Addressed issues¶
The following issues have been addressed in the Mirantis OpenStack for Kubernetes 21.5 release:
[17115][Update] Fixed the issue with the
status.providerStatus.releaseRefs.previous.name
field in the Cluster object for Ceph not changing during the MOS cluster update. If you have previously applied the workaround as described in [17115] Cluster update does not change releaseRefs in Cluster object for Ceph, manually add thesubresources
section back to theclusterworkloadlock
CRD:kubectl edit crd clusterworkloadlocks.lcm.mirantis.com # add here 'subresources' section: spec: versions: - name: v1alpha1 subresources: status: {}
[17477][Update] Fixed the issue with StackLight in HA mode placed on controller nodes being not deployed or cluster update being blocked. Once you update your MOS cluster from the Cluster release 6.18.0 to 6.19.0, roll back the workaround applied as described in [17477] StackLight in HA mode is not deployed or cluster update is blocked:
Remove
stacklight
labels from worker nodes. Wait for the labels to be removed.Remove the custom
nodeSelector
section from the cluster spec.
[16103][OpenStack] Fixed the issue with the Glance client returning the
HTTPInternalServerError
error while operating with a volume if Glance was configured with the Cinder backend TechPreview.[14678][OpenStack] Fixed the issue with instance being inaccessible through floating IP upon floating IP quick reuse when using a small floating network.
[16963][OpenStack] Fixed the issue with Ironic failing to provide nodes on deployments with OpenStack Victoria.
[16241][Tungsten Fabric] Fixed the issue causing failure to update a port, or security group assigned to the port, through the Horizon web UI.
[17045][StackLight] Fixed the issue causing the
fluentd-notifications
pod failing to track the RabbitMQ credentials updates in the Secret object.[17573][StackLight] Fixed the issue with OpenStack notifications missing in Elasticsearch and the Kibana
notification-*
index being empty.