Skip to main content

OpenSearch cluster fault tolerance

Last update:

The fault tolerance of an OpenSearch managed database cluster is affected by:

Index replication

The cluster uses a standard index replication mechanism—for each primary shard, a number of replica shards equal to the number you specified when creating the index is created. Writes to the index are only performed through primary shards, while reads can be performed concurrently from both primary and replica shards.

Number of nodes in node groups

A node group with the Manager role consists of three nodes. You cannot change the number of nodes in this group or add another node group with the Manager role.

A node group with the Data role consists of one node by default. To increase cluster fault tolerance, you can:

Node allocation

The type of node allocation in groups depends on the number of nodes in the group, the type of pool where the cluster is located, and the number of segments in the pool:

  • Single-Segment — in one segment of a single-zone pool. Applicable to:

    • for a node group with the Data role, which has only one node;
    • node groups with multiple nodes, if the cluster is located in a pool with only one segment;
    • node groups with multiple nodes in HighFreq configurations;
note

For example, you created an OpenSearch cluster in the ru-2 pool and used configurations from the HighFreq line for node groups with Manager and Data roles. All nodes in both groups will be allocated in one segment of the ru-2c pool.

  • Multi-Segment — in different segments of a single-zone pool. Applicable for node groups with multiple nodes if the cluster is located in a pool with multiple segments. Segments are located in different racks of the same data center. Nodes are distributed across segments sequentially;
note

For example, you created an OpenSearch cluster in the ru-1 pool and specified one node for the node group with the Data role. All three nodes of the group with the Manager role will be allocated sequentially in the segments of the ru-1a, ru-1b, and ru-1c pools. The node of the group with the Data role will be allocated in the ru-1a segment. If you add a second node to the group with the Data role, it will be allocated in the ru-1b pool segment.

  • Single-AZ — in one segment of a multi-zone pool. Applicable for a node group with the Data role that has only one node, if the cluster is located in the ru-6 pool. If you increase the number of nodes in the group, the node allocation type will change to Multi-AZ;

  • Multi-AZ — in different segments of a multi-zone pool. Applicable for node groups with multiple nodes if the cluster is located in the ru-6 pool. Each pool segment is located in a separate data center. Nodes are distributed across segments sequentially.

note

For example, you created an OpenSearch cluster in the ru-6 pool and specified one node for the node group with the Data role. All three nodes of the group with the Manager role will be allocated sequentially in the segments of the ru-6a, ru-6b, and ru-6c pools. The node of the group with the Data role will be allocated in the ru-6a segment. If you add a second node to the group with the Data role, it will be allocated in the ru-6b pool segment.

You can view the list of pools and segments in the Selectel Infrastructure table of the Locations: countries, regions, availability zones, pools, and data center addresses guide.

Increase the number of nodes in the group with the Data role

You can add no more than ten nodes to a node group with the Data role.

  1. In the Control panel, click Products in the top menu and select Managed Databases.
  2. Open the Active tab.
  3. Open the database cluster page → Settings tab.
  4. In the menu of the node group with the Data role, select Change configuration.
  5. Specify the number of nodes in the group.
  6. Click Save.

Add a node group with the Data role

The total number of node groups with the Data role in a cluster cannot exceed two.

  1. In the Control panel, click Products in the top menu and select Managed Databases.

  2. Open the Active tab.

  3. Open the database cluster page → Settings tab.

  4. In the Cluster block, click Add node group.

  5. Optional: change the node group name.

  6. Select a node group with the Data role.

  7. Select a configuration line.

  8. If you selected a line with custom configurations, specify the number of vCPUs, RAM size, and disk size.

  9. If you selected a line with fixed configurations, select a configuration.

  10. Specify the number of nodes in the group.

  11. Optional: to make the group nodes accessible from the internet, connect a public IP address for them:

    • if you selected an existing private subnet, enable the Public access toggle. The private subnet must meet the requirements;
    • if you are creating a new private subnet, connect a public IP address after adding the node group.

    IP addresses are automatically assigned to all group nodes.

  12. Click Save.