Google Distributed Cloud air-gapped 1.12.2 release notes

Cluster management:

• User clusters with Kubernetes version 1.27.x might have node pools that fail to initialize.
• The required IPv4 PodCIDR is not available.

File and block storage:

• When upgrading from 1.11.1 to 1.12.2, the file-netapp-trident subcomponent rollout might fail.

Firewall:

• bm-system-machine-init is failing on the first node. Default IDPS firewall policies don't support the organization custom IPs for the Direct Connect (DX) Interconnect.

Hardware security module (HSM):

• A rotatable secret for hardware security modules is in an unknown state.

Lower networking:

• Network switches preloaded with version lower than 9.3.10 may fail to bootstrap.
• Some connections to the org-admin node time out.

Networking:

• GDC fails to create switch ACLs from traffic policies during the initial bootstrapping process.

NTP server:

• The Node OS has unsynchronized time.

Object storage:

• Object storage buckets might not be ready after root org upgrade.

Monitoring:

• Configuring the ServiceNow webhook results in Lifecycle Management (LCM) re-reconciling and reverting the changes made to the ConfigMap object mon-alertmanager-servicenow-webhook-backend and the Secret object mon-alertmanager-servicenow-webhook-backend in the mon-system namespace.
• The mon-common subcomponent doesn't deploy the Istio Telemetry object on the mon-system namespace.
• The metrics storage class is incorrectly defined in the configuration.
• The mon-prober-backend-prometheus-config ConfigMap gets reset to include no probe jobs, and alert MON-A0001 is triggered.

Physical servers:

• A NodePool has a server in unknown state during creation.
• The node firmware upgrade fails on an org.

System artifact registry:

• The job service pod is not ready.

Ticketing system:

• The ticketing system knowledge base sync fails.
• The ticketing system has no healthy upstream.

Vertex AI:

• The MonitoringTarget shows a Not Ready status when user clusters are being created, causing pre-trained APIs to continually show an Enabling state in the user interface.

Virtual machines:

• The importer pods are failing or stuck.
• Virtual machine disks might take a long time to provision.
• VMRuntime might not be ready due to network-controller-manager installation failure.

Upgrade:

Upgrade:

• The unet-nodenetworkpolicy-infra subcomponent fails during upgrade.
• System cluster fails during upgrade from 1.11.x to 1.12.2.
• The file-observability subcomponent fails on the org-1-system-cluster when upgrading from 1.11.x to 1.12.2.
• The HSMupgrade fails when upgrading from 1.11.x to 1.12.2.
• Loki pods are stuck in a terminating state for more than 1.5 hours when upgrading from 1.11.x to 1.12.2.
• SSH for a VM with management IP and the cilium logs fails when upgrading from 1.11.x to 1.12.2.
• The object storage upgrade shows an error during the postflight or preflight check.
• The mz-etcd subcomponent updates spec.deployTarget and spec.Namespace causing the upgrade from 1.11.x to 1.12.x to fail.
• An NTP OSPolicy failure blocks all other OSPolicies from running.