Addressed issues

Issues addressed in the MKE 3.8.18 release include:

  • [FIELD-8397][MKE-15108] Fixed an issue wherein the ucp backup command hung on air-gapped clusters with usage reporting enabled. The anonymous usage-reporting network call had no timeout.

  • [FIELD-8342][MKE-14820] Fixed an issue wherein the ucp-rethinkdb-exporter metrics endpoint on TCP port 9055 accepted unauthenticated plain HTTP requests. The endpoint now requires mutual TLS authentication with a client certificate issued by the MKE CA, and ucp-metrics scrapes it over HTTPS.

  • [FIELD-8370][MKE-14819] Fixed an issue wherein calico-kube-controllers failed to start its controllers on clusters that use the Calico Kubernetes datastore (KDD), because the deployment ran under the calico-node service account, which lacks access to the KubeControllersConfiguration resource.

  • [MKE-14582] Fixed an issue wherein restoring MKE from a backup did not restore the Swarm Ingress setting. MKE now stores the Swarm Ingress enablement state in the MKE configuration, which the backup includes, and redeploys Interlock during the restore if Swarm Ingress was enabled when the backup was taken.