elasticsearch

On the third day of Christmas, ObjectRocket gave to me… 3 Master Nodes!

By December 16, 2019 March 16th, 2020 No Comments

Hooray, yet more new features on our ObjectRocket Elasticsearch offerings!

We’ve had great success with the Elasticsearch product on ObjectRocket, and are excited to introduce dedicated master nodes! For those in need of larger, more reliable clusters, separating out the master role responsibilities and assigning them to a dedicated node increases stability across your entire cluster.

What’s a dedicated master node?

A master node is responsible for cluster management tasks, such as creating/deleting an index, keeping track of all nodes, and sharding. Dedicating one node to take care of the cluster management work means that this dedicated master can focus on its work, and allow the data nodes to do the heavy lifting of searches, indexing, and data-specific work.

For much smaller clusters, one can get away with configuring a node to have both master and data roles. However, because the master node is key for stability and smooth operations of the entire cluster, it is highly recommended that a dedicated master node is used, especially for larger clusters.

Following standard guidelines of having an odd number of master nodes, we always create three master nodes when you create a new instance and choose to add master nodes. Having an odd number of master nodes means that one node can always act as a tie-breaker, and thus avoid cases of ‘split brain’ which can cause loss of data integrity. You always get three master nodes, and the number of data nodes created is based on the capacity (storage/memory) that you select when creating the cluster.

How to configure dedicated masters

At ObjectRocket, we always strive to make spinning up new databases with the features you want as easy as possible. No need to configure your cluster manually; everything can be done through Mission Control, our UI, within a few minutes.

You can select the option to configure dedicated master nodes when creating a new instance in our Mission Control. You can also specify the size of the dedicated master nodes by clicking on the ellipses (…) icon in the upper right of the UI. This is done on Step 2 of creating a new instance, as shown in the image below.

Our dedicated master nodes come in three different sizes:

  • 2GB RAM / 8GB Disk
  • 4GB RAM / 16GB Disk
  • 8GB RAM / 32GB Disk

Generally speaking, if the data nodes (your storage nodes) in your cluster are 8Gb to 32GB in size, you should choose 2GB RAM master nodes. With bigger storage nodes, between 64GB and 128Gb, you will want to choose the 4GB RAM master node. And for clusters with data nodes over 128GB, go for the 8GB RAM master nodes.

If you have any questions about sizing a dedicated master node or any other elasticsearch questions, please reach out to us and we will help you.

What’s next

Dedicated master nodes are currently only available for new instances and can not currently be added for old instances. But fret not, we’re hard at work and the ability to convert multirole nodes to dedicated masters is right around the corner!

As always, we’re always improving our products here at ObjectRocket to give our customers the very best offerings and services. Please feel free to reach out to us if you have any questions or feature suggestions not only for Elasticsearch, but for any of our products! Last but not least, if you have any databases you’d like us to support, we’d be more than happy to hear about it as well.

James Hong

James Hong

James has always been interested in anything and everything technology since he was 5 years old. Started out with just exploring the internet and tinkering with computers during the dial-up era to learning programming in junior high and eventually graduating from University of Illinois Urbana-Champaign with a bachelor's in Computer Science. James is currently a Product Manager at ObjectRocket creating exciting new features for Redis and Elasticsearch.