/

Console

• English
• Deutsch
• Español – América Latina
• Français
• Português – Brasil
• 中文 – 简体
• 日本語
• 한국어

Sign in

• Google Cloud SDK

Start free

gcloud container node-pools update

gcloud container node-pools update node-pool-1 --cluster=sample-cluster --enable-autoupgrade

gcloud container node-pools update node-pool-1 --cluster=example-cluster --consolidation-delay=3600s

For detailed information on the configuration usage, please refer to https://cloud.google.com/kubernetes-engine/docs/how-to/customize-containerd-configuration.

Note: Updating the containerd configuration of an existing cluster or node pool requires recreation of the existing nodes, which might cause disruptions in running workloads.

gcloud container node-pools update node-pool-1 --enable-kernel-module-signature-enforcement

gcloud container node-pools update node-pool-1 --cluster=example-cluster --enable-queued-provisioning
… and other required parameters, for more details see:
https://cloud.google.com/kubernetes-engine/docs/how-to/provisioningrequest

gcloud container node-pools update node-pool-1 --cluster=example-cluster --flex-start
and other required parameters, for more details see:
https://cloud.google.com/kubernetes-engine/docs/how-to/provisioningrequest

gcloud container node-pools update node-pool-1 --cluster=example-cluster --labels=label1=value1,label2=value2

gcloud container node-pools update node-pool-1 --cluster=example-cluster --max-run-duration=3600s

Supported values:
  * unspecified: Default behavior, currently the same as `arm`.
  * arm: kubernetes.io/arch=arm:NoSchedule taint will be added for ARM nodes.
  * none: No architecture taint will be applied.

gcloud container node-pools update node-pool-1 --cluster=example-cluster --node-architecture-taint-behavior=none

To read more about node-taints, see https://cloud.google.com/kubernetes-engine/docs/node-taints.

Examples:

gcloud container node-pools update node-pool-1 --cluster=example-cluster --node-labels=label1=value1,label2=value2

Updating the node pool's --node-labels flag applies the labels to the Kubernetes Node objects for existing nodes in-place; it does not re-create or replace nodes. New nodes, including ones created by resizing or re-creating nodes, will have these labels on the Kubernetes API Node object. The labels can be used in the nodeSelector field. See https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/ for examples.

gcloud container node-pools update node-pool-1 --cluster=sample-cluster --node-locations=us-central1-a,us-central1-b

Examples:

gcloud container node-pools update node-pool-1 --cluster=example-cluster --node-taints=key1=val1:NoSchedule,key2=val2:PreferNoSchedule

Examples:

gcloud container node-pools update example-node-pool --resource-manager-tags=tagKeys/1234=tagValues/2345
gcloud container node-pools update example-node-pool --resource-manager-tags=my-project/key1=value1
gcloud container node-pools update example-node-pool --resource-manager-tags=12345/key1=value1,23456/key2=value2
gcloud container node-pools update example-node-pool --resource-manager-tags=

Examples:

kubeletConfig:
  cpuManagerPolicy: static
  memoryManager:
    policy: Static
  topologyManager:
    policy: BestEffort
    scope: pod
  crashLoopBackOff:
    maxContainerRestartPeriod: '300s'
linuxConfig:
  accurateTimeConfig:
    enablePtpKvmTimeSync: true
  sysctl:
    net.core.somaxconn: '2048'
    net.ipv4.tcp_rmem: '4096 87380 6291456'
  hugepageConfig:
    hugepage_size2m: '1024'
    hugepage_size1g: '2'
  swapConfig:
    enabled: true
    bootDiskProfile:
      swapSizeGib: 8
  cgroupMode: 'CGROUP_MODE_V2'
  nodeKernelModuleLoading:
    policy: 'ENFORCE_SIGNED_MODULES'

List of supported kubelet configs in 'kubeletConfig'.

KEY	VALUE
cpuManagerPolicy	either 'static' or 'none'
cpuCFSQuota	true or false (enabled by default)
cpuCFSQuotaPeriod	interval (e.g., '100ms'. The value must be between 1ms and 1 second, inclusive.)
memoryManager	specify memory manager policy
topologyManager	specify topology manager policy and scope
podPidsLimit	integer (The value must be greater than or equal to 1024 and less than 4194304.)
containerLogMaxSize	positive number plus unit suffix (e.g., '100Mi', '0.2Gi'. The value must be between 10Mi and 500Mi, inclusive.)
containerLogMaxFiles	integer (The value must be between [2, 10].)
imageGcLowThresholdPercent	integer (The value must be between [10, 85], and lower than imageGcHighThresholdPercent.)
imageGcHighThresholdPercent	integer (The value must be between [10, 85], and greater than imageGcLowThresholdPercent.)
imageMinimumGcAge	interval (e.g., '100s', '1m'. The value must be less than '2m'.)
imageMaximumGcAge	interval (e.g., '100s', '1m'. The value must be greater than imageMinimumGcAge.)
evictionSoft	specify eviction soft thresholds
evictionSoftGracePeriod	specify eviction soft grace period
evictionMinimumReclaim	specify eviction minimum reclaim thresholds
evictionMaxPodGracePeriodSeconds	integer (Max grace period for pod termination during eviction, in seconds. The value must be between [0, 300].)
shutdownGracePeriodSeconds	integer (Grace period for pods terminating on node shutdown, in seconds. Allowed values: 0, 30, 120.)
shutdownGracePeriodCriticalPodsSeconds	integer (Grace period for critical pods terminating on node shutdown, in seconds. The value must be between [0, 120] and less than shutdownGracePeriodSeconds.)
allowedUnsafeSysctls	list of sysctls (Allowlisted groups: 'kernel.shm*', 'kernel.msg*', 'kernel.sem', 'fs.mqueue.*', and 'net.*', and sysctls under the groups.)
singleProcessOomKill	true or false
maxParallelImagePulls	integer (The value must be between [2, 5].)
crashLoopBackOff	specify crashloopbackoff thresholds

KEY	VALUE
policy	either 'Static' or 'None'

KEY	VALUE
policy	either 'none' or 'best-effort' or 'single-numa-node' or 'restricted'
scope	either 'pod' or 'container'

KEY	VALUE
memoryAvailable	quantity (e.g., '100Mi', '1Gi'. Represents the amount of memory available before soft eviction. The value must be at least 100Mi and less than 50% of the node's memory.)
nodefsAvailable	percentage (e.g., '20%'. Represents the nodefs available before soft eviction. The value must be between 10% and 50%, inclusive.)
nodefsInodesFree	percentage (e.g., '20%'. Represents the nodefs inodes free before soft eviction. The value must be between 5% and 50%, inclusive.)
imagefsAvailable	percentage (e.g., '20%'. Represents the imagefs available before soft eviction. The value must be between 15% and 50%, inclusive.)
imagefsInodesFree	percentage (e.g., '20%'. Represents the imagefs inodes free before soft eviction. The value must be between 5% and 50%, inclusive.)
pidAvailable	percentage (e.g., '20%'. Represents the pid available before soft eviction. The value must be between 10% and 50%, inclusive.)

KEY	VALUE
memoryAvailable	duration (e.g., '30s', '1m'. The grace period for soft eviction for this resource. The value must be positive and no more than '5m'.)
nodefsAvailable	duration (e.g., '30s', '1m'. The grace period for soft eviction for this resource. The value must be positive and no more than '5m'.)
nodefsInodesFree	duration (e.g., '30s', '1m'. The grace period for soft eviction for this resource. The value must be positive and no more than '5m'.)
imagefsAvailable	duration (e.g., '30s', '1m'. The grace period for soft eviction for this resource. The value must be positive and no more than '5m'.)
imagefsInodesFree	duration (e.g., '30s', '1m'. The grace period for soft eviction for this resource. The value must be positive and no more than '5m'.)
pidAvailable	duration (e.g., '30s', '1m'. The grace period for soft eviction for this resource. The value must be positive and no more than '5m'.)

KEY	VALUE
memoryAvailable	percentage (e.g., '5%'. Represents the minimum reclaim threshold for memory available. The value must be positive and no more than 10%.)
nodefsAvailable	percentage (e.g., '5%'. Represents the minimum reclaim threshold for nodefs available. The value must be positive and no more than 10%.)
nodefsInodesFree	percentage (e.g., '5%'. Represents the minimum reclaim threshold for nodefs inodes free. The value must be positive and no more than 10%.)
imagefsAvailable	percentage (e.g., '5%'. Represents the minimum reclaim threshold for imagefs available. The value must be positive and no more than 10%.)
imagefsInodesFree	percentage (e.g., '5%'. Represents the minimum reclaim threshold for imagefs inodes free. The value must be positive and no more than 10%.)
pidAvailable	percentage (e.g., '5%'. Represents the minimum reclaim threshold for pid available. The value must be positive and no more than 10%.)

KEY	VALUE
maxContainerRestartPeriod	duration (e.g., '30s', '1m'. The maximum duration the backoff delay can accrue to for container restarts. The value must be between [1s, 300s].)

KEY	VALUE
net.core.netdev_max_backlog	Any positive integer, less than 2147483647
net.core.rmem_default	Must be between [2304, 2147483647]
net.core.rmem_max	Must be between [2304, 2147483647]
net.core.wmem_default	Must be between [4608, 2147483647]
net.core.wmem_max	Must be between [4608, 2147483647]
net.core.optmem_max	Any positive integer, less than 2147483647
net.core.somaxconn	Must be between [128, 2147483647]
net.ipv4.tcp_rmem	Any positive integer tuple
net.ipv4.tcp_wmem	Any positive integer tuple
net.ipv4.tcp_tw_reuse	Must be {0, 1, 2}
net.ipv4.tcp_mtu_probing	Must be {0, 1, 2}
net.ipv4.tcp_max_orphans	Must be between [16384, 262144]
net.ipv4.tcp_max_tw_buckets	Must be between [4096, 2147483647]
net.ipv4.tcp_syn_retries	Must be between [1, 127]
net.ipv4.tcp_ecn	Must be {0, 1, 2}
net.ipv4.tcp_congestion_control	Supported values for COS: 'reno', 'cubic', 'bbr', 'lp', 'htcp'. Supported values for Ubuntu: 'reno', 'cubic', 'bbr', 'lp', 'htcp', 'vegas', 'dctcp', 'bic', 'cdg', 'highspeed', 'hybla', 'illinois', 'nv', 'scalable', 'veno', 'westwood', 'yeah'.
net.ipv4.neigh.default.gc_thresh1	Must be between [0, 262144]
net.ipv4.neigh.default.gc_thresh2	Must be between [512, 524288]
net.ipv4.neigh.default.gc_thresh3	Must be between [1024, 1048576]
net.netfilter.nf_conntrack_max	Must be between [65536, 4194304]
net.netfilter.nf_conntrack_buckets	Must be between [65536, 524288]. Recommend setting: nf_conntrack_max = nf_conntrack_buckets * 4
net.netfilter.nf_conntrack_tcp_timeout_close_wait	Must be between [60, 3600]
net.netfilter.nf_conntrack_tcp_timeout_time_wait	Must be between [1, 600]
net.netfilter.nf_conntrack_tcp_timeout_established	Must be between [600, 86400]
net.netfilter.nf_conntrack_acct	Must be {0, 1}
kernel.keys.maxkeys	Must be between [200, 1048576]
kernel.keys.maxbytes	Must be between [20000, 2097152]
kernel.shmmni	Must be between [4096, 32768]
kernel.shmmax	Must be between [0, 18446744073692774399]
kernel.shmall	Must be between [0, 18446744073692774399]
kernel.perf_event_paranoid	Must be {-1, 0, 1, 2, 3}
kernel.sched_rt_runtime_us	Must be [-1, 1000000]
kernel.softlockup_panic	Must be {0, 1}
kernel.yama.ptrace_scope	Must be {0, 1, 2, 3}
kernel.kptr_restrict	Must be {0, 1, 2}
kernel.dmesg_restrict	Must be {0, 1}
kernel.sysrq	Must be [0, 511]
fs.aio-max-nr	Must be between [65536, 4194304]
fs.file-max	Must be between [104857, 67108864]
fs.inotify.max_user_instances	Must be between [8192, 1048576]
fs.inotify.max_user_watches	Must be between [8192, 1048576]
fs.nr_open	Must be between [1048576, 2147483584]
vm.dirty_background_ratio	Must be between [1, 100]
vm.dirty_background_bytes	Must be between [0, 68719476736]
vm.dirty_expire_centisecs	Must be between [0, 6000]
vm.dirty_ratio	Must be between [1, 100]
vm.dirty_bytes	Must be between [0, 68719476736]
vm.dirty_writeback_centisecs	Must be between [0, 1000]
vm.max_map_count	Must be between [65536, 2147483647]
vm.overcommit_memory	Must be one of {0, 1, 2}. Not supported on machines with less than 15 GB memory.
vm.overcommit_ratio	Must be between [0, 100]
vm.vfs_cache_pressure	Must be between [0, 100]
vm.swappiness	Must be between [0, 200]
vm.watermark_scale_factor	Must be between [10, 3000]
vm.min_free_kbytes	Must be between [67584, 1048576]

KEY	VALUE
enablePtpKvmTimeSync	boolean

KEY	VALUE
hugepage_size2m	Number of 2M huge pages, any positive integer
hugepage_size1g	Number of 1G huge pages, any positive integer

KEY	VALUE
enabled	boolean
encryptionConfig	specify encryption settings for the swap space
bootDiskProfile	specify swap on the node's boot disk
ephemeralLocalSsdProfile	specify swap on the local SSD shared with pod ephemeral storage
dedicatedLocalSsdProfile	specify swap on a new, separate local NVMe SSD exclusively for swap

KEY	VALUE
disabled	boolean

KEY	VALUE
swapSizeGib	integer
swapSizePercent	integer

KEY	VALUE
swapSizeGib	integer
swapSizePercent	integer

KEY	VALUE
diskCount	integer

KEY	VALUE
policy	ENFORCE_SIGNED_MODULES, DO_NOT_ENFORCE_SIGNED_MODULES, POLICY_UNSPECIFIED

• On machines with less than 30 GB memory: 60% of the total memory. For example, on e2-standard-2 machine with 8 GB of memory, you can't allocate more than 4.8 GB for hugepages.
• On machines with more than 30 GB memory: 80% of the total memory. For example, on c4a-standard-8 machines with 32 GB of memory, hugepages cannot exceed 25.6 GB.

1G hugepages are only available in following machine familes: c3, m2, c2d, c3d, h3, m3, a2, a3, g2.

Supported values for 'cgroupMode' under 'linuxConfig'.

• CGROUP_MODE_V1: Use cgroupv1 on the node pool.
• CGROUP_MODE_V2: Use cgroupv2 on the node pool.
• CGROUP_MODE_UNSPECIFIED: Use the default GKE cgroup configuration.

Supported values for 'transparentHugepageEnabled' under 'linuxConfig' which controls transparent hugepage support for anonymous memory.

• TRANSPARENT_HUGEPAGE_ENABLED_ALWAYS: Transparent hugepage is enabled system wide.
• TRANSPARENT_HUGEPAGE_ENABLED_MADVISE: Transparent hugepage is enabled inside MADV_HUGEPAGE regions. This is the default kernel configuration.
• TRANSPARENT_HUGEPAGE_ENABLED_NEVER: Transparent hugepage is disabled.
• TRANSPARENT_HUGEPAGE_ENABLED_UNSPECIFIED: Default value. GKE will not modify the kernel configuration.

Supported values for 'transparentHugepageDefrag' under 'linuxConfig' which defines the transparent hugepage defrag configuration on the node.

• TRANSPARENT_HUGEPAGE_DEFRAG_ALWAYS: It means that an application requesting THP will stall on allocation failure and directly reclaim pages and compact memory in an effort to allocate a THP immediately.
• TRANSPARENT_HUGEPAGE_DEFRAG_DEFER: It means that an application will wake kswapd in the background to reclaim pages and wake kcompactd to compact memory so that THP is available in the near future. It is the responsibility of khugepaged to then install the THP pages later.
• TRANSPARENT_HUGEPAGE_DEFRAG_DEFER_WITH_MADVISE: It means that an application will enter direct reclaim and compaction like always, but only for regions that have used madvise(MADV_HUGEPAGE); all other regions will wake kswapd in the background to reclaim pages and wake kcompactd to compact memory so that THP is available in the near future.
• TRANSPARENT_HUGEPAGE_DEFRAG_MADVISE: It means that an application will enter direct reclaim and compaction like always, but only for regions that have used madvise(MADV_HUGEPAGE); all other regions will wake kswapd in the background to reclaim pages and wake kcompactd to compact memory so that THP is available in the near future.
• TRANSPARENT_HUGEPAGE_DEFRAG_NEVER: It means that an application will never enter direct reclaim or compaction.
• TRANSPARENT_HUGEPAGE_DEFRAG_UNSPECIFIED: Default value. GKE will not modify the kernel configuration.

Supported values for 'policy' under 'nodeKernelModuleLoading'.

• POLICY_UNSPECIFIED: Default behavior. GKE selects the image based on node type. For CPU and TPU nodes, the image will not allow loading external kernel modules. For GPU nodes, the image will allow loading any module, whether it is signed or not.
• ENFORCE_SIGNED_MODULES: Enforced signature verification: Node pools will use a Container-Optimized OS image configured to allow loading of Google-signed external kernel modules. Loadpin is enabled but configured to exclude modules, and kernel module signature checking is enforced.
• DO_NOT_ENFORCE_SIGNED_MODULES: Do not enforce kernel module signature enforcement. Mirrors existing DEFAULT behavior.

Note, updating the system configuration of an existing node pool requires recreation of the nodes which which might cause a disruption.

Examples:

gcloud container node-pools update node-pool-1 --cluster=example-cluster --tags=tag1,tag2

Specifies the number of extra (surge) nodes to be created during this node pool's upgrades. For example, running the following command will result in creating an extra node each time the node pool is upgraded:

gcloud container node-pools update node-pool-1 --cluster=example-cluster --max-surge-upgrade=1   --max-unavailable-upgrade=0

Specifies the number of nodes that can be unavailable at the same time during this node pool's upgrades. For example, assume the node pool has 5 nodes, running the following command will result in having 3 nodes being upgraded in parallel (1 + 2), but keeping always at least 3 (5 - 2) available each time the node pool is upgraded:

gcloud container node-pools update node-pool-1 --cluster=example-cluster --max-surge-upgrade=1   --max-unavailable-upgrade=2

gcloud container node-pools update node-pool-1 --cluster=example-cluster  --node-pool-soak-duration=600s

gcloud container node-pools update node-pool-1 --cluster=example-cluster  --standard-rollout-policy=batch-node-count=3,batch-soak-duration=60s

gcloud container node-pools update node-pool-1 --cluster=example-cluster  --standard-rollout-policy=batch-percent=0.3,batch-soak-duration=60s

gcloud compute machine-types list

gcloud container node-pools update high-mem-pool --machine-type=custom-2-12288

• BALANCED - Is a best effort policy that aims to balance the sizes of available zones.
• ANY - Instructs the cluster autoscaler to prioritize utilization of unused reservations, and reduces preemption risk for Spot VMs.

gcloud container node-pools update node-pool-1 --cluster=example-cluster --enable-autorepair

gcloud container node-pools update node-pool-1 --cluster=example-cluster --enable-autoupgrade

Run $ gcloud help for details.

gcloud alpha container node-pools update

gcloud beta container node-pools update