### Run Kimchi Setup Wizard

Source: https://docs.cast.ai/docs/ai-enabler-serverless-kimchi-cli

Launch the interactive setup wizard for Kimchi if you installed it manually.

```shell
kimchi
```

--------------------------------

### Install Kimchi with One-Line Installer

Source: https://docs.cast.ai/docs/ai-enabler-serverless-kimchi-cli

Use this command to quickly install Kimchi and automatically launch the setup wizard.

```shell
curl -fsSL https://github.com/castai/kimchi/releases/latest/download/install.sh | bash
```

--------------------------------

### Install OpenCode

Source: https://docs.cast.ai/docs/ai-enabler-tutorial-serverless-opencode

Install OpenCode using curl. This command fetches the installation script and executes it.

```bash
curl -fsSL https://opencode.ai/install | bash
```

--------------------------------

### Install GSD Framework

Source: https://docs.cast.ai/docs/ai-enabler-tutorial-serverless-opencode

Install the Get Shit Done (GSD) framework using npm. This command is used to set up the necessary tools for multi-agent orchestration.

```bash
npx gsd-opencode
```

--------------------------------

### Install castctl using a script

Source: https://docs.cast.ai/docs/connect-with-castctl

Install the castctl CLI on macOS or Linux using a provided installation script. The script defaults to installing in `~/.local/bin`.

```shell
curl -fsSL https://get.cast.ai/linux | bash
```

--------------------------------

### Spot Configuration Examples

Source: https://docs.cast.ai/docs/pod-mutations-reference

Examples of configuring Spot instance behavior with different modes and distribution percentages.

```yaml
# 80% Spot with fallback to On-Demand
spotConfig:
  mode: preferred-spot
  distributionPercentage: 80
```

```yaml
# 100% Spot, strict (no fallback)
spotConfig:
  mode: only-spot
  distributionPercentage: 100
```

```yaml
# 50/50 split, no preference when both available
spotConfig:
  mode: optional-spot
  distributionPercentage: 50
```

```yaml
# No Spot configuration (empty object)
spotConfig: {}
```

--------------------------------

### Overprovisioning Example Calculation

Source: https://docs.cast.ai/docs/cluster-efficiency-report

An example demonstrating the overprovisioning percentage calculation with sample CPU values.

```text
100% - (490 ÷ 688 × 100%) = 100% - 71.22% = 28.78%
```

--------------------------------

### Install castctl on Windows

Source: https://docs.cast.ai/docs/connect-with-castctl

Install the castctl CLI on Windows using PowerShell. This method downloads and executes the installer.

```powershell
iwr -useb https://get.cast.ai/windows | iex
```

--------------------------------

### Install kubectl using Homebrew

Source: https://docs.cast.ai/docs/omni-getting-started

Installs the Kubernetes CLI tool using the Homebrew package manager. Ensure you have Homebrew installed and configured.

```bash
brew install kubectl
```

--------------------------------

### Install Google Cloud CLI using Homebrew

Source: https://docs.cast.ai/docs/omni-getting-started

Installs the Google Cloud SDK using Homebrew, necessary for interacting with Google Cloud Platform edge locations. Homebrew must be installed.

```bash
brew install --cask google-cloud-sdk
```

--------------------------------

### Install egressd Demo

Source: https://docs.cast.ai/docs/egressd

Install a demo version of egressd with preconfigured Grafana and Prometheus metrics by running a script.

```shell
curl -fsSL <https://raw.githubusercontent.com/castai/egressd/master/demo.sh> | bash
```

--------------------------------

### Install castctl to a specific location

Source: https://docs.cast.ai/docs/connect-with-castctl

Install the castctl CLI to a specified directory, such as `/usr/local/bin`, using the installation script.

```shell
curl -fsSL https://get.cast.ai/linux \
  | bash
```

--------------------------------

### Install Live Migration Controller

Source: https://docs.cast.ai/docs/helm-charts

Installs the castai-live Helm chart for the live migration controller. Requires API key and cluster ID.

```shell
helm upgrade -i castai-live castai-helm/castai-live -n castai-agent \
  --set castai.apiKey="<your-api-key>" \
  --set castai.clusterID="<your-cluster-id>"
```

--------------------------------

### Install Prometheus using Helm

Source: https://docs.cast.ai/docs/observability-metrics-integrating-prometheus-metrics-with-new-relic

Installs Prometheus in the 'monitoring' namespace using Helm. Ensure the namespace is created if it doesn't exist.

```shell
helm install prometheus prometheus-community/prometheus \
  --namespace monitoring \
  --create-namespace
```

--------------------------------

### Install egressd using Helm

Source: https://docs.cast.ai/docs/egressd

Install egressd using the Cast AI Helm chart with a values.yaml file.

```shell
helm upgrade -i castai-egressd castai-helm/egressd -n castai-agent -f values.yaml
```

--------------------------------

### Install Kentroller for Karpenter

Source: https://docs.cast.ai/docs/helm-charts

Installs the castai-kentroller Helm chart to coordinate Cast AI features with Karpenter. Requires API key and cluster ID.

```shell
helm upgrade -i castai-kentroller castai-helm/castai-kentroller -n castai-agent \
  --set castai.apiKey="<your-api-key>" \
  --set castai.clusterID="<your-cluster-id>"
```

--------------------------------

### Example Evictor Configuration Labels (JSON)

Source: https://docs.cast.ai/docs/evictor

This JSON object represents the same configuration labels as the YAML example, useful for programmatic access or validation.

```json
{
  "labels": {
    "app.kubernetes.io/instance": "castai-evictor",
    "app.kubernetes.io/managed-by": "Helm",
    "app.kubernetes.io/name": "castai-evictor",
    "app.kubernetes.io/version": "v1",
    "charts.cast.ai/managed": "true",
    "helm.sh/chart": "castai-evictor-0.31.20"
  }
}
```

--------------------------------

### cURL GET Request Example

Source: https://docs.cast.ai/docs/authentication

A basic cURL command to make a GET request to a specified URL. This is a foundational example for interacting with the API.

```bash
curl --request GET \
     --url https://example.com/
```

--------------------------------

### Start OpenCode and Verify Configuration

Source: https://docs.cast.ai/docs/ai-enabler-tutorial-serverless-opencode

Navigates to a project directory and starts OpenCode to verify that the configuration is working correctly. A simple prompt is used to test the connection and model routing.

```bash
cd ~/your-project
opencode
```

--------------------------------

### RBAC Permissions for Terraform Installation

Source: https://docs.cast.ai/docs/castai-operator

This YAML defines the necessary RBAC permissions for a pre-installation job required by the castware-components chart during Terraform installations. It grants the job permissions to get, list, and watch deployments and pods.

```yaml
apiGroups: ["apps"]
resources: ["deployments"]
verbs: ["get", "list"]

apiGroups: [""]
resources: ["pods"]
verbs: ["get", "list", "watch"]
```

--------------------------------

### Namespace Filter Example (Regex Match)

Source: https://docs.cast.ai/docs/pod-mutations-reference

Filters mutations to apply to pods in namespaces matching specific regex patterns, such as those starting with 'team-' or 'env-prod-'.

```yaml
filterV2:
  workload:
    namespaces:
      - type: regex
        value: "^team-.*$"
      - type: regex
        value: "^env-prod-.*$"
```

--------------------------------

### Making an API Call with curl

Source: https://docs.cast.ai/docs/api-access

Example of how to authenticate and make a GET request to the Cast AI API using curl. Ensure you replace 'your-api-key-here' with your actual API key.

```shell
curl -X GET "https://api.cast.ai/v1/kubernetes/external-clusters" -H "X-API-Key: your-api-key-here" | jq
```

--------------------------------

### Understanding Resource Cost Distribution Example

Source: https://docs.cast.ai/docs/cluster-efficiency-report

Illustrates the relationship between total cost and different cost metrics (per provisioned, requested, and used resources) with a simplified example.

```text
(Cost per Provisioned × Provisioned Resources) = (Cost per Requested × Requested Resources) = (Cost per Used × Used Resources) = Total Cost
```

```text
Provisioned Resources > Requested Resources > Used Resources
```

```text
Cost per Provisioned < Cost per Requested < Cost per Used
```

```text
Cost per Provisioned CPU = $100 ÷ 10 = $10
Cost per Requested CPU = $100 ÷ 8 = $12.50
Cost per Used CPU = $100 ÷ 4 = $25
```

--------------------------------

### Full Deployment with Network Bandwidth Configuration

Source: https://docs.cast.ai/docs/network-bandwidth

This example demonstrates how to integrate network bandwidth requests and limits into a complete Kubernetes Deployment manifest. Ensure the `castai-omni-agent` is v1.12.5 or higher.

```yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx
  name: nginx
spec:
  replicas: 1
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
        - image: nginx
          name: nginx
          resources:
            requests:
              scheduling.cast.ai/network-bandwidth: "750"
            limits:
              scheduling.cast.ai/network-bandwidth: "750"
```

--------------------------------

### EKS MIG Partition Size Configuration Examples

Source: https://docs.cast.ai/docs/gpu-sharing-mig

Examples of configuring different MIG partition sizes on EKS by changing the node selector key.

```yaml
nodeSelector:
  nvidia.com/gpu.mig-partition-1g.5gb: "true"
```

```yaml
nodeSelector:
  nvidia.com/gpu.mig-partition-2g.10gb: "true"
```

```yaml
nodeSelector:
  nvidia.com/gpu.mig-partition-3g.20gb: "true"
```

--------------------------------

### Example OMNI pod status output

Source: https://docs.cast.ai/docs/omni-getting-started

This is an example of the expected output when verifying OMNI onboarding. It shows the status of various OMNI and liqo components running in the cluster.

```shell
NAME                                       READY   STATUS    RESTARTS   AGE
liqo-controller-manager-7cf59bcc64-xxxxx   1/1     Running   0          2m
liqo-crd-replicator-687bdc6f66-xxxxx       1/1     Running   0          2m
liqo-fabric-xxxxx                          1/1     Running   0          2m
liqo-ipam-8667dbccbb-xxxxx                 1/1     Running   0          2m
liqo-metric-agent-55cd8748c5-xxxxx         1/1     Running   0          2m
liqo-proxy-77c66dfb88-xxxxx                1/1     Running   0          2m
liqo-webhook-6f648484cc-xxxxx              1/1     Running   0          2m
omni-agent-595c4b97d9-xxxxx                1/1     Running   0          2m
```

--------------------------------

### Install jq on macOS

Source: https://docs.cast.ai/docs/cast-ai-anywhere-getting-started

Installs the jq command-line JSON processor required by the installation script on macOS systems.

```shell
brew install jq
```

--------------------------------

### Create Scaling Policy

Source: https://docs.cast.ai/docs/woop-scaling-policies-create

This example demonstrates how to create a scaling policy with immediate apply type, managed policy, CPU/memory targets, and specific workload and namespace assignment rules.

```APIDOC
## Create Scaling Policy

### Description
Creates a new scaling policy with specified optimization settings and assignment rules.

### Method
POST (Assumed based on creation context)

### Endpoint
/websites/cast_ai/scaling-policies (Assumed based on project path)

### Parameters
#### Request Body
- **name** (string) - Required - The name of the scaling policy.
- **applyType** (string) - Required - The type of scaling application (e.g., IMMEDIATE).
- **managementPolicy** (string) - Required - The management policy type (e.g., MANAGED).
- **cpu** (object) - Optional - CPU optimization settings.
  - **target** (string) - Required - The CPU target (e.g., p95, max).
  - **overhead** (number) - Optional - CPU overhead percentage.
- **memory** (object) - Optional - Memory optimization settings.
  - **target** (string) - Required - The memory target (e.g., max).
  - **overhead** (number) - Optional - Memory overhead percentage.
- **assignmentRules** (array) - Required - A list of rules to determine policy assignment.
  - Each rule can contain a `workload` matcher, a `namespace` matcher, or both.
  - Multiple matchers within a single rule are combined with AND logic.
  - Multiple rules within a policy are combined with OR logic.
  
  **Workload Matcher:**
  - **gvk** (array of strings) - Required - Kubernetes workload types to match (e.g., ["Deployment", "StatefulSet"]).
  - **labelsExpressions** (array of objects) - Optional - Label-based matching conditions.
    - **key** (string) - Required - The label key.
    - **operator** (string) - Required - The label operator (e.g., KUBERNETES_LABEL_SELECTOR_OP_IN).
    - **values** (array of strings) - Required if operator requires values - The list of values for the operator.

  **Namespace Matcher:**
  - **names** (array of strings) - Required - Specific namespace names to match (e.g., ["production", "prod-frontend"]).

### Request Example
```json
{
  "name": "production-frontend-policy",
  "applyType": "IMMEDIATE",
  "managementPolicy": "MANAGED",
  "cpu": {
    "target": "p95",
    "overhead": 0.05
  },
  "memory": {
    "target": "max",
    "overhead": 0.10
  },
  "assignmentRules": [
    {
      "workload": {
        "gvk": ["Deployment", "StatefulSet"],
        "labelsExpressions": [
          {
            "key": "tier",
            "operator": "KUBERNETES_LABEL_SELECTOR_OP_IN",
            "values": ["frontend", "web"]
          }
        ]
      }
    },
    {
      "namespace": {
        "names": ["production", "prod-frontend"]
      }
    }
  ]
}
```

### Response
#### Success Response (200)
(Response structure not provided in source text)

#### Response Example
(Response example not provided in source text)
```

--------------------------------

### Install jq on Debian/Ubuntu

Source: https://docs.cast.ai/docs/cast-ai-anywhere-getting-started

Installs the jq command-line JSON processor required by the installation script on Debian/Ubuntu systems.

```shell
apt-get install jq
```

--------------------------------

### Install Cline CLI

Source: https://docs.cast.ai/docs/ai-enabler-tutorial-serverless-cline

Install the Cline CLI globally using npm. Ensure you have Node.js 20 or higher installed.

```shell
npm install -g cline
```

--------------------------------

### Basic API Policy Creation

Source: https://docs.cast.ai/docs/woop-scaling-policies-create

This example demonstrates the creation of a basic scaling policy via API, including immediate scaling, CPU/memory targets, and assignment rules based on workload labels and namespaces.

```json
{
  "name": "production-frontend-policy",
  "applyType": "IMMEDIATE",
  "managementPolicy": "MANAGED",
  "cpu": {
    "target": "p95",
    "overhead": 0.05
  },
  "memory": {
    "target": "max",
    "overhead": 0.10
  },
  "assignmentRules": [
    {
      "workload": {
        "gvk": ["Deployment", "StatefulSet"],
        "labelsExpressions": [
          {
            "key": "tier",
            "operator": "KUBERNETES_LABEL_SELECTOR_OP_IN",
            "values": ["frontend", "web"]
          }
        ]
      }
    },
    {
      "namespace": {
        "names": ["production", "prod-frontend"]
      }
    }
  ]
}
```

--------------------------------

### Cloud SQL Proxy Configuration Example

Source: https://docs.cast.ai/docs/performance-advisor-installing-agent

Example values.yaml configuration for enabling Cloud SQL Proxy. Set the database host to the instance connection name and configure the cloudSqlProxy section.

```yaml
database:
  host: <Connection name>
  
cloudSqlProxy:
  enabled: true
  port: 10000
  privateIp: true
  autoIamAuthn: true
collectors:
  excludedDatabases: "cloudsqladmin"
```

--------------------------------

### Install NVIDIA DRA Driver on GKE

Source: https://docs.cast.ai/docs/dra

This script installs GPU drivers, creates the 'nvidia' namespace, applies a ResourceQuota, and installs the NVIDIA DRA driver using Helm. Ensure you have kubectl and Helm installed.

```shell
#!/bin/bash

# Install GPU drivers (GKE documentation: https://cloud.google.com/kubernetes-engine/docs/how-to/gpus#installing_drivers)
kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/container-engine-accelerators/master/nvidia-driver-installer/cos/daemonset-preloaded-latest.yaml

# Install NVIDIA DRA driver
kubectl create namespace nvidia

kubectl apply -n nvidia -f - << EOF
apiVersion: v1
kind: ResourceQuota
metadata:
  name: nvidia
spec:
  hard:
    pods: 100
  scopeSelector:
    matchExpressions:
    - operator: In
      scopeName: PriorityClass
      values:
      - system-node-critical
      - system-cluster-critical
EOF

cat <<EOF > values.yaml
resources.gpus.enabled: "true"
gpuResourcesEnabledOverride: "true"
nvidiaDriverRoot: /home/kubernetes/bin/nvidia/

controller:
  affinity: null

kubeletPlugin:
  tolerations:
  - key: nvidia.com/gpu
    operator: Exists
    effect: NoSchedule
  nodeSelector:
    nvidia.com/gpu.present: "true"
  affinity:
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
        - matchExpressions:
          - key: "nvidia.com/gpu.present"
            operator: "Exists"
          - key: nvidia.com/gpu.dra
            operator: In
            values:
            - "true"
EOF

helm upgrade -i nvidia-dra-driver-gpu \
  oci://registry.k8s.io/dra-driver-nvidia/charts/dra-driver-nvidia-gpu \
  --namespace nvidia \
  -f values.yaml
```

--------------------------------

### Recommendation Object with Startup Phase

Source: https://docs.cast.ai/docs/woop-startup-recommendations

This YAML defines a Recommendation object that includes both the startup phase (original CPU requests) and the post-startup phase (optimized CPU requests) for a container. The startup duration is also specified.

```yaml
apiVersion: autoscaling.cast.ai/v2
kind: Recommendation
metadata:
  name: nginx-deployment
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: nginx
  ...

  # startup phase: original workload CPU requests
  recommendation:
    - containerName: nginx
      requests:
        cpu: 300m
        memory: 300Mi

  startup:
    duration: "5m"

  postStartup:
    verticalHash: abc1e2323asda
    recommendation:
      - containerName: nginx
        requests:
          cpu: 150m
          memory: 150Mi
```

--------------------------------

### Install jq on Red Hat/CentOS

Source: https://docs.cast.ai/docs/cast-ai-anywhere-getting-started

Installs the jq command-line JSON processor required by the installation script on Red Hat/CentOS systems.

```shell
yum install jq
```

--------------------------------

### View Downloaded Init Script

Source: https://docs.cast.ai/docs/omni-custom-edge-location

Inspect the init script that was downloaded and executed by the agent.

```shell
cat /var/lib/castai-edge-initd/init_script

```

--------------------------------

### Install Cast AI Evictor

Source: https://docs.cast.ai/docs/helm-charts

Install the Evictor component to reduce cluster size. Set dryRun to false for actual eviction. This command is for basic installation.

```shell
helm upgrade -i castai-evictor castai-helm/castai-evictor -n castai-agent --set dryRun=false
```

--------------------------------

### Install AWS CLI using Homebrew

Source: https://docs.cast.ai/docs/omni-getting-started

Installs the AWS Command Line Interface using Homebrew, required for managing AWS resources in edge locations. Ensure you have Homebrew installed.

```bash
brew install awscli
```

--------------------------------

### Install Hibernate Component with Helm

Source: https://docs.cast.ai/docs/pausing-a-cluster

Install the Cast AI hibernate component using Helm. Ensure you have Helm installed and configured. Update the 'cloud' and 'apiKey' variables with your specific values.

```shell
helm upgrade -i castai-hibernate castai-helm/castai-hibernate -n castai-agent --set cloud=<AKS|EKS|GKE> --set apiKey=<CASTAI-API-KEY-REPLACE-ME-WITH-BASE64_ENCODE>
```

--------------------------------

### Complete Kubernetes Deployment for GPU Time-Slicing Workload

Source: https://docs.cast.ai/docs/gpu-sharing-time-slicing

A full Kubernetes Deployment example demonstrating how to schedule multiple pods on a single GPU using time-slicing. It includes node selectors and tolerations to target GPU-sharing enabled nodes.

```yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: gpu-timeslicing-workload
spec:
  replicas: 10  # Can schedule 10 pods on a single GPU with 10x sharing
  selector:
    matchLabels:
      app: gpu-timeslicing-workload
  template:
    metadata:
      labels:
        app: gpu-timeslicing-workload
    spec:
      nodeSelector:
        scheduling.cast.ai/node-template: "gpu-sharing-template"
      tolerations:
        - key: "scheduling.cast.ai/node-template"
          operator: "Exists"
          effect: "NoSchedule"
      containers:
      - name: gpu-workload
        image: your-gpu-image
        resources:
          limits:
            nvidia.com/gpu: 1
```

--------------------------------

### Install OCI CLI using curl script

Source: https://docs.cast.ai/docs/omni-getting-started

Installs the Oracle Cloud Infrastructure Command Line Interface by downloading and executing an installation script from a GitHub repository. This is for OCI edge locations.

```bash
bash -c "$(curl -L https://raw.githubusercontent.com/oracle/oci-cli/master/scripts/install/install.sh)"
```

--------------------------------

### Complete filterV2 Example

Source: https://docs.cast.ai/docs/pod-mutations-reference

This example demonstrates a comprehensive filterV2 configuration, specifying workload and pod matching criteria including namespaces, names, kinds, labels, and exclusions.

```yaml
filterV2:
  workload:
    namespaces:
      - type: exact
        value: production
    names:
      - type: regex
        value: "^frontend-.*$"
    kinds:
      - type: exact
        value: Deployment
      - type: exact
        value: StatefulSet
    excludeNames:
      - type: exact
        value: frontend-canary
  pod:
    labels:
      matchers:
        - keyMatcher:
            type: exact
            value: tier
          valueMatcher:
            type: exact
            value: web
      operator: and
    excludeLabels:
      matchers:
        - keyMatcher:
            type: exact
            value: skip-mutation
          valueMatcher:
            type: exact
            value: "true"
      operator: or
```

--------------------------------

### Make Kimchi Binary Executable

Source: https://docs.cast.ai/docs/ai-enabler-serverless-kimchi-cli

Set execute permissions for the downloaded Kimchi binary.

```shell
chmod +x kimchi
```

--------------------------------

### Install Exporter with Helm

Source: https://docs.cast.ai/docs/woop-custom-metrics

Install the castai-workload-autoscaler-exporter using Helm if it was not included during cluster onboarding.

```shell
helm repo add castai-helm https://castai.github.io/helm-charts
helm repo update
helm install castai-workload-autoscaler-exporter castai-helm/castai-workload-autoscaler-exporter -n castai-agent
```

--------------------------------

### Create OpenCode Configuration File

Source: https://docs.cast.ai/docs/ai-enabler-tutorial-serverless-opencode

Creates an empty JSON file to store the OpenCode configuration. This file will be populated with provider details.

```bash
touch ~/.config/opencode/opencode.json
```

--------------------------------

### Manual Evictor Installation

Source: https://docs.cast.ai/docs/evictor

Helm command for manual installation of the Cast AI Evictor.

```shell
helm repo add castai-helm https://castai.github.io/helm-charts
helm upgrade --install castai-evictor castai-helm/castai-evictor -n castai-agent --set dryRun=false
```

--------------------------------

### Install GPU Metrics Exporter Helm Chart

Source: https://docs.cast.ai/docs/gpu-metrics-exporter

Installs the gpu-metrics-exporter Helm chart using the specified namespace and values file. Ensure 'values.yaml' is correctly configured with cluster details.

```shell
helm install --generate-name castai-helm/gpu-metrics-exporter -n castai-agent -f values.yaml
```

--------------------------------

### Verify Kimchi Installation

Source: https://docs.cast.ai/docs/ai-enabler-serverless-kimchi-cli

Check if Kimchi is installed correctly by running the version command.

```shell
kimchi version
```

--------------------------------

### ipDetails JSON Example

Source: https://docs.cast.ai/docs/setup-notification-webhook

This is an example JSON object representing IP address details.

```json
{
  "ip_address": "118.25.6.39",
  "ip_version": 4,
  "country_code": "CN",
  "isp": "Tencent Cloud Computing (Beijing) Co. Ltd",
  "domain": "tencent.com",
  "hostnames": [],
  "is_tor": false
}
```

--------------------------------

### Evictor Installation with Aggressive Mode

Source: https://docs.cast.ai/docs/evictor

Helm command to install Evictor with aggressive mode enabled.

```shell
helm upgrade --install castai-evictor castai-helm/castai-evictor -n castai-agent \
  --set dryRun=false,aggressiveMode=true
```

--------------------------------

### GKE MIG Partition Size Configuration Examples

Source: https://docs.cast.ai/docs/gpu-sharing-mig

Examples of configuring different MIG partition sizes on GKE by changing the node selector value.

```yaml
nodeSelector:
  cloud.google.com/gke-gpu-partition-size: 1g.5gb
```

```yaml
nodeSelector:
  cloud.google.com/gke-gpu-partition-size: 2g.10gb
```

```yaml
nodeSelector:
  cloud.google.com/gke-gpu-partition-size: 3g.20gb
```

--------------------------------

### Verify castctl installation

Source: https://docs.cast.ai/docs/connect-with-castctl

Check if the castctl CLI is installed correctly by running the version command.

```shell
castctl version
```

--------------------------------

### Verify Cline Installation

Source: https://docs.cast.ai/docs/ai-enabler-tutorial-serverless-cline

Check if the Cline CLI is installed correctly by running the version command.

```shell
cline version
```

--------------------------------

### Example DCGM Metrics for Fractional GPUs

Source: https://docs.cast.ai/docs/gpu-sharing-fractional

This snippet shows example output from DCGM metrics, illustrating the limited metrics available for fractional GPUs compared to full GPUs. These metrics are useful for basic monitoring.

```shell
# HELP DCGM_FI_DEV_MEMORY_TEMP Memory temperature for the device.
# TYPE DCGM_FI_DEV_MEMORY_TEMP gauge
DCGM_FI_DEV_MEMORY_TEMP{gpu="0",UUID="GPU-e5de654e-ae57-11f0-92fa-d4c3671e2bf4",pci_bus_id="00000000:31:00.0",device="nvidia0",modelName="NVIDIA L4-3Q",Hostname="ip-192-168-119-184.eu-central-1.compute.internal",container="gpu-test",namespace="default",pod="gpu-test-685ff6d954-5jdfh"} 0
# HELP DCGM_FI_DEV_FB_TOTAL Total Frame Buffer of the GPU in MB.
# TYPE DCGM_FI_DEV_FB_TOTAL gauge
DCGM_FI_DEV_FB_TOTAL{gpu="0",UUID="GPU-e5de654e-ae57-11f0-92fa-d4c3671e2bf4",pci_bus_id="00000000:31:00.0",device="nvidia0",modelName="NVIDIA L4-3Q",Hostname="ip-192-168-119-184.eu-central-1.compute.internal",container="gpu-test",namespace="default",pod="gpu-test-685ff6d954-5jdfh"} 3072
# HELP DCGM_FI_DEV_FB_FREE Framebuffer memory free (in MiB).
# TYPE DCGM_FI_DEV_FB_FREE gauge
DCGM_FI_DEV_FB_FREE{gpu="0",UUID="GPU-e5de654e-ae57-11f0-92fa-d4c3671e2bf4",pci_bus_id="00000000:31:00.0",device="nvidia0",modelName="NVIDIA L4-3Q",Hostname="ip-192-168-119-184.eu-central-1.compute.internal",container="gpu-test",namespace="default",pod="gpu-test-685ff6d954-5jdfh"} 2418
# HELP DCGM_FI_DEV_FB_USED Framebuffer memory used (in MiB).
# TYPE DCGM_FI_DEV_FB_USED gauge
DCGM_FI_DEV_FB_USED{gpu="0",UUID="GPU-e5de654e-ae57-11f0-92fa-d4c3671e2bf4",pci_bus_id="00000000:31:00.0",device="nvidia0",modelName="NVIDIA L4-3Q",Hostname="ip-192-168-119-184.eu-central-1.compute.internal",container="gpu-test",namespace="default",pod="gpu-test-685ff6d954-5jdfh"} 0
```

--------------------------------

### Combined Init Script and NodeConfig

Source: https://docs.cast.ai/docs/node-configuration

Pass both a `nodeadm` NodeConfig and a shell script together using MIME multipart. This example configures Kubelet eviction settings and runs a custom shell script.

```yaml
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="BOUNDARY"

--BOUNDARY
Content-Type: application/node.eks.aws

---
apiVersion: node.eks.aws/v1alpha1
kind: NodeConfig
spec:
  kubelet:
    config:
      evictionMaxPodGracePeriod: 30
--BOUNDARY
Content-Type: text/x-shellscript; charset="us-ascii"

#!/bin/bash

echo "a script has run during node provisioning" > /var/log/node-configuration-init-script-out.txt

--BOUNDARY--

```

--------------------------------

### Start Claude Code in a Project Directory

Source: https://docs.cast.ai/docs/ai-enabler-tutorial-serverless-claude-code

Navigate to your project directory and start Claude Code to verify the configuration. This command initiates the AI-assisted coding environment.

```shell
cd ~/your-project
claude
```

--------------------------------

### Evictor Installation with Combined Modes

Source: https://docs.cast.ai/docs/evictor

Helm command to install Evictor with both aggressive and scoped modes enabled.

```shell
helm upgrade --install castai-evictor castai-helm/castai-evictor -n castai-agent \
  --set dryRun=false,aggressiveMode=true,scopedMode=true
```

--------------------------------

### Install Workload Autoscaler

Source: https://docs.cast.ai/docs/helm-charts

Installs the castai-workload-autoscaler Helm chart. Requires API key and cluster ID.

```shell
helm upgrade -i castai-workload-autoscaler castai-helm/castai-workload-autoscaler -n castai-agent \
  --set castai.apiKey="<your-api-key>" \
  --set castai.clusterID="<your-cluster-id>"
```