Dell ObjectScale 1.3 Administration Guide

Infrastructure Upgrade for ObjectScale Appliance - Operating System Upgrade

These steps are applicable only for ObjectScale Appliance deployment.

1. Upload the required lcm_manifest.json and linux-sles-15.4.tar.gz to the http-share service.

   # get http-share ingress-ip address
   # get IP address to upload/download files from http-share
   kubectl get svc rke2-ingress-nginx-controller -n kube-system --output jsonpath='{.status.loadBalancer.ingress[0].ip}'
    
   # upload the required artifacts
   cd objectscale-1.3.0-appliance-bundle/artifacts
   curl -i -k --create-dirs https://<ingress-ip>/httpshare/upload/bundle/objectscale/upgrade/1.3.0/ -F "file=@lcm_manifest.json"
   # upload the OS bundle as well for the appliance upgrade
   curl -i -k https://<ingress-ip>/httpshare/upload/bundle/objectscale/upgrade/1.3.0 -F "file=@linux-sles-15.4.tar.gz"

2. Upgrade the operating system using CLI.
   1. Ensure all the nodes are in Ready state.

      kubectl get nodes -o wide
      NAME    STATUS   ROLES                       AGE   VERSION          INTERNAL-IP     EXTERNAL-IP   OS-IMAGE                              KERNEL-VERSION          CONTAINER-RUNTIME
      cmo42   Ready    control-plane,etcd,master   25h   v1.24.7+rke2r1   10.249.253.42   <none>        SUSE Linux Enterprise Server 15 SP4   5.3.18-24.107-default   containerd://1.6.8-k3s1
      cmo43   Ready    control-plane,etcd,master   25h   v1.24.7+rke2r1   10.249.253.43   <none>        SUSE Linux Enterprise Server 15 SP4   5.3.18-24.107-default   containerd://1.6.8-k3s1

   2. Ensure all the pods within the system are in Running phase.
   3. Create the following os-update.yaml file for updating the operating system.
      NOTE:

      • You can assign a unique name for <lcmupdate-name>.
      • All the nodes are updated through the management (host/IP) names.
      • Identify the namespace of the CSI using the command helm list -A| grep -i csi-baremetal-operator. In the sample output below, the second column denotes the CSI namespace value, which is csi.

        csi-baremetal-operator                  csi             2               2023-09-29 07:27:11.013043982 +0000 UTC deployed        csi-baremetal-operator-1.3.0-121.2e006fb                1.3.0admin@dhcp-10-236-65-131:~>

        You can identify the namespace of the ObjectScale component using the command helm list -A| grep -i objectscale-manager. In the sample output below, the second column denotes the ObjectScale namespace value, which is objectscale.

        objectscale-manager                     objectscale     2               2023-09-29 07:31:45.231404122 +0000 UTC deployed        objectscale-manager-1.3.0-4184                          1.3.0-4184admin@dhcp-10-236-65-131:~>

        You can identify the namespace of the object store component using the command helm list -A| grep -i ecs-cluster. In the sample output below, the second column denotes the object store namespace value, which is objectscale.

        wilson-str1                             objectscale     2               2023-09-29 07:34:50.143272032 +0000 UTC deployed        ecs-cluster-1.3.0-4184                                  1.3.0-4184

      • In the below example, ObjectScale component is installed in the <objectscale-namespace>, CSI in <csi-namespace>, and object store in <object-store-namespace>. If both ObjectScale and object store components are installed in the <objectscale-namespace>, the additionalParamValue section changes to additionalParamValue: '{"lcm_nodehook_untaint_custom_namespaces":"<objectscale-namespace>, <csi-namespace>"}'.

      apiVersion: lcm-cluster.dell.com/v1
      kind: LCMUpdate
      metadata:
        name: <lcmupdate-name>
        namespace: kube-system
      spec:
          bundlePathBaseDirectory: "https://http-share.cmo:443/download/bundle/objectscale/upgrade/<OBJECTSCALE_VERSION>"
          infrastructureConfig:
            componentsToUpgrade :
            - componentType: "os"
              componentName: "sles"
              componentVersion: "15-SP4"
            customConfig:
              taint:
                key: "node.dell.com/drain"
                value: "planned-downtime"
                effect: "NoSchedule"
            nodeHookConfig:
              customPostUpgradeNodeHook:
              - env:
                  mode: sequential
                  profileName: lcm-cmo-post-maintenance
                  additionalParams:
                  - additionalParamName: optional_json
                    additionalParamValue: '{"lcm_nodehook_untaint_custom_namespaces":"<objectscale-namespace>, <csi-namespace>, <object-store-namespace>"}'
              customPreUpgradeNodeHook:
              - env:
                  additionalParams:
                  - additionalParamName: optional_json
                    additionalParamValue: '{"lcm_nodehook_taint_key":"node.dell.com/drain","lcm_nodehook_taint_value":"planned-downtime","lcm_nodehook_taint_effect":"NoSchedule"}'
                  mode: sequential
                  profileName: lcm-cmo-pre-maintenance
              skipNodeHooks: false
            nodeList: ["<comma_separated_cluster_node_ssh_IP_addresses>"]

      Below is a sample nodeList:

      nodeList: ["10.236.126.118","10.236.126.119","10.236.126.120","10.236.126.121","10.236.126.122","10.236.126.123","10.236.126.124","10.236.126.125"]

   4. Apply the resource to the target cluster and monitor the status.

      kubectl apply -f os-update.yaml
      kubectl get lcmupdate <lcmupdate-name> -o yaml
      

      NOTE:

      • This may take a few hours as this is a rolling update for each node.
      • If operating system LCM update fails abruptly, before retrying the upgrade on other nodes delete the existing operating system LCM Update and proceed further:

        kubectl delete lcmupdate <lcmupdate-name>