Hiver

Deploy on Azure

Run Hive on your own Azure infrastructure using Azure Kubernetes Service (AKS).

Prerequisites

  • An Azure subscription
  • az CLI installed and authenticated
  • kubectl installed

Create an AKS cluster

az group create --name hive-rg --location eastus

az aks create \
  --resource-group hive-rg \
  --name hive-cluster \
  --node-count 3 \
  --node-vm-size Standard_D4s_v3 \
  --generate-ssh-keys

az aks get-credentials --resource-group hive-rg --name hive-cluster

Deploy the Hive controller

kubectl apply -f https://install.hive.run/operator/azure.yaml

Create a service principal and store its credentials:

az ad sp create-for-rbac --name hive-sp --role Contributor \
  --scopes /subscriptions/$SUBSCRIPTION_ID > sp.json

kubectl create secret generic hive-azure-credentials \
  --from-literal=client-id=$(jq -r .appId sp.json) \
  --from-literal=client-secret=$(jq -r .password sp.json) \
  --from-literal=tenant-id=$(jq -r .tenant sp.json)

Apply the cluster config:

file.yaml
apiVersion: hive.run/v1
kind: HiveCluster
metadata:
  name: hive
spec:
  provider: azure
  subscriptionId: your-subscription-id
  resourceGroup: hive-rg
  region: eastus
  credentials:
    secretRef: hive-azure-credentials
  sandboxes:
    maxConcurrent: 100
    defaultImage: ubuntu:24.04
  storage:
    type: blob
    account: myhivestorage
    container: hive-data

Verify

kubectl get hivecluster hive
# NAME   STATUS   ENDPOINT                        AGE
# hive   Ready    https://hive.internal.example   2m

Connect the TypeScript client

Pass gatwayUrl to point the SDK at your deployed controller:

index.ts
import { getOrCreateSandbox, shutdown } from "hive";

const sandbox = await getOrCreateSandbox("my-sandbox", {
  fs: [{ backend: "local", mount: "/workspace", acls: [{ path: "/workspace/**", access: "rw" }] }],
}, {
  gatwayUrl: "https://hive.internal.example",
});

Back to GCP