OpenSearch provider v0.2.0 adds dedicated node roles, UI version upgrades, and safer rollouts

•

The OpenSearch provider v0.2.0 brings new topology options, more control over credentials and certificates, and safer upgrades.

  • Dedicated node roles: split the cluster into separate cluster-manager, data, and optional coordinator nodes instead of a single all-in-one node pool.
  • Bring your own secrets: seed the admin password from spec.userSecretRef, and supply the HTTP certificate from a Kubernetes Secret or a cert-manager issuer.
  • Version changes from the UI: edit a cluster’s OpenSearch version directly in the OpenEverest UI, without the API or manifests.
  • Safer upgrades: helm upgrade runs a preflight check that stops if a cluster is mid-upgrade or mid-restart, and changes a new provider would make to existing clusters wait for spec.maintenance approval.
  • Gentler node drains: a PodDisruptionBudget ensures that draining Kubernetes nodes evicts only one OpenSearch node at a time.
  • Dashboards fix: OpenSearch Dashboards no longer get stuck with “Another OpenSearch Dashboards instance appears to be migrating” after a slow first cluster start. Dashboards now start only once the cluster is fully initialized.

Upgrades from 0.1.0 were tested on Kubernetes 1.35; existing clusters stay Ready without restarts. To learn more, see the v0.2.0 release notes.