Unable to deploy Elasticsearch 8 with helm as single node

[robertvanloenhout]

I'd like to deploy Elasticsearch 8.5.1 using helm and only use a single node to reduce memory usage on minikube.
This issue is similar to #312, but I'll summarize it here because that issue is closed
because originally it was not an issue regarding Elasticsearch 8.

The deployment is on minikube 1.28.0 with helm 3.8.2

When I set replicas to 1 then elasticsearch pod will produce these warning messages and never become operational:

master not discovered or elected yet, an election requires at least 2 nodes

We can set discovery type to single-node by adding it to the values file

We can do that like this:

esConfig:
  elasticsearch.yml: |
    discovery:
      type: single-node

Or we can do it like this:

extraEnvs:
  - name: discovery.type
    value: single-node

This produces the following error:

setting [cluster.initial_master_nodes] is not allowed when [discovery.type] is set to [single-node]

We can try setting the initial master nodes it to empty like this:

esConfig:
  elasticsearch.yml: |
    discovery:
      type: single-node
    cluster:
      initial_master_nodes: ""

This still produce the following error:

setting [cluster.initial_master_nodes] is not allowed when [discovery.type] is set to [single-node]

We can also try it like this:

extraEnvs:
  - name: discovery.type
    value: single-node
  - name: cluster.initial_master_nodes
    value: ""

This will produce this error:

null-valued setting found for key [cluster.initial_master_nodes]

We can try setting the initial master nodes to null like this:

esConfig:
  elasticsearch.yml: |
    discovery:
      type: single-node
    cluster:
      initial_master_nodes: null

or like this:

extraEnvs:
  - name: discovery.type
    value: single-node
  - name: cluster.initial_master_nodes
    value: null

Either way the elasticsearch pod will faill immediately with the following error:

null-valued setting found for key [cluster.initial_master_nodes] found

[dpfaffenbauer]

I figured it out:

values:
      roles: []
      extraEnvs:
        - name: discovery.type
          value: single-node
        - name: node.roles
          value: master,data,data_content,data_hot,data_warm,data_cold,ingest,ml,remote_cluster_client,transform

[robertvanloenhout]

@dpfaffenbauer Thanks! This seems to work. I'll try it our for some time and close this issue if I don't encounter any problems.
Still hope that this can be made easier though some day.

[robertvanloenhout]

@dpfaffenbauer Unfortunately after a restart it didn't work anymore. I have the following warning message:

This node is a fully-formed single-node cluster with cluster UUID [uK64OL3nSTyOiW9wPZWBPQ], but it is configured as if to discover other nodes and form a multi-node cluster via the [discovery.seed_hosts=[elasticsearch-master-headless]] setting. Fully-formed clusters do not attempt to discover other nodes, and nodes with different cluster UUIDs cannot belong to the same cluster. The cluster UUID persists across restarts and can only be changed by deleting the contents of the node's data path(s).

[dpfaffenbauer]

@robertvanloenhout Haven't had that issue yet, will try to fix it once I get it.

[yonzmeer]

Any update on this?

[dpfaffenbauer]

We had the issue too now, I manually patched the statefulset to fix it.

[ShrawanLakhe]

@dpfaffenbauer how did you work on the fixes?

i have similar issues with elastic search 8.7

[justmike1]

same issue
@dpfaffenbauer what did you manually change?