Upscaling a stand-alone cluster to a multi-node cluster in Assets Server

Assets Server can be run as a stand-alone cluster consisting of 1 node or as a cluster consisting of many nodes (see The Assets Server structure: clusters and nodes).

This article describes how to upscale an existing stand-alone cluster to a multi-node cluster.

Note: It is assumed here that the stand-alone cluster is already fully up and running and uses manual discovery.

Step 1. Stop the current node.

Step 2. Install Assets Server on as many additional machines as needed.

Step 3. On each system, access the config folder of Assets Server by doing one of the following:

• Windows: Start menu > Assets Server > Assets Server > click Open config
• MacOS: Applications > Assets Server > click Open config
• Linux: Access /srv/elvis-server/config

Step 4. In the node-config.properties.txt file, configure the following options:

• Set the cluster name to the same name by setting the clusterName option:

Example: clusterName=AssetsServerDemo

• Set the shared data location by setting the sharedDataLocation option:

Example: sharedDataLocation=${appDataDir}/Elvis Shared Data

• Set the IP addresses of the other nodes in the cluster by updating the cluster.join.tcpip.members option.

Step 5. In the cluster-config.properties.txt file, configure the following options:

Note: The cluster-config.properties.txt file needs to be identical for all node configurations. After setting up the first node and configuring subsequent nodes, make sure that any changes to a cluster property are implemented in the cluster-config.properties file for all nodes. See also Changing the Assets Server configuration for a running cluster.

• Set the cluster count to the total number of nodes in the cluster:

Example: cluster.nodeCount=3

• Set the Search node count to the total number of Search nodes in the cluster:

Example: cluster.searchNodeCount=3

• When the sharedDataLocation is different on each node, make sure that the elasticsearch.backup.location (which references this location by default) is also configured correctly for the cluster.

Step 6. By default, the index.numberOfShards setting matches the cluster.searchNodeCount setting. When changing the cluster.searchNodeCount setting therefore, the index.numberOfShards setting should also be changed.

One shard per Search node is best from a performance point of view.

If you expect your number of Search nodes to grow, configure the index.numberOfShards setting to match your expected maximum number of Search nodes.

If your indices have already been created and you change the number of Search nodes or change the index.numberOfShards setting, you will need to update the Assets Server index.

For more information, see the comment for the index.numberOfShards setting in the cluster-config.properties.txt file.

Step 7. (Optional) Set the preferred number of replicas for the cluster. Each replica creates a copy of the index and helps to improve search capabilities and can function as a backup.

A minimum and maximum limit can be set for replicas. By default the replicas will scale so that for each node in the cluster -1 a replica will be created until the maximum limit is reached.

Note: Adding more replicas better balances search capacity over the nodes and thereby improves search speeds. However as every replica is a copy of the primary index they take up the same amount of disk space as the primary; when updating data, both the primary and all replicas need to be updated.

Replicas are controlled through the following option:

• Location: The cluster properties page in the Management Console
• Option name: autoExpandReplicas
• Possible values: a dash-delimited lower and upper bound, 'all', or 'false'
• Default value: 0-3
• Example:

0-all

Step 8. Start the original node.

Step 9. Wait a minimum of 15 seconds and start the other nodes.

Step 10. On any of the nodes, access the Paramedic page in Assets Server and notice that only the config indices are replicated:

• Access the Management Console and in the menu on the left, choose Tools > Support tools > Paramedic (in the Elasticsearch section under Plugins).

Step 11. Access the Management Console by doing one of the following:

• Access the following URL and log in using system administrator credentials (a 'Super user' account):

<Assets Server URL>/console

• In Assets, access the Avatar menu and choose Management Console.

Step 12. In the menu on the left, choose Tools > Index revision tool.

Step 13. In the "Choose index to update" section, re-index each index except "config".

This will increase the current revision for each of the indices and trigger the shards to replicate on the other node.