5.5. 注册新集群

流程

1. 可选：运行以下命令，检查辅助安装程序中可用的 Operator：

   $ curl -s "https://api.openshift.com/api/assisted-install/v2/supported-operators" -H "Authorization: Bearer ${API_TOKEN}" | jq .

   Copy to Clipboard Toggle word wrap

   输出示例

   [
     "node-feature-discovery",
     "openshift-ai",
     "osc",
     "self-node-remediation",
     "mtv",
     "odf",
     "cnv",
     "mce",
     "pipelines",
     "servicemesh",
     "node-healthcheck",
     "node-maintenance",
     "lso",
     "fence-agents-remediation",
     "kube-descheduler",
     "serverless",
     "nvidia-gpu",
     "authorino",
     "nmstate",
     "amd-gpu",
     "kmm",
     "lvm"
   ]

   Copy to Clipboard Toggle word wrap

2. 运行以下命令，检查 Operator 是否支持指定的 OCP 版本、CPU 架构或平台：

   $ curl -s "https://api.openshift.com/api/assisted-install/v2/support-levels/features?openshift_version=4.13&cpu_architecture=x86_64&platform_type=baremetal" -H "Authorization: Bearer ${API_TOKEN}" | jq .features.SNO 

   1

   2

   Copy to Clipboard Toggle word wrap

   1

   按如下方式替换属性：

   • 对于 openshift_version，请指定 OpenShift Container Platform 版本号。此属性是必需的。
   • 对于 cpu_architecture，指定 x86_64,aarch64,arm64,ppc64le,s390x, 或 multi。此属性是可选的。
   • 对于 platform_type，指定 baremetal、none、nutanix、vsphere 或外部。此属性是可选的。

   2

   对于 Operator，为节点功能发现指定 .NODE-FEATURE-DISCOVERY, .OPENSHIFT-AI for OpenShift AI, .OSC for OpenShift 沙盒容器 .SELF-NODE-REMEDIATION for Self Node Remediation, .MTV for Migration Toolkit for Virtualization, .ODF for OpenShift Data Foundation, . .CNV for OpenShift Virtualization, .MCE for multicluster engine, .PIPELINES for OpenShift Pipelines, .SERVICEMESH for OpenShift Service Mesh, .NODE-HEALTHCHECK for Node Health Check, .NODE -MAINTENCE for Node Maintenance, .LSO for Local Storage Operator, .FENCE-AGENTS-REMEDIATION for Fence Agents 修复, .KUBE-DESCHEDULER for Kube Descheduler, .SERVERLESS for OpenShift Serverless, .NVIDIA-GPU for NVIDIA GPU, .AUTHORINO for Authorino, .NMSTATE for NMState, .AMD-GPU for AMD GPU, .KMM' for Kernel Module Management, 或 .LVM 用于 LVM 存储。

   输出示例

   "supported" 

   1

   Copy to Clipboard Toggle word wrap

   1

   可能的状态是"支持"、"dev-preview"、"tech-preview"和"不可用"。

3. 运行以下命令，检查指定的 OCP 版本、CPU 架构或平台的完整列表：

   $ curl -s "https://api.openshift.com/api/assisted-install/v2/support-levels/features?openshift_version=4.13&cpu_architecture=x86_64&platform_type=baremetal" -H "Authorization: Bearer ${API_TOKEN}" | jq

   Copy to Clipboard Toggle word wrap

   输出示例

   {
     "features": {
       "AMD_GPU": "dev-preview",
       "AUTHORINO": "dev-preview",
       "CLUSTER_MANAGED_NETWORKING": "supported",
       "CNV": "supported",
       "CUSTOM_MANIFEST": "supported",
       "DUAL_STACK": "supported",
       "DUAL_STACK_VIPS": "supported",
       "FENCE_AGENTS_REMEDIATION": "dev-preview",
       "FULL_ISO": "supported",
       "KMM": "dev-preview",
       "KUBE_DESCHEDULER": "dev-preview",
       "LSO": "supported",
       "LVM": "supported",
       "MCE": "supported",
       "MINIMAL_ISO": "supported",
       "MTV": "unavailable",
       "NMSTATE": "supported",
       "NODE_FEATURE_DISCOVERY": "dev-preview",
       "NODE_HEALTHCHECK": "dev-preview",
       "NODE_MAINTENANCE": "dev-preview",
       "NON_STANDARD_HA_CONTROL_PLANE": "unavailable",
       "NVIDIA_GPU": "dev-preview",
       "ODF": "supported",
       "OPENSHIFT_AI": "dev-preview",
       "OSC": "tech-preview",
       "OVN_NETWORK_TYPE": "supported",
       "PIPELINES": "dev-preview",
       "PLATFORM_MANAGED_NETWORKING": "unsupported",
       "SDN_NETWORK_TYPE": "supported",
       "SELF_NODE_REMEDIATION": "dev-preview",
       "SERVERLESS": "dev-preview",
       "SERVICEMESH": "dev-preview",
       "SINGLE_NODE_EXPANSION": "supported",
       "SKIP_MCO_REBOOT": "unavailable",
       "SNO": "supported",
       "USER_MANAGED_LOAD_BALANCER": "unavailable",
       "USER_MANAGED_NETWORKING": "supported",
       "VIP_AUTO_ALLOC": "dev-preview"
     }
   }

   Copy to Clipboard Toggle word wrap

4. 运行以下命令指定要安装的 Operator：

   $ curl -s -X POST https://api.openshift.com/api/assisted-install/v2/clusters \
   -H "Authorization: Bearer ${API_TOKEN}" \
   -H "Content-Type: application/json" \
   -d "$(jq --null-input \
      --slurpfile pull_secret ~/Downloads/pull-secret.txt '
   {
      "name": "testcluster",
      "openshift_version": "4.15",
      "cpu_architecture" : "x86_64",
      "base_dns_domain": "example.com",
      "olm_operators": [
     { "name": "mce" } 

   1

   
     ,
     { "name": "odf" }
     ,
    { "name": "amd-gpu" }
       ]
      "pull_secret": $pull_secret[0] | tojson
   }
   ')" | jq '.id'

   Copy to Clipboard Toggle word wrap

   1

   列出要安装的 Operator。为 OpenShift Virtualization 指定 cnv for OpenShift Virtualization, mtv for Virtualization, mce for multicluster engine, odf for Red Hat OpenShift Data Foundation, lvm for Logical Volume Manager Storage, openshift-ai for OpenShift AI, osc for OpenShift 沙盒容器, nmstate for Kubernetes NMState, amd-gpu for AMD GPU, kmm for Kernel Module Management, node-feature-discovery for Node Feature Discovery, nvidia-gpu for NVIDIA and self-node-remediation 用于自助服务修复。安装 Operator 会自动激活任何依赖的 Operator。

流程

1. 可选：运行以下命令来检查辅助安装程序中可用的 Operator 捆绑包：

   $ curl -s "https://api.openshift.com/api/assisted-install/v2/operators/bundles" -H "Authorization: Bearer ${API_TOKEN}" | jq .

   Copy to Clipboard Toggle word wrap

   输出示例

   [
     {
       "description": "Run virtual machines alongside containers on one platform.",
       "id": "virtualization",
       "operators": [
         "kube-descheduler",
         "node-maintenance",
         "mtv",
         "nmstate",
         "fence-agents-remediation",
         "cnv",
         "node-healthcheck",
       ],
       "title": "Virtualization"
     },
     {
       "description": "Train, serve, monitor and manage AI/ML models and applications using GPUs.",
       "id": "openshift-ai",
       "operators": [
         "authorino",
         "odf",
         "openshift-ai",
         "amd-gpu",
         "node-feature-discovery",
         "nvidia-gpu",
         "pipelines",
         "servicemesh",
         "serverless",
         "kmm"
       ],
       "title": "OpenShift AI"
     }
   ]

   Copy to Clipboard Toggle word wrap

2. 可选：运行以下命令来检查哪些 Operator 与一个特定的捆绑包关联：

   $ curl -s "https://api.openshift.com/api/assisted-install/v2/operators/bundles/virtualization" -H "Authorization: Bearer ${API_TOKEN}" | jq . 

   1

   Copy to Clipboard Toggle word wrap

   1

   为 Virtualization Operator 捆绑包指定 虚拟化，或为 OpenShift AI Operator 捆绑包指定 openshift-ai。示例指定了 Virtualization Operator 捆绑包。

   输出示例

   {
     "description": "Run virtual machines alongside containers on one platform.",
     "id": "virtualization",
     "operators": [
       "kube-descheduler",
       "node-maintenance",
       "mtv",
       "nmstate",
       "fence-agents-remediation",
       "cnv",
       "node-healthcheck",
     ],
     "title": "Virtualization"
   }

   Copy to Clipboard Toggle word wrap

3. 运行以下命令，安装与捆绑包关联的 Operator：

   $ curl -s -X POST https://api.openshift.com/api/assisted-install/v2/clusters \
   -H "Authorization: Bearer ${API_TOKEN}" \
   -H "Content-Type: application/json" \
   -d "$(jq --null-input \
      --slurpfile pull_secret ~/Downloads/pull-secret.txt '
   {
      "name": "testcluster",
      "openshift_version": "4.15",
      "cpu_architecture" : "x86_64",
      "base_dns_domain": "example.com",
      "olm_operators": [ 

   1

   
     { "name": "node-healthcheck" }
   
     { "name": "fence-agents-remediation" }
   
     { "name": "kube-descheduler" }
   
     { "name": "mtv" }
   
     { "name": "nmstate" }
   
     { "name": "node-maintenance" }
   
     { "name": "cnv" } 

   2

   
   
     { "name": "nmstate" }
       ]
        "pull_secret": $pull_secret[0] | tojson
   }
   ')" | jq '.id'

   Copy to Clipboard Toggle word wrap

   1

   在您要安装的 Operator 捆绑包中指定 Operator。示例列出了虚拟化捆绑包的 Operator。

   2

   注意以下几点：

   • 在虚拟化捆绑包中，指定 cnv 会在后台自动安装 lso。

   • 在 OpenShift AI Operator 捆绑包中：

     • 指定 nvidia-gpu 会自动安装 node-feature-discovery。
     • 指定 amd-gpu 会自动安装 kmm。

流程

1. 将负载均衡器配置为可以从所有主机访问，并可访问以下服务：

   • OpenShift Machine Config Operator (MCO)- 在 control plane 节点上。
   • OpenShift API - 在 control plane 节点上。
   • Ingress Controller - 在 compute (worker)节点上。

   详情请参阅 配置用户管理的负载均衡器（第 1 步和 2 步 ）。

2. 配置集群的 DNS 记录，使其以用户管理的负载均衡器的前端 IP 地址为目标。您必须在负载均衡器上为集群 API 和应用程序的 DNS 服务器更新记录：

   • 配置 DNS 记录以使您的主 API 可以访问：

     <load_balancer_ip_address> <record_name> api.<cluster_name>.<base_domain>

     Copy to Clipboard Toggle word wrap

   • 配置 DNS 记录，通过入口控制器将外部流量路由到您的应用程序：

     <load_balancer_ip_address> <record_name> apps.<cluster_name>.<base_domain>

     Copy to Clipboard Toggle word wrap

   • 对于 vSphere，配置 DNS 记录以支持网络中的内部 API 访问：

     <load_balancer_ip_address> <record_name> api-int.<cluster_name>.<base_domain>

     Copy to Clipboard Toggle word wrap

   详情请参阅 配置用户管理的负载均衡器（第 3 步）。

3. 在 Assisted Installer API 集群定义中添加以下配置：

   1. 将 user_managed_networking 和 load_balancer 字段设置为以下值：

      "user_managed_networking": false,
      "load_balancer": {
              "type": "user-managed"
      }

      Copy to Clipboard Toggle word wrap

      详情请参阅 更改网络管理类型。

   2. 指定 Ingress 和 API VIP。它们应当与负载均衡器 IP 地址对应：

          "ingress_vips": [
              {
                  "cluster_id": "<cluster-id>",
                  "ip": "<load-balancer-ip>"
              }
          ],
          "api_vips": [
              {
                  "cluster_id": "<cluster-id>",
                  "ip": "<load-balancer-ip>"
              }
          ]

      Copy to Clipboard Toggle word wrap

   3. 指定机器网络列表以确保以下内容：

      • 每个节点至少有一个网卡(NIC)，至少有一台机器网络 IP 地址。

      • 负载均衡器 IP （也是 API VIP 和 Ingress VIP）至少包含在其中一个机器网络中。

        Example

         "machine_networks": [
            {
                "cidr": "<hosts-cidr-1>",
                "cluster_id": "<cluster-id>"
            },
            {
                "cidr": "<hosts-cidr-2>",
                "cluster_id": "<cluster-id>"
            },
            {
                "cidr": "<load-balancer-cidr>",
                "cluster_id": "<cluster-id>"
            }
        ]

        Copy to Clipboard Toggle word wrap

      如需了解更多详细信息，请参阅 机器网络。