required for a cluster that you plan to promote.

  4. Promote a cluster if you want to change the group leader.

Create a cluster group

  1. Open the Backup manager page and navigate to Data flow → Cluster groups.

  2. Click Create cluster group.

    Cluster groups page
    Cluster groups page

  3. In the window that opens, configure the following parameters:

    • Group name — a unique name for the new cluster group.

    • Initial leader — the first cluster to be added to the group. The selected cluster becomes the initial leader of the group. Its backup configuration becomes the shared configuration for the entire group, and its storage is used as the shared repository.

    Create cluster group dialog
    Create cluster group dialog

  4. Click Create. As a result, the cluster group is displayed on the Data flow → Cluster groups tab.

    Cluster group created
    Cluster group created

Add clusters to a group

For a cluster to be added to a group:

  • It must have a backup configuration.

  • It must not be in a Down state.

  • It must not belong to another cluster group.

To add clusters:

  1. Open the Backup manager page and navigate to Data flow → Cluster groups.

  2. In the Action column, click added actions dark added actions light and select Manage clusters.

    Manage clusters action
    Manage clusters action

  3. In the window that opens, select clusters to be added to the group and click Save.

    Manage clusters dialog
    Manage clusters dialog

The initial state of a new cluster is INIT. Adding a cluster starts the Add cluster to group action that will apply the leader backup configuration to the newly added clusters. When it completes, the cluster state changes to CONFIGURED.

NOTE

The original backup configuration of the added clusters is preserved, but it cannot be modified. If a cluster is later removed from the group, it will revert to this original configuration.

Preview a cluster group

To quickly preview the group content, use the information dark information light icon in the Action column of the Cluster groups page.

Cluster group details
Cluster group details

To get details about the latest restore point in a cluster, open the group and click the information dark information light icon in the Latest restore point column.

Restore point details
Restore point details

To view actions and backups performed on the group clusters, use the Actions and Backups tabs. Selecting an action in the list redirects you to the Backup → Actions page where you can get more details about subactions and end results.

Actions in the cluster group
Actions in the cluster group

Get the backup from the leader

Once a cluster has been added to a group, it can be synchronized with the group shared repository through a copy or streaming data flow.

  1. On the Data flow → Configurations tab, click Create and create a data flow configuration with the leader as the source cluster and a newly added cluster as a target cluster.

    Create a data flow configuration
    Create a data flow configuration

  2. On the Data flow → Data flows tab, click Create data flow and run the data flow with the created configuration. For more information on the data flow configuration parameters, refer to Disaster recovery.

    Optionally, select the Promote flag to make the target cluster the leader after the data flow is completed.

    Run a data flow
    Run a data flow

This will start the Copying cluster (for the Copy data flow type) or Streaming cluster (for the Streaming data flow type) action. When it completes, the cluster state changes to JOINED.

Promote a new leader

Only one cluster in the group can be the leader. You can change the leader through a promotion process — either manually or automatically at the end of a data flow.

When a cluster becomes the new leader, its backup configuration is used as the group configuration, and its schedule becomes active. The state of the previous leader changes to CONFIGURED.

You can only promote a cluster in the JOINED state. Note that whether the cluster data is up to date with the leader is determined by the restore point, not by the JOINED state.

Manual promotion

  1. Open the Backup manager page and navigate to Data flow → Cluster groups.

  2. In the Action column, click added actions dark added actions light and select Promote cluster.

    Promoting a cluster
    Promoting a cluster

  3. In the window that opens, in the Candidate list, select the cluster to promote and click Run.

    Manual promotion
    Manual promotion

After the Promote new group leader action completes successfully, the selected cluster becomes the leader.

Automatic promotion

When running a data flow between two clusters that belong to the same cluster group, you can enable the Promote option in the data flow creation form.

  1. Open the Backup manager page and navigate to Data flow → Data flows.

  2. Click Create data flow.

  3. In the window that opens, select the Promote flag.

    Automatic promotion
    Automatic promotion

After the data flow completes successfully, the target cluster becomes the new group leader.

Rename a group

  1. Open the Backup manager page and navigate to Data flow → Cluster groups.

  2. In the Action column, click added actions dark added actions light and select Edit cluster group.

    Renaming a group
    Renaming a group

  3. In the window that opens, enter the new name for the group and click Apply.

Remove a cluster

For removing clusters, two limitations apply:

  • The current leader cannot be removed. If you need to remove the leader, first promote another cluster to become the new leader.

  • The initial leader can never be removed from the group as its storage is used by all clusters in the group. Delete the entire group if you don’t want the initial leader to be part of a group.

To remove a cluster:

  1. Open the Backup manager page and navigate to Data flow → Cluster groups.

  2. Click the name of the group to open the group page.

  3. In the Action column, click the delete dark delete light icon.

    Removing a cluster
    Removing a cluster

  4. In the window that opens, click Delete to confirm deletion.

Delete a group

When you delete a group, all clusters except for the leader and the initial leader receive a new draft backup configuration that reflects their state before joining the group. You can save these configurations to be able to create new ones. The current leader and initial leader clusters automatically have their new backup configurations applied.

  1. Open the Backup manager page and navigate to Data flow → Cluster groups.

  2. In the Action column, click added actions dark added actions light and select Delete cluster group.

    Group deletion
    Group deletion

  3. In the window that opens, click Delete to confirm deletion.

Found a mistake? Seleсt text and press Ctrl+Enter to report it