Fix GKE autopilot reco

deploy-manage/deploy/cloud-on-k8s/deploy-eck-on-gke-autopilot.md

@@ -19,25 +19,57 @@
 1. It is recommended that each Kubernetes host’s virtual memory kernel settings be modified. Refer to [Virtual memory](virtual-memory.md).
 2. It is recommended that {{es}} Pods have an `initContainer` that waits for virtual memory settings to be in place.
 3. For Elastic Agent/Beats there are storage limitations to be considered.
-4. Ensure you are using a node class that is applicable for your workload by adding a `cloud.google.com/compute-class` label in a `nodeSelector`. Refer to [GKE Autopilot documentation.](https://cloud.google.com/kubernetes-engine/docs/concepts/autopilot-compute-classes).
+4. Ensure you are using a node class that is applicable for your workload by adding a `cloud.google.com/compute-class` label in a `nodeSelector`. Refer to [GKE Autopilot documentation](https://cloud.google.com/kubernetes-engine/docs/concepts/autopilot-compute-classes).
 
 ## Ensuring virtual memory kernel settings [k8s-autopilot-setting-virtual-memory]
 
-If you are intending to run production workloads on GKE Autopilot then `vm.max_map_count` should be set. The recommended way to set this kernel setting on the Autopilot hosts is with a `Daemonset` as described in the [Virtual memory](virtual-memory.md) section. You must be running at least version 1.25 when on the `regular` channel or using the `rapid` channel, which currently runs version 1.27.
+If you are intending to run production workloads on GKE Autopilot then `vm.max_map_count` should be set. The recommended way to set this kernel setting on the Autopilot hosts depends on your GKE version:
 
-::::{warning}
-Only use the provided `Daemonset` exactly as specified or it could be rejected by the Autopilot control plane.
-::::
+* **GKE 1.30.3-gke.1451000 or later**: [Use a custom ComputeClass](/deploy-manage/deploy/cloud-on-k8s/virtual-memory.md#k8s_using_a_computeclass_to_set_virtual_memory). Using a custom ComputeClass allows you to set a higher value for `vm.max_map_count`, avoiding the limitations of the `DaemonSet` approach.
+* **Earlier versions**: [Use a DaemonSet](/deploy-manage/deploy/cloud-on-k8s/virtual-memory.md#k8s_using_a_daemonset_to_set_virtual_memory). You must be running at least version 1.25 when on the `regular` channel or using the `rapid` channel, which currently runs version 1.27.
+
+  ::::{warning}
+  Use the provided `Daemonset` exactly as specified, with a `vm.max_map_count` value of `262144`, or it could be rejected by the Autopilot control plane.
+  ::::
 
 ## Install the ECK Operator [k8s-autopilot-deploy-the-operator]
 
 Refer to [*Install ECK*](install.md) for more information on installation options.
 
 ## Deploy an {{es}} cluster [k8s-autopilot-deploy-elasticsearch]
 
-Create an {{es}} cluster. If you are using the `Daemonset` described in the [Virtual memory](virtual-memory.md) section to set `max_map_count` you can add the `initContainer` below is also used to ensure the setting is set prior to starting {{es}}.
+Create an {{es}} cluster. The information that you need to provide in your spec depends on whether you've increased your virtual memory kernel setting, and the method that you used.
+
+::::{tab-set}
+
+:::{tab-item} Using a custom ComputeClass
+If you used a custom ComputeClass to set `vm.max_map_count`, then you need to reference the custom ComputeClass as part of your template spec.
 
-```shell subs=true
+```yaml subs=true
+cat <<EOF | kubectl apply -f -
+apiVersion: elasticsearch.k8s.elastic.co/v1
+kind: Elasticsearch
+metadata:
+  name: elasticsearch-sample
+spec:
+  version: {{version.stack}}
+  nodeSets:
+  - name: default
+    count: 1
+    podTemplate:
+      spec:
+        nodeSelector:
+          cloud.google.com/compute-class: "elasticsearch"
+EOF
+```
+:::
+
+
+:::{tab-item} Using a DaemonSet
+
+If you used a DaemonSet to set `max_map_count`, you can add the following `initContainer` to ensure the setting is set prior to starting {{es}}.
+
+```yaml subs=true
 cat <<EOF | kubectl apply -f -
 apiVersion: elasticsearch.k8s.elastic.co/v1
 kind: Elasticsearch
@@ -48,23 +80,44 @@
   nodeSets:
   - name: default
     count: 1
-    # Only uncomment the below section if you are not using the Daemonset to set max_map_count.
-    # config:
-    #  node.store.allow_mmap: false
     podTemplate:
       spec:
         # This init container ensures that the `max_map_count` setting has been applied before starting Elasticsearch.
-        # This is not required, but is encouraged when using the previously mentioned Daemonset to set max_map_count.
+        # This is not required, but is encouraged when using the Daemonset to set max_map_count.
         # Do not use this if setting config.node.store.allow_mmap: false
         initContainers:
         - name: max-map-count-check
-          command: ['sh', '-c', "while true; do mmc=$(cat /proc/sys/vm/max_map_count); if [ ${mmc} -eq 262144 ]; then exit 0; fi; sleep 1; done"]
+          command: ['sh', '-c', "while true; do mmc=$(cat /proc/sys/vm/max_map_count); if [ ${mmc} -eq 262144 ]; then exit 0; fi; sleep 1; done"]     
+EOF
+```
+:::
+::::
+
+### Deploy without custom virtual memory
+
+If you didn't increase your virtual memory, then you need to set `node.store.allow_mmap` to `false`.
+
+```yaml subs=true
+cat <<EOF | kubectl apply -f -
+apiVersion: elasticsearch.k8s.elastic.co/v1
+kind: Elasticsearch
+metadata:
+  name: elasticsearch-sample
+spec:
+  version: {{version.stack}}
+  nodeSets:
+  - name: default
+    count: 1 
+    config:
+      node.store.allow_mmap: false
 EOF
 ```
+:::
+::::
 
 ## Deploy a standalone Elastic Agent and/or Beats [k8s-autopilot-deploy-agent-beats]
 
-When running Elastic Agent and Beats within GKE Autopilot there are storage constraints to be considered. No `HostPath` volumes are allowed, which the ECK operator defaults to when unset for both `Deployments` and `Daemonsets`. Instead use [Kubernetes ephemeral volumes](https://kubernetes.io/docs/concepts/storage/ephemeral-volumes).
+When running Elastic Agent and Beats within GKE Autopilot there are storage constraints to be considered. No `HostPath` volumes are allowed, which the ECK operator defaults to when unset for both `Deployments` and `DaemonSets`. Instead use [Kubernetes ephemeral volumes](https://kubernetes.io/docs/concepts/storage/ephemeral-volumes).
 
 Refer to [Recipes to deploy {{es}}, {{kib}}, Elastic Fleet Server and Elastic Agent and/or Beats within GKE Autopilot](https://github.com/elastic/cloud-on-k8s/tree/main/config/recipes/autopilot).
 

deploy-manage/deploy/cloud-on-k8s/virtual-memory.md

@@ -12,10 +12,14 @@
 
 By default, {{es}} uses memory mapping (`mmap`) to efficiently access indices. Default values for virtual address space on Linux distributions can be too low for {{es}} to work properly, which may result in out-of-memory exceptions. This is why [the quickstart example](/deploy-manage/deploy/cloud-on-k8s/elasticsearch-deployment-quickstart.md) disables `mmap` through the `node.store.allow_mmap: false` setting. For production workloads, we recommended you increase the kernel setting `vm.max_map_count` to `1048576` and leave `node.store.allow_mmap` unset.
 
-The kernel setting `vm.max_map_count=1048576` can be set on the host directly, by a dedicated init container which must be privileged, or a dedicated Daemonset.
+The kernel setting `vm.max_map_count=1048576` can be set on the host directly, by a dedicated init container which must be privileged, a dedicated Daemonset, or a custom ComputeClass.
 
 :::{important}
-For {{es}} version 8.16 and later, set the `vm.max_map_count` kernel setting to `1048576`; for {{es}} version 8.15 and earlier, set `vm.max_map_count` to `262144`.
+For {{es}} version 8.16 and later, set the `vm.max_map_count` kernel setting to `1048576`; for {{es}} version 8.15 and earlier, set `vm.max_map_count` to `262144`. 
+
+The exception is in GKE Autopilot environments. Your options depend on your GKE version:
+* **GKE 1.30.3-gke.1451000 or later**: Use a custom `ComputeClass`, rather than a `DaemonSet`, to override the kernel setting.
+* **Earlier versions**: `vm.max_map_count` must be set to `262144`.
 :::
 
 For more information, check the {{es}} documentation on [Virtual memory](/deploy-manage/deploy/self-managed/vm-max-map-count.md).
@@ -91,13 +95,14 @@
           securityContext:
             privileged: true
             runAsUser: 0
-          command: ['/usr/local/bin/bash', '-e', '-c', 'echo 262144 > /proc/sys/vm/max_map_count']
+          command: ['/usr/local/bin/bash', '-e', '-c', 'echo 1048576 > /proc/sys/vm/max_map_count'] <1>
       containers:
         - name: sleep
           image: docker.io/bash:5.2.21
           command: ['sleep', 'infinity']
 EOF
 ```
+1. In GKE Autopilot environments, `vm.max_map_count` must be set to 262144 when using a DaemonSet.
 
 To run an {{es}} instance that waits for the kernel setting to be in place:
 
@@ -122,8 +127,60 @@
         # Do not use this if setting config.node.store.allow_mmap: false
         initContainers:
         - name: max-map-count-check
-          command: ['sh', '-c', "while true; do mmc=$(cat /proc/sys/vm/max_map_count); if [ ${mmc} -eq 262144 ]; then exit 0; fi; sleep 1; done"]
+          command: ['sh', '-c', "while true; do mmc=$(cat /proc/sys/vm/max_map_count); if [ ${mmc} -eq 262144 ]; then exit 0; fi; sleep 1; done"] <1>
 EOF
 ```
+1. In GKE Autopilot environments, `vm.max_map_count` must be set to 262144 when using a DaemonSet.
+
+
+## Using a custom ComputeClass to set virtual memory [k8s_using_a_computeclass_to_set_virtual_memory]
+```{applies_to}
+deployment:
+  eck: ga 3.2+
+```
+
+If you're using GKE to run ECK, then you can use a [custom ComputeClass](https://docs.cloud.google.com/kubernetes-engine/docs/concepts/about-custom-compute-classes), rather than a DaemonSet, to increase the `vm.max_map_count` setting. On [GKE Autopilot](/deploy-manage/deploy/cloud-on-k8s/deploy-eck-on-gke-autopilot.md) this allows you to set a higher value, which is not possible with a DaemonSet.
 
+1. Create a ComputeClass that changes the host kernel setting on all nodes:
 
+    ```yaml
+    cat <<EOF | kubectl apply -f -
+    apiVersion: cloud.google.com/v1
+    kind: ComputeClass
+    metadata:
+      name: elasticsearch
+    spec:
+      whenUnsatisfiable: "DoNotScaleUp" <1>
+      nodePoolAutoCreation:
+        enabled: true
+      priorityDefaults: <2>
+        nodeSystemConfig:
+          linuxNodeConfig:
+            sysctls:
+              vm.max_map_count: 1048576
+      priorities:
+        - machineFamily: n2
+    EOF
+    ```
+    1. Default since GKE 1.33
+    2. `priorityDefaults` is available only since GKE 1.32.1-gke.1729000
+
+2. Create your {{es}} instance using the custom ComputeClass:
+
+    ```yaml subs=true
+    cat <<'EOF' | kubectl apply -f -
+    apiVersion: elasticsearch.k8s.elastic.co/v1
+    kind: Elasticsearch
+    metadata:
+      name: elasticsearch
+    spec:
+      version: {{version.stack}}
+      nodeSets:
+      - name: default
+        count: 1
+        podTemplate:
+          spec:
+            nodeSelector:
+              cloud.google.com/compute-class: "elasticsearch"
+    EOF
+    ```