OpenSearch Dashboards 3.1: Unable to deploy via Helm chart

OpenSearch Dashboards
1

Versions: OpenSearch Dashboards 3.1 (OpenSearch 3.1; Kubernetes 1.31)

Describe the issue: I have a working OpenSearch 3.1 instance which was deployed via Helm chart with the only customization being setting the admin password. But I am having no luck deploying OpenSearch Dashboards 3.1 via the Helm chart (version 3.1). I am using all of the default values with no customizations. Neither the documentation nor the Helm chart README file indicate that any additional configuration or setup is required.

Unfortunately, while the OpenSearch Dashboards pod starts, it generates no log messages. So there are no clues why it is failing. :frowning: After about 5 minutes, the pod restarts…and the process repeats itself.

I suspect the issue may be related to TLS certificates although without any log messages it is hard to confirm that. And, according to the documentation, the default deployment uses HTTP and shouldn’t require any certs. I do NOT see any log messages on the OpenSearch side that indicates the OpenSearch Dashboards pod has attempted to connect.

Has anyone successfully deployed OpenSearch Dashboards 3.1 (or 3.0) via the Helm chart?

Configuration:

Relevant Logs or Screenshots:

2

@GSmith What platform did you use for your Kuberentes cluster (k8s, k3 Rancher, Minikube, AKS, EKS)?

3

@GSmith I’ve just tested and I had no issues running 3.1 OpenSearch and OpenSearch Dashboards on k3 Rancher.

image

image
image1057Ă—484 65.4 KB

4

@pablo Thanks for responding so quickly. It is good to know that the Helm charts are working for someone. My environment is a Kubernetes 1.31.2 cluster composed of nodes running Rocky Linux 8.7 running inside our corporate environment. I should be able to try things out with a different cluster tomorrow.

UPDATE: Your response reminded me that we have occasionally seen an issue with running OSD in one of our internal availability zones. I checked and some of the nodes in this cluster were, in fact, running in that AZ. I cordoned those nodes off, forcing the OSD pod to a different AZ…and OSD appears to have come up fine!

Thanks for the confirmation things should be working which forced me to look at my local environmental factors.

Regards,

Greg

1 Like