# Would be a good idea to turn all data nodes into non-eligible masters?

**URL:** <https://discuss.elastic.co/t/would-be-a-good-idea-to-turn-all-data-nodes-into-non-eligible-masters/329261>\
**Category:** Elasticsearch\
**Created:** [April 4, 2023, 1:03am UTC](https://discuss.elastic.co/t/would-be-a-good-idea-to-turn-all-data-nodes-into-non-eligible-masters/329261 "2023-04-04T01:03:44Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Bruno\_Arruda](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/bruno_arruda/32/104769_2.png) [@Bruno\_Arruda](https://discuss.elastic.co/u/Bruno_Arruda)\
**Post date:** [April 4, 2023, 1:03am UTC](https://discuss.elastic.co/t/would-be-a-good-idea-to-turn-all-data-nodes-into-non-eligible-masters/329261/1 "2023-04-04T01:03:44Z")

</div>

Hi,

Actually I have a cluster with 4 master nodes and 10 data nodes on a Kubernetes Cluster. Basically, each node causes my cluster to scale new hosts because of the anti-affinity default behavior, so my k8s got 14 nodes too. I was thinking about how to downsize my Elasticsearch cluster (and consequently my k8s cluster) and a idea I had was to deploy the stack with 3 (or even 2) master nodes and promote the 10 data nodes to non-eligible master, in addiction to their other roles. I guess that many voting nodes could prevent a split brain scenario, because at any time, even with no communication between 2 master nodes, the other nodes would vote and get into a consensus about which one is the master.

Is that a advisable step or shouldn't I do it?

I saw this [related question](https://discuss.elastic.co/t/whats-the-downside-of-making-all-data-nodes-master-eligible/175495/1) but the asker was talking about turning all data nodes into eligible master nodes, which I know is not the best idea because of the risk of a highly demanded data node being elected as the cluster master and not having enough resources to act as a master.

---

<div class="post-metadata">

**Author:** ![warkolm](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/warkolm/32/39224_2.png) [@warkolm](https://discuss.elastic.co/u/warkolm)\
**Post date:** [April 4, 2023, 1:07am UTC](https://discuss.elastic.co/t/would-be-a-good-idea-to-turn-all-data-nodes-into-non-eligible-masters/329261/2 "2023-04-04T01:07:36Z")

</div>

Welcome to our community! 😃

> [@Bruno\_Arruda](#):
>
> 4 master nodes

Take a look at [Voting configurations | Elasticsearch Guide [8.7] | Elastic](https://www.elastic.co/guide/en/elasticsearch/reference/8.7/modules-discovery-voting.html#_even_numbers_of_master_eligible_nodes) for some commentary about using an even number of master eligible nodes.

TLDR - uneven numbers are better for master eligible counts and it's totally ok to make the others data only.

---

<div class="post-metadata">

**Author:** ![Christian\_Dahlqvist](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/christian_dahlqvist/32/4617_2.png) [@Christian\_Dahlqvist](https://discuss.elastic.co/u/Christian_Dahlqvist)\
**Post date:** [April 4, 2023, 3:57am UTC](https://discuss.elastic.co/t/would-be-a-good-idea-to-turn-all-data-nodes-into-non-eligible-masters/329261/3 "2023-04-04T03:57:08Z")

</div>

> [@Bruno\_Arruda](#):
>
> Is that a advisable step or shouldn't I do it?

I do not see the point in this and don't think you should ever have more than one voting-only master node (if any). I would instead recommend setting up 3 small dedicated master nodes that you do not scale in or out and then only adjust the number of dedicated data nodes

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\
**Post date:** [May 2, 2023, 3:57am UTC](https://discuss.elastic.co/t/would-be-a-good-idea-to-turn-all-data-nodes-into-non-eligible-masters/329261/4 "2023-05-02T03:57:09Z")

</div>

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.
