Cluster Expansion

As your workload changes, you can expand your cluster to improve its query performance, system availability, and fault tolerance. Expanding a cluster adds more nodes to the cluster. During an expansion, you can also change the replication factor of your cluster.

Before you begin

If you are expanding to a single-node installation, TigerGraph must already be installed on the new node in exactly the same version as the original. For expansion to a target cluster, the target cluster should be running a cluster installation of TigerGraph Server exactly the same version as the original that you are expanding from. You must expand to all nodes in the target cluster, not a portion of them.

  • Ensure that the cluster is not using shared storage. Cluster expansion does not support shared storage.

  • Ensure that no loading jobs, queries, or REST requests are running on the original or the expansion target.

  • Obtain a few key measures for the state of your data before the expansion, such as vertex counts/edge counts or certain query results. This will be useful in verifying data integrity after the expansion completes.

  • If the original cluster is a single node installation, make sure the IP used is not a local loopback address such as 127.0.0.1.

Known issue:

If you have ever run a schema change job to delete a vertex attribute or edge attribute, cluster expansion may lead to issues reading attributes of that vertex or edge type. Prior to expansion, run the following command as the TigerGraph Linux user on your instance to check if you can proceed safely. Make sure all TigerGraph services are running when you run the script:

$ if grep "AttributeName: .*_deleted" $(gadmin config get System.DataRoot)/gstore/0/part/config.yaml > /dev/null; then printf "\nContact support\n\n"; else printf "\nProceed with expansion\n\n"; fi

If the script returns "Proceed with expansion", you can proceed with cluster expansion. If you see "Contact support", do not proceed with the expansion, and contact TigerGraph support to for assistance.

Procedure

Step 1: Identify new cluster replication and partition

Before running any commands to expand a cluster, make sure you have a clear idea of how the new cluster should be distributed. You should have the following information:

  • The new replication factor of the cluster

  • The new partitioning factor of the cluster

  • The IP addresses of the new nodes to be added to the cluster

Step 2: Expand the cluster

To expand the cluster, run the gadmin cluster expand command like below. If the expansion involves changing the replication factor, use the --ha option to indicate the new replication factor:

$ gadmin cluster expand node_ip_list [--ha <replication_factor>]

node_ip_list is the machine aliases of the nodes you are adding to the cluster mapped to their IP addresses with a colon(:), and separated by a comma. Below is an example:

$ gadmin cluster expand m3:10.128.0.81,m4:10.128.0.82 --ha 1

We suggest naming the new nodes following the convention of m<serial>, such as m1, m2, and m3 for a 3-node cluster. If you are adding a fourth node, then the fourth node would be named m4. If you decide to name them differently, make sure that all names are unique within the cluster.

Supply a staging location

Extra disk space is required during cluster expansion. If more space is not available on the same disk, you can supply a staging location on a different disk to hold temporary data:

$ gadmin cluster expand m3:192.168.1.3,m4:192.168.1.4 --stagingPath /tmp/

If you choose to supply a staging location, make sure that the TigerGraph Linux user has write permission to the path you provide. The overall amount of space required for expansion on each node is (1 + Math.ceil(oldPartition/newPartition) ) * dataRootSize. oldPartition and newPartition each stand for the partitioning factor of the cluster before and after expansion; dataRootSize stands for the size of the data root folder on the node.

For example, if you are expanding from a 6 node cluster with a replication factor of 2 and a partitioning factor of 3, to a 10-node cluster with a replication factor of 2 and a partitioning factor of 5, and the size of the data root folder on a node is 50GB. Then you would need more than (1 + Math.ceil(3/5)) * 50) = 100 GB of free space on the staging path.

Step 3: Verify success and delete temporary files

When the expansion completes, you should see a message confirming the completion of the cluster change. The message will also include the location of the temporary files created during the expansion.

Verify data integrity by comparing vertex/edge counts or query results. After confirming a successful expansion, delete the temporary files to free up disk space.