Addressed issues

The following issues have been addressed in the Mirantis OpenStack for Kubernetes 21.3 release:

  • [13422][OpenStack] Fixed the issue with some Redis pods remaining in Pending state and causing failure to update the Cluster release.

  • [12511][OpenStack] Fixed the issue with Kubernetes nodes getting stuck in the Prepare state during the MOS managed cluster update.

  • [13233][StackLight] Fixed the issue with low memory limits for StackLight Helm controller causing update failure.

  • [12917][StackLight] Fixed the issue with prometheus-tf-vrouter-exporter pods failing to start on Tungsten Fabric nodes with DPDK. To remove the nodeSelector definition specified when applying the workaround:

    1. Remove the nodeSelector.component.tfVrouterExporter definition from the stackLight helmRelease (.spec.providerSpec.value.helmReleases) values of the Cluster resource.

    2. Remove the label using the following command. Do not remove the last dash sign.

      kubectl label node <node_name> tfvrouter-fix-
      
  • [11961][Tungsten Fabric] Fixed the issue with members failing to join the RabbitMQ cluster after the tf-control nodes reboot.