VMware Telco Cloud Automation 2.1 | 21 JUL 2022 | Build - VM-based: 20142564, HA-based: 20142097

Check for additions and updates to these release notes.

What's New

  • CaaS Infrastructure Upgrade and Redesign

    VMware Telco Cloud Automation 2.1 offers improved CaaS operability by upgrading the current CaaS infrastructure API version to v2. With this upgrade, users now have increased control over cluster failures and can view the status of all the components at a granular level, giving them the ability to act on a failure while the cluster creation is in progress. Also, during the cluster creation process, the user can edit or delete a node pool or an add-on at any point in the event of an error.

    The v2 upgrade provides the following enhancements:

    NOTE: These new features are only available on v2 clusters.

    • Granular upgrade support - The user can now upgrade VMware Telco Cloud Automation Control Plane and Worker node pools separately.
    • AKO Operator support for clusters - AVI Load-Balancer Kubernetes Operator Lifecycle Management offers improved network life cycle management.
    • Support for Tanzu Kubernetes Grid extensions Prometheus and Fluentbit.
    • The user can now edit Calico and Antrea while the cluster creation is in progress.
    • Removal of tasks moves events to a Declarative Model.
    • Stretch Cluster Support on the UI - Stretched cluster on VMware Cloud allows a Service Provider to manage multiple vCenter Kubernetes clusters on the VMware Cloud.
    • A new user interface that is based on the upgraded v2 APIs.

    NOTE: The existing CaaS v1 APIs care still available with VMware Telco Cloud Automation 2.1.

  • SSH Access to Appliances

    The user can access the SSH terminal of the backing VMware Telco Cloud Automation Control Plane (TCA-CP) appliance through the Virtual Infrastructure tab.

    NOTE: This functionality is available for users with System Administrator permissions.

  • Improved Concurrency and Throughput Capabilities within VMware Telco Cloud Automation

    VMware Telco Cloud Automation 2.1 improves concurrency-handling across the product. Features such as Infrastructure Automation (Cell Site Hosts), CNFs, and Cluster Automation now have improved numbers for quicker and stable production-facing deployments. For more information on the scale and concurrency limits of VMware Telco Cloud Automation, see https://configmax.vmware.com.

  • Support for Helm Dependency

    VMware Telco Cloud Automation supports Helm dependency for scaling, re-configuring, and terminating a CNF instance.

  • Diagnosis on a CaaS Cluster or a Node Pool

    The user can now run diagnosis on a CaaS Cluster or a Node Pool from the VMware Telco Cloud Automation UI, with multiple preset tests to choose from. The user can use these tests for a health check before an operation such as pre-upgrade or post-upgrade, or perform health check on a specific component such as ETCD or CNI. VMware Telco Cloud Automation provides users with separate diagnosis tests for Management Clusters, Workload Clusters, and Node Pools, enabling a thorough diagnosis throughout the stack.

  • Kubectl Access

    Secure Kubectl access enables restricted access to VMware Tanzu Kubernetes Grid through VMware Telco Cloud Automation. This feature gives users the ability to issue one-time passwords and tokens. These tokens, moreover, have expiry dates and system administrators have the ability to revoke them.

  • Active Directory Support

    Integration with external Active Directory system for user authentication, in addition to vCenter Server authentication.

  • VIO Support Extension

    VIO support extension allows enhanced platform awareness features to be configured automatically on VMware Telco Cloud Platform through VMware Telco Cloud Automation. Importantly, automated VIO configurations and customizations support new high-performance applications and related services (5G, vRAN) that require these applications. New configuration options are now available through the VNFD and CSAR.

  • HA-Based VMware Telco Cloud Automation Improvements

    • Support for upgrading cloud-native VMware Telco Cloud Automation deployed in an airgapped environment.
    • Support for deploying cloud-native VMware Telco Cloud Automation with customized cluster sizes.
  • IPV6 Support

    VMware Telco Cloud Automation supports IPv6 for a new deployment in an airgapped environment. The user can either use IPv6 or IPv4, but cannot use a combination of both IPv6 and IPv4.

    Note:

    • When using the IPv6, the user can register components only with FQDN.
    • The user must deploy DNS, DHCP, and NTP using IPv6.
    • The user must deploy vCenter and VMware ESXi server using IPv6.
    • VMware NSX-T and VMware vRO do not support IPv6. Any feature of VMware Telco Cloud Automation that uses these products cannot work in an IPv6 environment.
    • At present, IPv6 does not support the following:
      • Network Slicing
      • Cloud-native deployment using user interface
      • Infrastructure Automation
  • EKS-A Validation Support

    You can now instantiate CNFs on EKS-A Kubernetes clusters. This improvement in EKS functionality allows the Service Provider to extend the environments in which their workloads can be managed, supporting lifecycle management of CNFs running on EKS-A.

Important Notes

  • Appliance Size Limit Updates and One off Upgrade for 2.1

    VMware Telco Cloud Automation 2.1 has updated the VM Appliance Size limits. CPU, Memory, and HDD are all increased from the default limits that were present until VMware Telco Cloud Automation 2.0. The limits are increased for improved scale performance numbers, new services, and better support.

    All newly deployed VMware Telco Cloud Automation 2.1 appliances (TCA-M, TCA-CP and TCA-Bootstrapper) have the new limits.

    Upgrading from VMware Telco Cloud Automation 2.0 to 2.1 comes with an additional step to ensure that the new limits are applied correctly on the new appliances.

    For details on upgrade, see the VMware Telco Cloud Automation User Guide.

  • AVI objects (aviinfrasetting, gatewayclass, gateway)

    VMware Telco Cloud Automation 2.1 does not support creation of AVI objects (aviinfrasetting, gatewayclass, gateway CRs) on Workload clusters. You can still leverage the NF onboarding ability for a customized PaaS configuration until a generic cluster policy is available.

  • Customize Prometheus add-on

    VMware Telco Cloud Automation 2.1 only exposes a few variables of Prometheus to the UI. However, through the YAML content, you can customize additional variables that are not available through the UI. The steps are:

    1. Log on to the management cluster and download imgpkg executable binary from https://vmwaresaas.jfrog.io/artifactory/generic-cnf-dev/airgap/imgpkg-linux-amd64-v0.12.0.
    2. Run kubectl get package -A | grep prometheus to get Prometheus package name with version:
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ kubectl get package -A | grep prometheus
      tanzu-package-repo-global   prometheus.tanzu.vmware.com.2.27.0+vmware.1-tkg.1                         prometheus.tanzu.vmware.com                          2.27.0+vmware.1-tkg.1           22h23m13s
      tanzu-package-repo-global   prometheus.tanzu.vmware.com.2.27.0+vmware.2-tkg.1                         prometheus.tanzu.vmware.com                          2.27.0+vmware.2-tkg.1           22h23m13s
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$
    3. Select one package name  and set image_url environment variable that points to the Prometheus package image URL with command image_url=$(kubectl -n tanzu-package-repo-global get packages prometheus.tanzu.vmware.com.2.27.0+vmware.1-tkg.1 -o jsonpath='{.spec.template.spec.fetch[0].imgpkgBundle.image}').
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ image_url=$(kubectl -n tanzu-package-repo-global get packages prometheus.tanzu.vmware.com.2.27.0+vmware.1-tkg.1 -o jsonpath='{.spec.template.spec.fetch[0].imgpkgBundle.image}')
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ echo $image_url
      projects-stg.registry.vmware.com/tkg/packages/standard/prometheus@sha256:27af034c1c77bcae4e1f7f6d3883286e34419ea2e88222642af17393cd34e46a
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$
    4. Use imgpkg tool to download the Prometheus bundle and get the default values.yaml file for the Prometheus add-on.
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ imgpkg pull -b $image_url -o /tmp/prometheus-package-2.27.0+vmware.1-tkg.1
      Pulling bundle 'projects-stg.registry.vmware.com/tkg/packages/standard/prometheus@sha256:27af034c1c77bcae4e1f7f6d3883286e34419ea2e88222642af17393cd34e46a'
        Extracting layer 'sha256:44798ebd112b55ea792f0198cf220a7eaed37c2abc531d6cf8efe89aadc8bff2' (1/1)
       
      Locating image lock file images...
      One or more images not found in bundle repo; skipping lock file update
       
      Succeeded
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ cp /tmp/prometheus-package-2.27.0+vmware.1-tkg.1/config/values.yaml prometheus-data-values.yaml
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ ls prometheus-data-values.yaml
      prometheus-data-values.yaml
    5. To install the Prometheus add-on with the customized configuration, customize the default values.yaml and copy the new YAML content (from the line prometheus: onwards) to the VMware Telco Cloud Automation UI.
      	#@data/values
      ---
       
      #! The namespace in which to deploy prometheus.
      namespace: tanzu-system-monitoring
       
      prometheus:                --- from this line onwards
        deployment:
          #! Number of prometheus servers
          replicas: 1
       
          #! Replaces the prometheus-server container args with the ones provided below (i.e. Does not append).
          containers:
            args: []
            resources: {}
          podAnnotations: {}
          podLabels: {}
          configmapReload:
            containers:
              args: []
              resources: {}
        #! Prometheus service configuration
        service:
          type: "ClusterIP"
          port: 80
          targetPort: 9090
          labels: {}
          annotations: {}
        pvc:
          annotations: {}
          storageClassName: null
          accessMode: ReadWriteOnce
          storage: "150Gi"
        #! The prometheus configuration
        config:
          prometheus_yml: |
            global:
              evaluation_interval: 1m
              scrape_interval: 1m
              scrape_timeout: 10s
            rule_files:
            - /etc/config/alerting_rules.yml
              ...

    Note:

    • Prometheus add-on depends on CSI add-on. Make sure that the CSI add-on is provisioned normally before you install Prometheus add-on.
    • VMware Telco Cloud Automation vsphere-csi add-on only supports create ReadWriteOnce PVC by default. If you want to use ReadWriteMany or ReadOnlyMany accessMode for Prometheus PVC, you must use VMware Telco Cloud Automation nfs-client add-on or other CSI to make sure that PVC can be created successfully.

  • Customize FluentBit Add-on

    VMware Telco Cloud Automation 2.1 only exposes a few variables of FluentBit to the UI. However, through the YAML content, you can customize additional variables that are not available through the UI. The steps are:

    1. Log on to the Management cluster and download imgpkg executable binary from https://vmwaresaas.jfrog.io/artifactory/generic-cnf-dev/airgap/imgpkg-linux-amd64-v0.12.0.
    2. Run kubectl get package -A | grep fluent to get the FluentBit package name with version:
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ kubectl get package -A | grep fluent
      tanzu-package-repo-global   fluent-bit.tanzu.vmware.com.1.7.5+vmware.1-tkg.1                          fluent-bit.tanzu.vmware.com                          1.7.5+vmware.1-tkg.1            22h43m33s
      tanzu-package-repo-global   fluent-bit.tanzu.vmware.com.1.7.5+vmware.2-tkg.1                          fluent-bit.tanzu.vmware.com                          1.7.5+vmware.2-tkg.1            22h43m33s
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$
    3. Select one package name  and set image_url environment variable that points to the FluentBit package image url with the command image_url=$(kubectl -n tanzu-package-repo-global get packages fluent-bit.tanzu.vmware.com.1.7.5+vmware.2-tkg.1 -o jsonpath='{.spec.template.spec.fetch[0].imgpkgBundle.image}')
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ image_url=$(kubectl -n tanzu-package-repo-global get packages fluent-bit.tanzu.vmware.com.1.7.5+vmware.2-tkg.1 -o jsonpath='{.spec.template.spec.fetch[0].imgpkgBundle.image}')
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ echo $image_url
      projects-stg.registry.vmware.com/tkg/packages/standard/fluent-bit@sha256:264bfbefb2430c422cb691637004ed5dbf4a4d0aac0b0cb06ee19a3c81b1779e
    4. Use imgpkg tool to download the FluentBit bundle and get the default values.yaml file for the FluentBit add-on.
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ imgpkg pull -b $image_url -o /tmp/fluent-bit.tanzu.vmware.com.1.7.5+vmware.2-tkg.1
      Pulling bundle 'projects-stg.registry.vmware.com/tkg/packages/standard/fluent-bit@sha256:264bfbefb2430c422cb691637004ed5dbf4a4d0aac0b0cb06ee19a3c81b1779e'
        Extracting layer 'sha256:9117de69d77240fa52e38e1a434bcda69b9185f5fafd2a85eedbd06c06beb57c' (1/1)
       
      Locating image lock file images...
      One or more images not found in bundle repo; skipping lock file update
       
      Succeeded
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ cp /tmp/fluent-bit.tanzu.vmware.com.1.7.5+vmware.2-tkg.1/config/values.yaml ./fluent-bit-data-values.yaml
      capv@mgmt-sb-sbm-control-plane-5wqtn [ ~ ]$ ls fluent-bit-data-values.yaml
      fluent-bit-data-values.yaml
    5. To install the FluentBit add-on with the customized configuration, customize the default values.yaml and copy the new YAML content (from the line fluent-bit: onwards) to the VMware Telco Cloud Automation UI .
      #@data/values
      #@overlay/match-child-defaults missing_ok=True
      ---
      namespace: "tanzu-system-logging"
      #! Required params for supported output plugins
      fluent_bit:                                    --------------- from this line onwards
      config:
      #! https://docs.fluentbit.io/manual/administration/configuring-fluent-bit/variables
      service: |
      [Service]
      Flush 1
      Log_Level info
      Daemon off...
  • Infrastructure Automation

    • Starting from 2.2, VMware Telco Cloud Automation deprecates the support creating central data center, regional data center, and compute cluster using Infrastructure Automation. However, you can add the pre-deployed data centers through Infrastructure Automation in a VM-based deployment.
    • Starting from 2.2, VMware Telco Cloud Automation deprecates the support for CN bootstrapping using Infrastructure Automation.

  • Delete Domain Behavior Changed

    In VMware Telco Cloud Automation 2.1, the process to delete the domains has changed. Current and previous behaviors are as follows:

    Previous behavior:

    To delete a domain, the user had to remove the domain definition from the domains list and the back-end used to take care of the deletion.

    Below is sample cloud spec file with two domains test1 and test2.

    {
        "domains": [
            {
                "name": "test1",
                ...
            },
            {
                "name": "test2",
                ...
            }
        ],
        "settings": {
            ...
        },
        "appliances": [
            ...
        ],
        "images": {
            ...
        }
    }

    Below is modified cloud spec to be posted for deleting the domain test1 (remove test1 from domains list).

    {
        "domains": [
            {
                "name": "test2",
                ...
            }
        ],
        "settings": {
            ...
        },
        "appliances": [
            ...
        ],
        "images": {
            ...
        }
    }

    Current behavior:

    To delete domains, the user has to add a list of strings (names of the domains you need to delete) to the deleteDomains field in the cloud spec. For example,"deleteDomains" : ["cdc1", "rdc1"].

    It is optional to include/exclude the domain to be deleted in the domains list. Below is an example for the new behavior. We provide test1 in deleteDomains list to delete that domain.

    {
        "domains": [
            {
                "name": "test1",
                ...
            },
            {
                "name": "test2",
                ...
            }
        ],
        "settings": {
            ...
        },
        "appliances": [
            ...
        ],
        "images": {
            ...
        },
        "deleteDomains": ["test1"]
    }
  • VM-Based New Resource Requirements

    The VM-based VMware Telco Cloud Automation deployments have additional resource requirement options. For more information, see the System Requirements > VM-Based Deployment section of the VMware Telco Cloud Automation 2.1 Deployment Guide.

Known Issues

Infrastructure Automation

  • Deleting multiple ESXi hosts at once is not recommended due to vSAN limitations. This is specifically for non-Cell Site hosts.

    Delete each ESXi host individually and let the deletion process complete before deleting the next ESXi host.

  • When adding a pre-deployed host to a pre-deployed Cell Site Group domain by its IP Address, the host gets provisioned, but the Hostconfig Profile fails to apply.

    Add the ESXi host to the vCenter Server using FQDN only. Then add the pre-deployed host to Infrastructure Automation using FQDN.

  • Infrastructure Automation requires to know the Single Sign On (SSO) credential of the VMware Telco Cloud Automation Manager, without which, the host profile application fails.

    The current implementation fetches this information from Central Data Center (CDC), which is configured with the vCenter account used in the VMware Telco Cloud Automation Manager SSO configuration. There could be deployment topologies where the user may not provision Central Data Center but directly deploy Regional Data Center (RDC). In such a circumstance, Infrastructure Automation fails to apply Host Profile.

    Create a Central Data Center by importing the vCenter and the account used to configure single sign on the VMware Telco Cloud Automation Manager.

  • Infrastructure Automation expects a password with minimum 13 characters for vCenter on the Global Appliance Configuration page. This might create a problem for pre-deployed setups where the vCenter password can be less than 13 characters.

    • On the Global Appliance Configuration page, provide a dummy 13 character password and save the configuration.
    • On the domain level, go to the Appliance Override section and change the vCenter password as required.
  • CSG creation on Predeployed domain fails with error message root password cannot be empty for VC in domain cdc.

    On a freshly installed VMware Telco Cloud Automation or a VMware Telco Cloud Automation that has no cloud spec defined so far, if we upload a valid cloud spec without appliance IDs, the cloud spec file gets accepted. However, further cloud spec change/uploads operations always require appliance passwords as there are no appliance IDs.

    From the API, we can upload passwords each time. But from the UI, the passwords are not included on any change and save operation. VMware Telco Cloud Automation displays the error message root/admin/audit password needed for {appliance} for {domain}.

    Re-uploading the same cloud spec generates the appliance IDs. To use the latest cloud spec with appliance IDs, from the Domain section, click Refresh. After completing the steps, you can perform the operations normally.

  • Infrastructure Automation Cloud-Native Upgrade - Release specific TKG image URL must be manually updated in the Images section of Infrastructure Automation.

    Upload the TKG image URL manually in the Images section of Infrastructure Automation.

Active Directory

  • New Active Directory users with the option to 'Change Password on next logon' cannot log in to VMware Telco Cloud Automation.

    Set the user password in Active Directory before logging in to VMware Telco Cloud Automation.

  • If the logon username is more than 20 characters, the login itself will work but the group retrieval of the user will fail, causing the login to VMware Telco Cloud Automation to fail.

    VMware Telco Cloud Automation does not support Active Directory users with username that include more than 20 characters.

    Ensure that the usernames are less than 20 characters in length.

Cluster Automation

  • The vsphere-csi daemonset does not always load the latest configuration after restart, due to which the nodes are not labeled with multi-zone information (zone and region information). In addition, the topologyKeys are not populated either.

    1. Login to the Workload Cluster Control Plane node.
    2. Restart the vsphere-csi node daemonset manually after applying the csi params from VMwareTelco Cloud Automation and verifying that the operation is complete from VMware Telco Cloud Automation.
    3. Run the following command to restart vsphere-csi daemonset: kubectl rollout restart ds vsphere-csi-node -n kube-system
    4. Verify the node labels after 2 or more minutes.
  • Node Pool Scale Out operation does not work on Workload Clusters (v2) that do not have their Control Plane and Node Pools upgraded to 1.22.x.

    Upgrade the Control Plane and Node Pool to 1.22.x and then perform the Scale Out operation.

  • Deployment of IPv6 based Workload Clusters with Kubernetes versions prior to 1.22.x will result in TKG known limitations where the VSPHERE_CONTROL_PLANE_ENDPOINT IP Address is assigned to the node and host network pods.

    This may further result in scenarios where there are potential IP Address conflicts when a Workload Cluster has more than one Control Plane node.

    It is recommended that IPv6 base Workload Cluster is deployed with Kubernetes version 1.22.x (or later).

  • Management Cluster IPv6  endpoint IP cannot be accessed at times.

    When a user logs in to the Control Plane VM console, and then runs ip addr, it displays eth0 with dadfailed tentative.

    [root@tca /home/admin]# kubectl config get-contexts
    CURRENT   NAME                  CLUSTER    AUTHINFO       NAMESPACE
    *         lcm-mc-admin@lcm-mc   lcm-mc     lcm-mc-admin
              minikube              minikube   minikube       default
    [root@tca /home/admin]# kubectl get pods -n tca-system
    kubectl get nodepolicy -A
    ^C
    [root@tca /home/admin]# kubectl get nodes
    ^C
    [root@tca /home/admin]# kubectl config use-context lcm-mc-admin@lcm-mc
    Switched to context "lcm-mc-admin@lcm-mc".
    [root@tca /home/admin]# kubectl get nodes
    Unable to connect to the server: dial tcp [2013:930::700]:6443: i/o timeout

    Reboot or restart the Cluster Control Plane nodes from the vCenter client.

  • Cluster LCM operations might fail due to a root cause where the api-server of the corresponding Management Cluster is restarted.

    This happens because the Control Plane nodes run on low performance hosts, which results in some pods with CrashLoopBackOff error messages such as no route to host, or leader election lost.

    Restart the pods within the Management Cluster that are in a CrashLoopBackOff state.

  • Objects on Avi Controller are not deleted automatically when the load-balance-and-ingress-service add-on is uninstalled from the Workload cluster.

    Delete the objects from the Avi Controller UI directly.

  • Updating ako-operator add-on configuration is not supported. For example: Avi Controller credentials, certificates, and so on.

    Uninstall and re-install the ako-operator add-on with the new configuration.

  • New Cluster deployments might become slow if there are other clusters with many Node Pools.

    None.

  • New vCenterSub configurations cannot be added to existing stretched Workload Clusters that already have other vCenterSub configurations.

    None.

  • Failed to upgrade Management Cluster: One control plane node displays the status as SchedulingDisabled.

    You cannot end the pods of an existing Control Plane node because it can not remove cgroup path. This is similar to the upstream Kubernetes issue: https://github.com/kubernetes/kubernetes/issues/97497

    1. Restart the affected Control Plane node. Post this, the upgrade will continue automatically in the backend.
    2. Retry the upgrade operation for the Cluster from VMware Telco Cloud Automation UI.
  • Transforming Management and Workload cluster results in the Node Policy status temporarily going to a Configuring state.

    This automatically transfers to a normal state after a few minutes.

    None.

  • Workload Cluster upgrade failes on Kubernetes version v1.20.8 to v1.22.9.

    It is strongly recommended to create / ensure that the Management Clusters have 3 or more Control Plane nodes before upgrading to VMware Telco Cloud Automation to 2.1.

    Case 1: Upgrade fails and reports an error message "timeout: poll control plane ready for removing SCTPSupport=true"

    Workaround:

    1. Retry upgrade on the user interface.
    2. If retry fails, check the CAPI log to find logs related to the new Control Plane machine and status.
      1. The new Control Plane machine status is Running. The new Control Plane node cloud-init log contains "could not find a JWS signature in the cluster-info ConfigMap for token ID".
      2. Contact tech-support.
      # Confirm the new Control Plane
      [root@10 /home/admin]# kubectl get machine -n cdc-work-cluster1-v1-21-2
      NAME                                                          CLUSTER                     NODENAME                                                      PROVIDERID                                       PHASE     AGE     VERSION
      cdc-work-cluster1-v1-21-2-work-master-control-plane-29xgb     cdc-work-cluster1-v1-21-2   cdc-work-cluster1-v1-21-2-work-master-control-plane-29xgb     vsphere://423b959d-59aa-47d9-56cf-d657b165ab0a   Running   2d14h   v1.21.2+vmware.1
      cdc-work-cluster1-v1-21-2-work-master-control-plane-9ptnm     cdc-work-cluster1-v1-21-2   cdc-work-cluster1-v1-21-2-work-master-control-plane-9ptnm     vsphere://423b1c01-4b7f-c27d-6458-fb6a4e3c3984   Running   2d10h   v1.21.2+vmware.1
      cdc-work-cluster1-v1-21-2-work-worker-np1-575bc4c56-h89d6     cdc-work-cluster1-v1-21-2   cdc-work-cluster1-v1-21-2-work-worker-np1-575bc4c56-h89d6     vsphere://423b7444-e75f-066a-eb67-d9631c939e3d   Running   2d14h   v1.21.2+vmware.1
      cdc-work-cluster1-v1-21-2-work-worker-np10-65b5c46bd6-2kcqm   cdc-work-cluster1-v1-21-2   cdc-work-cluster1-v1-21-2-work-worker-np10-65b5c46bd6-2kcqm   vsphere://423b993d-dbda-2dbd-0943-d0800ac6a0aa   Running   2d14h   v1.21.2+vmware.1
       
      # Look for the IP address of the new Control Plane node
      [root@10 /home/admin]# kubectl get machine -n cdc-work-cluster1-v1-21-2 cdc-work-cluster1-v1-21-2-work-master-control-plane-9ptnm -oyaml
       
      # log in through SSH
      [root@10 /home/admin]# ssh capv@10.176.119.36
       
      capv@cdc-work-cluster1-v1-21-2-work-master-control-plane-9ptnm [ ~ ]$ sudo su
      root [ /home/capv ]# cat /var/log/cloud-init-output.log
       
      # If you see "could not find a JWS signature in the cluster-info ConfigMap for token ID" message in the cloud-init log, and if kube-vip does not float on this faulty Control Plane, you can delete the faulty Control Plane.
      ***If kube-vip floats on the faulty Control Plane, contact TKG team to debug and provide a workaround.
       
      # Log in to TCA-CP and switch to root user
      [root@10 /home/admin]# kubectl edit  deployment -n capi-kubeadm-bootstrap-system       capi-kubeadm-bootstrap-controller-manager
       
      # Update bootstrap-token-ttl=15m to bootstrap-token-ttl=25m. This will cause capi-kubeadm-bootstrap-system pod to restart
       
      # Delete faulty Control Plane machine
      kubectl delete machine -n cdc-work-cluster1-v1-21-2 cdc-work-cluster1-v1-21-2-work-master-control-plane-9ptnm
       
      # After the faulty Control Plane machine is deleted, the new Control Plane is created automatically. After the faulty Control Plane is deleted and the new Control Plane is running, retry upgrade on the user interface.
       
      # If the New Control Plane node cloud-init still reports error "could not find a JWS signature ", increase the value of bootstrap-token-ttl.

    Case 2: Error: could not find server "cdc-mgmt-cluster"

    Workaround:

    1. Navigate to /root/.config/tanzu/config.yaml and confirm if Management cluster context exists. If not, contact TKG tech-support to debug.
    2. After finding the root cause, perform the following workaround.

      Old Value:

      [root@10 /home/admin]# cat /root/.config/tanzu/config.yaml
      apiVersion: config.tanzu.vmware.com/v1alpha1
      clientOptions:
        cli:
          discoverySources:
          - oci:
              image: projects.registry.vmware.com/tkg/packages/standalone/standalone-plugins:v0.11.6-1-g90440e2b_vmware.1
              name: default
          edition: tkg
        features:
          cluster:
            custom-nameservers: "false"
            dual-stack-ipv4-primary: "false"
            dual-stack-ipv6-primary: "false"
          global:
            context-aware-cli-for-plugins: "true"
          management-cluster:
            custom-nameservers: "false"
            dual-stack-ipv4-primary: "false"
            dual-stack-ipv6-primary: "false"
            export-from-confirm: "true"
            import: "false"
            network-separation-beta: "false"
            standalone-cluster-mode: "false"
      kind: ClientConfig
      metadata:
        creationTimestamp: null

      New Value:

      [root@10 /home/admin]# cat /root/.config/tanzu/config.yaml
      apiVersion: config.tanzu.vmware.com/v1alpha1
      clientOptions:
        cli:
          discoverySources:
          - oci:
              image: projects.registry.vmware.com/tkg/packages/standalone/standalone-plugins:v0.11.6-1-g90440e2b_vmware.1
              name: default
          edition: tkg
        features:
          cluster:
            custom-nameservers: "false"
            dual-stack-ipv4-primary: "false"
            dual-stack-ipv6-primary: "false"
          global:
            context-aware-cli-for-plugins: "true"
          management-cluster:
            custom-nameservers: "false"
            dual-stack-ipv4-primary: "false"
            dual-stack-ipv6-primary: "false"
            export-from-confirm: "true"
            import: "false"
            network-separation-beta: "false"
            standalone-cluster-mode: "false"
      current: cdc-mgmt-cluster
      kind: ClientConfig
      metadata:
        creationTimestamp: null
      servers:
      - managementClusterOpts:
          context: cdc-mgmt-cluster-admin@cdc-mgmt-cluster
          path: /root/.kube-tkg/config
        name: cdc-mgmt-cluster
        type: managementcluster

    3. Verify if the information is correct:
      [root@10 /home/admin]# tanzu login --server cdc-mgmt-cluster
      ✔  successfully logged in to management cluster using the kubeconfig cdc-mgmt-cluster
      Checking for required plugins...
      All required plugins are already installed and up-to-date
      [root@10 /home/admin]#

    It is strongly recommended to create / ensure that the Management Clusters have 3 or more Control Plane nodes before upgrading to VMware Telco Cloud Automation to 2.1.

  • Workload cluster of the Control Plane node fails when scaling out from 1 to 3.

    Retry the scale out operation on Control Plane nodes.

  • Node policy is still configuring after 30 minutes in CaaS kubecluster operator pipeline.

    Cluster node stays on a step during boot up with no IP Address being assigned to the node. This results in the node being in a NotReady status.

    Reset the node Virtual Machine through the vCenter client.

  • Unable to add Harbor as add-on.

    This issue can be reproduced by performing the following steps:

    1. Create or update the Harbor add-on of a cluster.
    2. Add or update Harbor on the same cluster through Partner Systems.

    Use the following workaround options:

    • Option 1: Add, update, or remove Harbor associations only through the Partner Systems tab.
    • Option 2: Perform add, update, or remove operations only on the Harbor add-on.
  • After deleting a transformed cluster, creating a v1 cluster using the same name or IP address fails.

    Transform a Workload cluster abc with endpoint IP 10.10.10.10.  After transforming the cluster, delete the Workload cluster abc using the v2 Delete API option. Now, when you create a cluster using v1 API with the name abc or with endpoint IP 10.10.10.10, it fails.

    None.

  • Collect Tech-Support Bundle UI issue.

    After deleting a transformed cluster, the Collect Tech-Support bundle UI still lists the deleted cluster names.

    None.

  • nodeconfig-operator add-on stays in Configuring and its AddonDeployReady condition stays in False.

    Cluster connection is not ready for a while and nodeconfig-operator add-on stays in configuring phase.

    Restart the tca-kubecluster-operator pod.

  • Workload Cluster upgrade failson a cloud-native environment.

    Cluster Upgrade fails intermittently with the following error message:

    unknown command "login" for "tanzu"
    Did you mean this?
    plugin
    : Error: unknown command "login" for "tanzu"
    Did you mean this?
    plugin
    Run 'tanzu --help' for usage.
    1. Log in to the VMware Telco Cloud Automation Control Plane SSH session using admin user and then switch to root user.
    2. Edit /root/ .config/tanzu/config.yaml.
    kind: ClientConfig  
      metadata:    
          creationTimestamp: null       
              network-separation-beta: "false"        
              standalone-cluster-mode: "false"  
     current: cdc-mgmt-cluster  
     kind: ClientConfig  
     metadata:    
      creationTimestamp: null
    • Remove the duplicate line for kind: ClientConfig.
    • Also fix the indentation for network-separation-beta and standalone-cluster-mode.
  • vsphere-csi-node is CrashLoopBackOff in TKG 1.5.3 RC1 build for Workload cluster spanning vCenters.

    vsphere-csi is not supported in scenarios where the Cluster is deployed across multiple vCenters.

    None.

  • Update AKO configuration with AKODeploymentConfig does not take effect.

    Updating load-balance-and-ingress-service (AKO) add-on configuration is not supported.

    Uninstall and re-install the load-balance-and-ingress-service (AKO) add-on with the new configuration.

  • Prometheus/Alertmanager has issues when deployed on TKG.

    Within the Prometheus add-on, the alertmanager pod might be in a CrashLoopBackOff state in vCenter 70u2 and vCenter 70u3 deployments.

    Provide the cluster.advertise-address in the alertmanager deployment YAML.

    containers:
      - name: prometheus-alertmanager
         image: "prom/alertmanager:v0.20.0"
         imagePullPolicy: "IfNotPresent"
         args:
         - --config.file=/etc/config/alertmanager.yml
         - --storage.path=/data
         - --cluster.advertise-address=127.0.0.1:9093
  • Fluent-bit pod remains in the CrashLoopBackoffstate on Worker nodes where the cpu-manager-policy is set to static.

    Change the cpu-manager-policy to none.

  • Workload Cluster upgrade failson a VM-based VMware Telco Cloud Automation environment.

    To fix this issue, log in to either the bootstrapper VM or Telco Cloud Automation Control Plane and remove duplicate entries or fix misaligned lines in the /root/.config/tanzu/config.yaml file.

  • On v2 clusters, the Delete Node Pool operation hangs if the v2 cluster contains a label with a dot in it.

    If delete Node Pool operation is not initiated, follow the steps:

    1. Edit the node pool and delete the label that contains a dot within it.
    2. Delete the node pool.

    If delete Node Pool operation is initiated and the delete operation gets hang, follow the steps from KB <<link to be updated>>.

  • Diagnosis on Management/Workload cluster fails.

    Diagnosis on Management/Workload cluster fails with the error message diagnosis on non-running management cluster cdc-mgmt-cluster-v1-22-9 is not supported. Status:

    When running diagnostics, VMware Telco Cloud Automation may display an error that the Management cluster is not running even when the cluster is running. 

    1. Log in to the Appliance Management portal for VMware Telco Cloud Automation Control Plane.
    2. Click Appliance Summary tab.
    3. Under Telco Cloud Automation Services, restart the TCA Diagnosis API Service.
  • On v2 clusters, if node pools and TBRs are not upgraded, VMware Telco Cloud Automation displays an error icon instead of info icon.

    VMware Telco Cloud Automation displays the following as an error: Current TBR is not compatible. Upgrade to latest TBR version. You can perform Edit Nodepool/Cluster and update the TBR reference. This message is displayed under the following scenarios:

    1. After transforming the imported clusters from v1 to v2.
    2. After upgrading the control plane in v2 clusters (on node pools).

    Cause: The availability of latest TBRs causes the existing TBRs to be marked as not compatible for cluster control plane and node pools.

    Purpose: The purpose of the error message is to indicate to users that newer versions of TBRs are available.

    The error message provides information that an upgrade is available. Users need not perform any action as long as the cluster is within the supported Kubernetes version.

HA-based VMware Telco Cloud Automation

  • Configuring Active Directory with TLS in Cloud-Native VMware Telco Cloud Automation environment fails.

    HA-based VMware Telco Cloud Automation does not support configuring Active Directory with TLS.

    None.

  • Cannot switch from ActiveDirectory to vCenter authentication for Appliance Manager UI if Cloud-Native VMware Telco Cloud Automation is deployed with Active Directory.

    You cannot switch the authentication model from Active Directory to vCenter authentication for HA-based VMware Telco Cloud Automation.

    None.

  • Unable to add VMware Cloud Director VIM.

    VIM registration is not working in a cloud-native VMware Telco Cloud Automation deployed with VMware Cloud Director.

    This issue will be addressed in an upcoming patch release of VMware Telco Cloud Automation.

Node Customization

  • vmconfig-operator only support access to vCenter Server with port 443.

    vmconfig-operator supports Clusters deployed within vCenter Server environments that run on port 443 only. vCenter Server environments with custom ports are not supported.

    None.

  • Post Control Plane BOM Upgrade, the Edit Node Pool operation lists Control Plane Kubernetes version templates instead of Node Pool Kubernetes version templates.

    Re-select the VMware Telco Cloud Automation BOM release version to the existing version. The VM templates will now be listed correctly.

  • In the CSAR Designer, Add Network Adapter with device type vmxnet3 does not show Resource Name.

    1. Design a CSAR with targetDriver set to [igb_uio/vfio_pci] and resourceName.
    2. Onboard the CSAR.
    3. Download the CSAR and delete from VMware Telco Cloud Automation.
    4. Edit the download CSAR in a text editor:
      1. Remove 'targetDriver'.
      2. Add 'interfaceName' [the interface name inside the GUEST OS for this adapter]. For example:
    network:
            devices:
              - deviceType: vmxnet3
                networkName: network1
                resourceName: res2
                interfaceName: net1
                count: 1
                isSharedAcrossNuma: false

    5. Upload the edited CSAR file to VMware Telco Cloud Automation.

    To edit a CSAR file:

    1. Unzip the CSAR.
    2. Edit the Definitions/NFD.yaml with the changes listed in the steps above.
    3. Run zip -r <new_name>.csar TOSCA-Metadata/ Definitions/ Artifacts/ NFD.mf

Kubectl and SSH

  • On a vim editor, open terminal contents display repeated characters at the end.

    The terminal opened through VMware Telco Cloud Automation displays repeated characters at the end within the vim editor.

    Within vim, set the following: set :nottyfast

    For details, see vim.

  • kubectl command overwrites from the beginning of the line after few characters on the remote ssh terminal.

    Resize the SSH terminal window before using lengthy commands on the console.

Network Slicing

  • Network slicing is not supported for Active Directory Authentication-enabled environments in greenfield and brownfield scenarios.

    None.

Partner System

  • Adding custom port Harbor to the v2 clusters is causing Harbor add-on to freeze.

    Add the custom port Harbor to Partner Systems first and then register the same to the cluster as an extension.

Appliance UI

  • Kernel Arguments applied to Node pool during NF instantiation are not visible on the UI.

    Cluster Automation CaaS UI does not render the list of Kernel arguments applied to Workload ClustersNode Pools as part of Node Customizations.

    The customizations are applied and are visible through the APIs.

CNF Lifecycle Management

  • Terminating and then instantiating an existing CNF on a transformed cluster fails.

    After transforming a cluster from from v1 to v2 , terminating and reinstatiating the network function displays the error Unable to find node pool: np1 for cluster.

    Reselect the cloud and the node pool during CNF reinstantiation.

  • CSAR deployment fails for 38+ helm charts at a time through VMware Telco Cloud Automation

    VMware Telco Cloud Automation does not support more than 25 values.yaml files in a single instantiation.

    Split the CSAR into 2 with at most 25 values file in each.

Backup and Restore

  • Configuring SFTP server with a passphrase throws an error.

    When configuring the SFTP server with a passphrase for Backup and Restore, the following error is displayed:

    “java.io.FileNotFoundException: /common/appliance-management/backup-restore/id_rsa_backup_restore.pub (No such file or directory)”, “Failed to configure FTP server!"

    None.

  • When configuring the FTP server, Backup Directory Path must be relative path.

    The tool tip suggests to provide absolute path, whereas the user must provide relative path.

    None.

  • After backing up and restoring from one VMware Telco Cloud Automation appliance to another, operations on the CaaS clusters fail.

    A KB article with a workaround will be published soon.

Resolved Issues

Infrastructure Automation

  • Must allow get/post/put/patch/delete API calls to VMware Telco Cloud Automation/Control Plane for clusters after Infrastructure Automation is migrated successfully.

  • CDC deployment requires the user to manually upload the certificate for provisioning to start.

  • When pre-deployed domain is added without using appliance overrides, state of domain should not become provisioned.

  • Apply Hostconfig profile task for firmware upgrade, failing with error "503 Service Unavailable".

  • Not providing passwords for appliance overrides during domain modification is successful but fails during provisioning.

  • Domain can be provisioned without any global settings present, but won't proceed further, status not conveyed in UI.

Cloud Native

  • WLD domain deployment failed during VMware Telco Cloud Automation Control Plane with error Failed to deploy TCA CLUSTER 69247203-c518-4bdf-95e4-1e77cf3d078d. Reason: compatibility file (/root/.config/tanzu/tkg/compatibility/tkg-compatibility.yaml) already exists.

  • No mongo dump in Techsupport bundle for VMware Telco Cloud Automation Control Plane only deployment.

  • Multiple issues post backup and restore.

CNF LCM

  • CNF instance should not have Heal and Scale to level in the list of available operations.

  • VMware Telco Cloud Automation shows the scaling details when reconfiguring a CNF.

Cluster Operator

  • No IP when you restart a Worker Node for CPI with multi vCenter.

Appliance

  • httpd does not start after upgrading from 1.9.1 to 2.0.

Security Fixes

  • Privilege escalation to root user.

  • Insecure sudo config - Privilege Escalation from admin user to root.

  • VMware Telco Cloud Automation Manager and Control Plane is exposed to Clickjack Attack.

check-circle-line exclamation-circle-line close-line
Scroll to top icon