What data does ensemble clustering need

Provision of a coordination service ensemble

The coordination service is based on the open source project Apache ZooKeeper (Link opens in a new window) and coordinates activities on the server. It guarantees a quorum in the event of a failure and serves as a reliable source for the server topology, configuration and status. The service is automatically installed on the first Tableau Server node, but no further instances are installed as you add more nodes. Because Tableau Server depends on a properly functioning Coordination Service to function properly, it is recommended that you add additional instances of the Coordination Service by deploying a new Coordination Service ensemble for server installations with three or more nodes. This ensures redundancy and higher availability if an instance of the coordination service has problems.

Important: The process for deploying a coordination service ensemble has changed from version 2020.1.0. If you're running an earlier version of Tableau Server, see the documentation for that version. Documentation on all supported versions can be found here: Tableau Help (Link opens in a new window)

Hardware requirements

The hardware that you use for Tableau Server can affect the quality of execution of the Coordination Service. In particular with regard to:

  • the random access memory: The coordination service manages status information in the working memory. The memory footprint is intentionally kept small and is usually not a factor in overall server performance.

  • the Disk speed: Because the service stores status information on disk, it takes advantage of the fast disk speed on single-node computers.

  • the Connection speed between nodes: The service communicates continuously between cluster nodes. A fast connection speed between nodes helps with efficient synchronization.

Because the Coordination Service is I / O intensive, if you are running Tableau Server on computers that meet or even meet the minimum hardware requirements, it is advisable to configure a Coordination Service ensemble that places the service on nodes that are not used by any other server processes exceed. This reduces the likelihood of delays due to I / O conflicts between server processes. For information on deploying an ensemble on nodes for dedicated coordination services, see Configuring Tableau Server for High Availability with Coordination Service Nodes.

The coordination service quorum

In order to ensure the proper functioning of the coordination service, a quorum is required for the service, i. H. a minimum number of instances of the service. As a result, the number of nodes in your installation affects how many instances of the coordination service you should configure in your ensemble.

Number of usable coordination service instances

The maximum number of Coordination Service instances for an ensemble in Tableau Server depends on how many Tableau Server nodes you have in your deployment. Configure a coordination services ensemble according to the following guidelines:

Total number of server nodesRecommended number of coordination service nodes in the ensemble (must be 1, 3, or 5)Hints
1-2 knots1 knotThis is the default and no changes are required unless you want to move the Coordination Service from your initial node to the additional node.
3-4 knots3 knots
5 or more knots3 knots
or 5 knots
Five is the maximum number of Coordination Service instances that you can install. In a three-node coordination services ensemble, one of the ensemble nodes can fail without a Tableau Server failure. In a five-node ensemble, two of the ensemble nodes can fail without a Tableau Server failure.

For most installations, three coordination service nodes are sufficient. Because of the I / O intensive nature of the coordination service, this is the most powerful configuration.

If high availability is a priority, consider deploying a five-node coordination services ensemble. This provides the most redundancy in the event that one or more nodes fail. However, this requires more system resources. A maximum of two of the ensemble nodes can fail without adversely affecting Tableau Server (as long as all other services on the node are also available on functioning nodes).

To reduce the impact on performance, find the Coordination Service on nodes that are less performing other services. Alternatively, it is advisable to use nodes on which only the coordination service is running. For more information, see Configuring Tableau Server for High Availability with Coordination Service Nodes.

Provision of a new coordination services ensemble

If you are installing three or more nodes, you should also deploy a coordination services ensemble. Otherwise, a warning message appears every time you change the server configuration or topology. You can safely ignore this message, but it is best practice to deploy a co-ordination services ensemble with multiple nodes.

When you install Tableau Server, a single instance of the Coordination Service is installed on the first node. The TSM and Tableau Server are dependent on a properly functioning coordination service. You should therefore configure additional instances of the coordination service by providing a coordination service ensemble in order to provide redundancy and guarantee availability in installations with multiple nodes. Coordination service ensembles are installed with one, three, or five instances of the coordination service. In a three-node Tableau Server installation, a number of three Coordination Service Instances is recommended, one for each node.

Do not attempt to use a co-ordinating service ensemble when there are further changes to be made. Discard or apply any pending changes before deploying a new Coordination Services ensemble.

Note: This process includes steps that you may need to complete from the TSM command line.

The following steps show how to provision a new Coordination Services ensemble in an existing three-node Tableau Server cluster and delete the old ensemble.

  1. On the first node, open a command prompt as an administrator.

  2. Stop Tableau Server:

    When prompted, log in as a TSM administrator.

    Some TSM processes are still running, including the Administration Controller and Administration Agent.

  3. Confirm that there are no pending changes:

    If there are any pending changes, either discard them or apply them. The pending changes will take some time to apply:

    • Discard the changes:

      tsm pending-changes discard
    • or

    • Apply the changes:

      tsm pending-changes apply

    Wait for the command to complete and return to the system prompt.

  4. Get the node IDs for each node in the cluster:

  5. Use the command to add a new Coordination Service ensemble by adding the Coordination Service on the specified node. You must identify the node (s) on which the Coordination Service should be added. The command also makes the new ensemble the "production ensemble" (the ensemble used). The old ensemble is removed unless the deployment fails. If that happens, see step 6 below for more information.

    Note: A "y / n" prompt to confirm a server restart is displayed. Include the option to run the command without typing.

    For example, deploy the Coordination Service on all three nodes in a three-node cluster, where the nodes are Node1, Node2, and Node3:

    Wait for the command to complete and return to the system prompt.

  6. (Optional) If the deployment fails, you must run the command to remove the unsuccessfully deployed ensemble. For more information about running the command, see tsm topology cleanup-coordination-service.

  7. Start Tableau Server: