Skip to content

opensearch-operator: dnsBase stays cluster.local, so securityadmin never reaches its target on the platform domain #4073

Description

Describe the bug
The vendored opensearch-operator chart sets manager.dnsBase: cluster.local (packages/system/opensearch-operator/charts/opensearch-operator/values.yaml:59, passed to the manager as an env var by the deployment template at line 90). packages/system/opensearch-operator/values.yaml does not override it and the package's only patch is leaderElection.diff. The platform runs on networking.clusterDomain, cozy.local by default (packages/core/platform/values.yaml:62).

At the pinned operator 2.8.0 the securityadmin job targets {svc}.{ns}.svc.{dnsBase}, so on every cluster with the default domain it retries a name that never resolves. Users and roles from the security config never apply, whichever CA signed the admin certificate. #2682 fixes the CA side and states this limitation; the DNS side is this issue.

Environment

  • Cozystack main, 2026-09-04
  • Any provider with the default cluster domain

To Reproduce

  1. Deploy an opensearch application with the security config enabled
  2. The securityadmin job in the release namespace retries <release>.<ns>.svc.cluster.local

Expected behaviour
The operator resolves in-cluster names on the platform's domain, and users and roles apply.

Actual behaviour
The job never completes.

Additional context
The platform hands the domain to system packages through the cozystack ConfigMap key cluster-domain (packages/core/platform/templates/apps.yaml:140); the operator package should set manager.dnsBase from it. The SANs of the certificates the operator issues use the same base and need the same fix.

Checklist

  • I have checked the documentation
  • I have searched for similar issues
  • I have included all required information
  • I have provided clear steps to reproduce
  • I have included relevant logs

Activity

  1. added
    kind/bugCategorizes issue or PR as related to a bug
    area/databaseIssues or PRs related to managed databases (postgres, mariadb, redis, etcd, kafka, clickhouse)
    on Sep 3, 2026
  2. lexfrei commented on Sep 17, 2026

    @lexfrei
    ContributorAuthor

    Fixed by #4185, merged 2026-09-10. It had no closing keyword, so this stayed open.

    The platform now passes networking.clusterDomain to the operator as manager.dnsBase, and two helm-unittest files pin it.

    On an existing installation two things stay manual, both are in the release note of #4185. A <release>-securityconfig-update Job already stuck on the old hostname has to be deleted once by hand. Old certificates keep their old SANs; nothing in-cluster verifies them, so they keep working. The operator never reissuing a certificate is #4211.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    area/databaseIssues or PRs related to managed databases (postgres, mariadb, redis, etcd, kafka, clickhouse)kind/bugCategorizes issue or PR as related to a bugtriage/needs-triageIndicates an issue needs triage by a maintainer

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions