[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間:2025-09-04 (世界標準時間)。"],[[["\u003cp\u003eMaintenance planning is essential for all deployed resources and their underlying components, as policies for each can be distinct.\u003c/p\u003e\n"],["\u003cp\u003eVirtual machine (VM) maintenance policies determine how VMs react to updates, crashes, or host events, offering options like live migration or restart.\u003c/p\u003e\n"],["\u003cp\u003eVM maintenance policies differ from service maintenance policies, even when the services are running on VMs, such as GKE clusters on Compute Engine VMs.\u003c/p\u003e\n"],["\u003cp\u003eVMs with GPUs or TPUs do not support live migration during host maintenance events, which can lead to downtime and workload disruptions.\u003c/p\u003e\n"],["\u003cp\u003eNetwork infrastructure maintenance, such as for Cloud Router and Cloud Interconnect, requires following specific recommendations to minimize downtime and retain connections.\u003c/p\u003e\n"]]],[],null,["# Consider resource maintenance\n\nBefore you create resources, you must consider maintenance requirements for each\nresource that you deploy. Additionally, you must consider maintenance on any\nunderlying resources.\n\nFor example, some services deploy to underlying Compute Engine VMs. The\nmaintenance policy that you set for your deployed service is distinct from the\nmaintenance policy on the underlying VMs.\n\nThe following example maintenance dynamics are provided to help you understand\nthe planning required to keep your resources available and running efficiently.\n\nSet virtual machine (VM) maintenance policies\n---------------------------------------------\n\nWhen you create VMs you set a maintenance policy that dictates VM behavior when\nthere is an update pending, a VM crashes, or other host events occur. For\nexample, you can create a policy to live migrate workloads to another VM, or\nshut down and restart the impacted VM.\n\nFor more information, see the following:\n\n- [About host events](/compute/docs/instances/host-maintenance-overview)\n- [Monitor and plan for a host maintenance event](/compute/docs/instances/monitor-plan-host-maintenance-event)\n\n### Distinguish between VM maintenance and service maintenance\n\nThe maintenance policy that you set for VMs is distinct from maintenance\npolicies that you set for services that run on your VMs.\n\nFor example, GKE deploys clusters on Compute Engine VMs. You\ncan set maintenance policies to control when some GKE cluster\nmaintenance happens, but those policies don't prevent automatic maintenance\ntriggered on the underlying Compute Engine VMs.\n\nTo learn more about maintenance policies for services running on\nCompute Engine VMs, review the respective documentation for those services.\n\n### Maintain workloads on VMs with GPUs or TPUs\n\nSome Compute Engine resources you create might have GPUs or TPUs attached. For\nexample, you might create VMs that use GPUs or TPUs to handle AI workloads. If a\nhost event occurs on these VMs, live migration from the impacted VM to a new VM\nis not supported. As a result, host maintenance events result in VM downtime and\npotential disruption to your workloads. To handle maintenance events on\nresources with GPUs or TPUs, see the following:\n\n- [Handle GPU host maintenance events](/compute/docs/gpus/gpu-host-maintenance)\n- [Introduction to Cloud TPU](/tpu/docs/intro-to-tpu)\n\nRetain connections during network infrastructure maintenance\n------------------------------------------------------------\n\nNetwork Connectivity products help you connect your peer networks to your\nVirtual Private Cloud networks. Google Cloud performs regular maintenance on this\ninfrastructure. To help prevent downtime during maintenance events, we recommend\nthat you follow the maintenance recommendations for each networking product, as\nin the following examples:\n\n- Cloud Router maintenance does not interrupt routing, but might require you\n to configure settings on your peer network router. For more information, see [Software maintenance and automated task restarts](/network-connectivity/docs/router/concepts/overview#maintenance).\n\n- Cloud Interconnect experiences regular automated maintenance, which might\n require you to set up notifications and create redundant connections. For more\n information, see [Infrastructure maintenance events](/network-connectivity/doocs/interconnect/support/infrastructure-maintenance-events)."]]