Version: v2.10.0

Create Cluster

A Yeedu cluster is the compute unit that runs your jobs and notebooks. This page documents every field on the New Cluster form as it appears in Yeedu v2.10.1.

Accessing Cluster Creation

Navigate to Clusters in the tenant navigation panel.
Click the + Cluster button (top-right, above the Actions column).
The form opens at /tenant/<tenant-id>/create-cluster with the breadcrumb Clusters > New Cluster.

Create Button

The page uses a two-column layout:

Left column – the cluster configuration form.
Right column – a live CLUSTER DETAILS summary card that updates as you fill the form.

Create Cluster

Field Order

The form presents fields in this order. Fields marked * are required.

#	Field	Control
1	Cluster Name*	Text
2	Description	Textarea
3	Cluster Type*	Dropdown
4	Cloud Environment*	Dropdown + inline create
5	Acceleration*	Radio group
6	Dependency Repository	Dropdown (clearable) + inline create
7	Run Time Version*	Dropdown
8	Compute*	Select Compute modal
9	Min Instances* / Max Instances* / Number of Parallel Execution*	Numeric
10	Auto Shutdown	Numeric (minutes)
11	Labels	Key/value rows
12	Metastore	Dropdown + inline create
13	Mounts, Spark Config, Advanced Options	Collapsible sections

Basic Details

Cluster Name

Required. Placeholder: example_cluster_1.
Max length: 64 characters.
Allowed: lowercase letters, -, _, @, .

Description

Optional free-text description of the cluster.

Cluster Type

The Cluster Type dropdown offers exactly three values (these are the literal values shown in the UI and used by the Cluster Type filter on the Clusters list):

Value	Notes
`YEEDU`	Fully managed Spark environment optimized for CPU performance; runs Spark in local mode.
`STANDALONE`	Standalone Spark deployment.
`CLUSTER`	Cluster-mode Spark deployment.

Note: Cluster type cannot be modified after creation.

Cloud Environment

The Cloud Environment* dropdown selects an existing cloud environment. Alongside the dropdown are:

Open in new – opens the selected cloud environment in a new tab.
Show details (eye) – expands the environment's details inline.
+ – opens the Create Cloud Environment modal without leaving the form.

Supported cloud providers

Yeedu v2.10.1 supports five providers:

Provider	Notes
OCI	Oracle Cloud Infrastructure.
OnPrem	On-premises environment backed by Proxmox VE.
GCP	Google Cloud Platform.
AWS	Amazon Web Services.
Azure	Microsoft Azure.

Cloud environment fields

Every cloud environment has these common fields:

Cloud Environment Name*
Description
Cloud Provider*
Credential*
Network*
Boot Disk Image*
Labels

The remaining fields depend on the provider:

Provider	Provider-specific fields
OCI	Availability Domain*
OnPrem	End Point* (for example `https://onprem-host:8006/`, the Proxmox VE port) and OnPrem Nodes* (select nodes, with a Fetch Nodes action)
GCP	Availability Zone, Project ID
AWS	Availability Zone, Account ID
Azure	Availability Zone, Resource Group

Credential types

When creating a credential inline, the Cloud Provider is locked to the parent form's provider. Each provider exposes exactly one credential type, except OnPrem, which exposes two.

Provider	Credential Type	Fields
OCI	OCI API Key Pair	Tenancy OCID, User OCID, Fingerprint, Compartment OCID, Private Key File* (`.pem` upload / drag-and-drop)
OnPrem	Proxmox Basic Auth	Username* (the UI appends `@pam` to the username), Password*
OnPrem	Custom S3a Credentials	Access Key ID, Custom S3a Endpoint, Secret Access Key*
GCP	Google Service Account	Credentials* (service-account JSON)
AWS	AWS Access Secret Key Pair	Access Key ID, Default Region (dropdown), Secret Access Key*
Azure	Azure Service Principal	Client ID, Tenant ID, Client Secret, Subscription ID, Storage Account Name, Container Name

Cloud Permissions Required by Each Credential

The credential you attach to a cloud environment is the principal Yeedu acts as. Our provisioning engine uses it in two passes: Terraform creates and destroys the node, its disks and its network attachment, and the cloud SDK then starts, stops and inspects that node for the rest of its life.

Everything below comes from the Terraform we actually run, one module per cloud, plus the SDK calls the engine makes around it. So the tables are short on purpose. We never create a VPC, a subnet, a security group, a resource group or a Proxmox datastore for you, because those are inputs you supply on the cloud environment and we only read them.

Missing permissions surface late. The form accepts the credential and the failure lands during cluster bootstrap, not at submit time.

Credential type	Cloud	What the credential provisions
OCI API Key Pair	OCI	Compute instances, block volumes, a VNIC on your subnet
Proxmox Basic Auth	OnPrem	Proxmox VE guests, their disks, a cloud-init snippet
Custom S3a Credentials	OnPrem	Nothing. Object storage access only
Google Service Account	GCP	Compute instances, persistent disks, a reserved internal address
AWS Access Secret Key Pair	AWS	EC2 instances, EBS volumes, an imported key pair, CloudWatch log groups
Azure Service Principal	Azure	Linux VMs, managed disks, a NIC, optional Azure Monitor wiring

note

We derive these permission sets from the resources our Terraform declares and the SDK calls our engine makes. Yeedu doesn't ship a ready-made IAM policy, role definition or Proxmox role for any cloud, so treat each list as the minimum we know a deployment needs and widen it if your organisation's guardrails require more.

OCI: OCI API Key Pair

The Tenancy OCID, User OCID, Fingerprint and private key authenticate an API signing key. The Compartment OCID scopes every resource we create.

Terraform declaration	Kind	Created when
`oci_identity_availability_domains`	data	Always. Resolves the Availability Domain you picked to its full OCI name
`oci_core_instance`	resource	Always. Carries the VNIC, which reads your subnet and optional NSGs
`oci_core_volume`	resource	One per disk in Disks Config, while the node runs or the scratch disk is kept
`oci_core_volume_attachment`	resource	One per volume above

Scope the policy to the compartment named on the credential. The availability-domain lookup is the one statement that has to sit at tenancy level:

Allow group YeeduClusterProvisioners to inspect availability-domains in tenancy
Allow group YeeduClusterProvisioners to manage instance-family in compartment <compartment-name>
Allow group YeeduClusterProvisioners to manage volume-family in compartment <compartment-name>
Allow group YeeduClusterProvisioners to use subnets in compartment <compartment-name>
Allow group YeeduClusterProvisioners to use vnics in compartment <compartment-name>

Add this only when your cloud environment supplies network security groups:

Allow group YeeduClusterProvisioners to use network-security-groups in compartment <compartment-name>

Spot needs nothing extra on OCI. The preemptible configuration is a field on the instance, so manage instance-family already covers it. Cloud logging setup is skipped entirely for OCI, which is why no logging statement appears above.

OnPrem: Proxmox Basic Auth

The Username, to which the UI appends @pam, and the Password authenticate against the Proxmox VE API at your End Point. Unlike every other provider, on-prem start and stop go through Terraform rather than an SDK, by toggling the guest's started flag.

Terraform declaration	Kind	Created when
`proxmox_virtual_environment_nodes`	data	Always. Lists cluster nodes so we can pick one with free CPU and memory
`proxmox_virtual_environment_vms`	data	Per selected node. Reads existing guests and their tags to compute headroom
`proxmox_virtual_environment_file`	resource	Always. The cloud-init user-data, written as a snippet on the `local` datastore
`proxmox_virtual_environment_vm`	resource	Always. Boot disk plus one disk per entry in Disks Config, on `local-lvm`

A role covering the guest lifecycle, both datastores and node audit is enough:

pveum role add YeeduClusterProvisioner -privs \
  "VM.Allocate,VM.Audit,VM.Config.CPU,VM.Config.Cloudinit,VM.Config.Disk,VM.Config.HWType,\
VM.Config.Memory,VM.Config.Network,VM.Config.Options,VM.Monitor,VM.PowerMgmt,\
Datastore.Audit,Datastore.AllocateSpace,Datastore.AllocateTemplate,Sys.Audit"

pveum acl modify / -user yeedu@pam -role YeeduClusterProvisioner

VM.Monitor is there because we enable the QEMU guest agent and read the guest's IPv4 address back through it. Sys.Audit covers the node listing. Datastore.AllocateTemplate is what permits the snippet upload.

note

The snippet upload is easy to overlook. We write the node's cloud-init user-data to the local datastore, so that datastore must both accept the Snippets content type and be writable by this account. Otherwise the guest gets created and never bootstraps.

OnPrem: Custom S3a Credentials

This one provisions nothing. It never reaches a Terraform provider, and no resource in any module reads it.

What it does is authenticate the cluster node against an S3-compatible object store, in Yeedu's own on-prem setup a Ceph RADOS Gateway. The node's bootstrap exports the key pair as AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and S3_ENDPOINT, then configures Spark with org.apache.hadoop.fs.s3a.S3AFileSystem and path-style access against that endpoint.

Field	Used as
Access Key ID	S3 access key on the node and in the Spark S3A config
Secret Access Key	S3 secret key
Custom S3a Endpoint	`spark.hadoop.fs.s3a.endpoint`, with `path.style.access` set to `true`

Its permissions come from your object store's own bucket policy, not from any cloud IAM. Grant read and write on the buckets you reference: the bucket behind a Dependency Repository, and the bucket in each s3:// or s3a:// mount source. Read-only mounts need read alone.

An OnPrem cloud environment still needs a Proxmox Basic Auth credential to provision anything.

GCP: Google Service Account

The service-account JSON authenticates Terraform, and by default the same account is attached to the node as its identity with the cloud-platform scope. That second use is why iam.serviceAccountUser appears below.

Terraform declaration	Kind	Created when
`google_compute_address`	resource	Always. A reserved internal IP in your subnet, purpose `GCE_ENDPOINT`
`google_compute_instance`	resource	Always. GPUs attached when CUDA acceleration is selected
`google_compute_disk`	resource	Per disk in Disks Config, unless the disk type is `local-ssd` or `scratch-disk`
`google_compute_attached_disk`	resource	One per disk above

Two predefined roles cover the baseline. Grant them on the project named on the cloud environment:

roles/compute.instanceAdmin.v1
roles/iam.serviceAccountUser

If you'd rather define a custom role, this is the permission set our modules and SDK calls exercise:

compute.instances.create
compute.instances.delete
compute.instances.get
compute.instances.start
compute.instances.stop
compute.instances.attachDisk
compute.instances.detachDisk
compute.instances.setMetadata
compute.instances.setLabels
compute.instances.setTags
compute.instances.setServiceAccount
compute.disks.create
compute.disks.delete
compute.disks.get
compute.disks.use
compute.addresses.create
compute.addresses.delete
compute.addresses.get
compute.addresses.use
compute.subnetworks.use
compute.images.useReadOnly
compute.zoneOperations.get
iam.serviceAccounts.actAs

Three cases add to that. When the cloud environment's network project differs from its Project ID, the account needs roles/compute.networkUser on the network project, because we configure a second provider pointed at it. When Enable Public IP is on, add compute.subnetworks.useExternalIp. When CUDA is selected, add compute.acceleratorTypes.get and make sure the GPU quota exists in the zone.

Nothing on the GCP path creates logging infrastructure. Our engine treats cloud logging setup as a no-op for GCP.

AWS: AWS Access Secret Key Pair

Terraform declaration	Kind	Created when
`aws_subnet`	data	Always. Reads the availability zone of the subnet you supplied
`aws_key_pair`	resource	Always. Imports Yeedu's public key, one key pair per node
`aws_instance`	resource	On-demand nodes only
`aws_spot_instance_request`	resource	Spot nodes only
`aws_ebs_volume`	resource	Per disk in Disks Config, while the node runs or the scratch disk is kept
`aws_volume_attachment`	resource	One per volume above
`aws_ec2_tag`	resource	Spot nodes only. Copies the label set onto the fulfilled instance

On top of that, the engine calls StartInstances, StopInstances and DescribeInstances directly, because restarting an instance that already exists through the SDK takes a couple of seconds where a full Terraform cycle takes closer to a minute, and it resolves an instance by its Name tag before starting it, which is why DescribeTags sits in the baseline alongside the describe calls for instances, images, volumes and subnets. It also creates nine CloudWatch log groups if they don't already exist: yeedu_bootstrap_log, yeedu_unstructured_log, yeedu_usi_reactors_logs, yeedu_log_collector_reactors, yeedu_log_collector_history_server, yeedu_copy_object_storage_logs, yeedu_workspace_files_sync_logs, yeedu_usi_reactors_monitor_logs and yeedu_usi_telegraf_logs.

Here's the baseline policy for an on-demand cluster.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid": "YeeduNodeLifecycle",
      "Effect": "Allow",
      "Action": [
        "ec2:RunInstances",
        "ec2:TerminateInstances",
        "ec2:StartInstances",
        "ec2:StopInstances",
        "ec2:DescribeInstances",
        "ec2:DescribeInstanceAttribute",
        "ec2:DescribeInstanceTypes",
        "ec2:DescribeImages",
        "ec2:DescribeSubnets",
        "ec2:ImportKeyPair",
        "ec2:DescribeKeyPairs",
        "ec2:DeleteKeyPair",
        "ec2:CreateVolume",
        "ec2:DeleteVolume",
        "ec2:DescribeVolumes",
        "ec2:AttachVolume",
        "ec2:DetachVolume",
        "ec2:CreateTags",
        "ec2:DescribeTags",
        "ec2:DeleteTags"
      ],
      "Resource": "*"
    },
    {
      "Sid": "YeeduLogGroups",
      "Effect": "Allow",
      "Action": [
        "logs:DescribeLogGroups",
        "logs:CreateLogGroup"
      ],
      "Resource": "*"
    }
  ]
}

Turning on Spot Instances (Experimental) switches the module to a spot request, and immediately after fulfilment we call ModifyInstanceMetadataOptions on the new instance to enable metadata tags:

{
  "Sid": "YeeduSpotNodes",
  "Effect": "Allow",
  "Action": [
    "ec2:RequestSpotInstances",
    "ec2:DescribeSpotInstanceRequests",
    "ec2:CancelSpotInstanceRequests",
    "ec2:ModifyInstanceMetadataOptions"
  ],
  "Resource": "*"
}

Attaching an instance profile through the cloud environment's cloud identity needs iam:PassRole for that role, since the profile name is passed straight to RunInstances.

warning

Leave the cloud identity empty and the node falls back to these same keys for its own work. The bootstrap script then exports them into the node's environment, and the credential additionally needs s3:GetObject and s3:ListBucket on the Yeedu software bucket, logs:CreateLogStream and logs:PutLogEvents for the awslogs agent, and ecr:GetAuthorizationToken plus ecr:BatchGetImage, ecr:GetDownloadUrlForLayer and ecr:BatchCheckLayerAvailability if your container images live in ECR. Attaching an instance profile is the cleaner option.

Azure: Azure Service Principal

The Client ID, Tenant ID, Client Secret and Subscription ID authenticate the service principal. The Resource Group on the cloud environment is where every VM resource lands.

Terraform declaration	Kind	Created when
`azurerm_subnet`	data	Always. Looked up by name in the network resource group
`azurerm_network_interface`	resource	Always
`azurerm_public_ip`	resource	Only when Enable Public IP is selected
`azurerm_linux_virtual_machine`	resource	Always. Priority flips to `Spot` for spot clusters
`azurerm_managed_disk`	resource	Per disk in Disks Config, while the node runs or the scratch disk is kept
`azurerm_virtual_machine_data_disk_attachment`	resource	One per disk above
`azurerm_virtual_machine_extension`	resource	Logging only. Installs `AzureMonitorLinuxAgent`
`azurerm_monitor_data_collection_rule`	data	Logging only. Reads `yeedu_usi_bootstrap_dcr` and `yeedu_usi_container_dcr`
`azurerm_monitor_data_collection_rule_association`	resource	Logging only. Three associations per node
`azurerm_monitor_data_collection_endpoint`	data	Logging only
`azurerm_log_analytics_workspace`	data	Logging only

Baseline actions for a custom role scoped to the subscription:

Microsoft.Resources/subscriptions/resourceGroups/read
Microsoft.Compute/virtualMachines/read
Microsoft.Compute/virtualMachines/write
Microsoft.Compute/virtualMachines/delete
Microsoft.Compute/virtualMachines/start/action
Microsoft.Compute/virtualMachines/deallocate/action
Microsoft.Compute/disks/read
Microsoft.Compute/disks/write
Microsoft.Compute/disks/delete
Microsoft.Network/networkInterfaces/read
Microsoft.Network/networkInterfaces/write
Microsoft.Network/networkInterfaces/delete
Microsoft.Network/networkInterfaces/join/action
Microsoft.Network/virtualNetworks/read
Microsoft.Network/virtualNetworks/subnets/read
Microsoft.Network/virtualNetworks/subnets/join/action

If you prefer built-in roles, Virtual Machine Contributor plus Network Contributor on the resource group is the closest fit to that list.

Public IP is the smallest add-on. Turn it on and the module creates and later deletes an azurerm_public_ip:

Microsoft.Network/publicIPAddresses/read
Microsoft.Network/publicIPAddresses/write
Microsoft.Network/publicIPAddresses/delete
Microsoft.Network/publicIPAddresses/join/action

Logging is the larger one. With it enabled, we create the custom log tables yeedu_usi_bootstrap_logs_CL and yeedu_usi_container_logs_CL in your Log Analytics workspace, create a data collection endpoint and two data collection rules, associate all three with the VM, and install the monitor agent extension:

Microsoft.Compute/virtualMachines/extensions/read
Microsoft.Compute/virtualMachines/extensions/write
Microsoft.Compute/virtualMachines/extensions/delete
Microsoft.Insights/dataCollectionEndpoints/read
Microsoft.Insights/dataCollectionEndpoints/write
Microsoft.Insights/dataCollectionRules/read
Microsoft.Insights/dataCollectionRules/write
Microsoft.Insights/dataCollectionRuleAssociations/read
Microsoft.Insights/dataCollectionRuleAssociations/write
Microsoft.Insights/dataCollectionRuleAssociations/delete
Microsoft.OperationalInsights/workspaces/read
Microsoft.OperationalInsights/workspaces/tables/read
Microsoft.OperationalInsights/workspaces/tables/write

Monitoring Contributor plus Log Analytics Contributor covers that block. One last case: if the cloud environment supplies a user-assigned managed identity as its cloud identity, the service principal also needs Microsoft.ManagedIdentity/userAssignedIdentities/assign/action on that identity. With no cloud identity set, the VM gets a system-assigned identity instead and nothing extra is required.

Acceleration

Acceleration* is a required radio group with three options:

Option	Description
No Acceleration	Standard CPU execution, no acceleration layer.
Turbo	Yeedu's Turbo acceleration engine.
CUDA	GPU (CUDA) acceleration.

The selected value is reflected in the CLUSTER DETAILS panel, and surfaces on the Job and Notebook forms as Turbo Acceleration: Enabled / Disabled.

Check the support badges on the Run Time Version row (see below) to confirm the runtime you pick advertises Cuda or Turbo support before selecting the matching acceleration mode.

Dependency Repository

Dependency Repository is an optional, clearable dropdown with a + button to create a new repository inline. It references a repository defined in the tenant-level Dependency Repositories module (/tenant/<tenant-id>/dependency-management).

The inline create form (Dependency Repository > New Repository) has these fields:

Field	Values / notes
Repository Name*	Text
Description	Text
Cloud Provider*	`OnPrem` \| `GCP` \| `AWS` \| `Azure` — note that OCI is not offered here, even though cloud environments do support OCI
Credential Name*	Dropdown of credentials for that provider, with a + to create one inline
Bucket Name*	Placeholder `e.g., my-deps-bucket`

Footer actions on that form: Reset, Create.

Dependencies

Run Time Version

Run Time Version* is a required dropdown. Each row in the list renders two groups:

Version: Spark, Scala, Python and Java versions — for example 3.5.3, 2.12.15, 3.9.5, 3.2.4.
Support: capability badges, each shown with a tick or a cross:
- Notebook
- Cuda
- Turbo
- Thrift Support

Use the badges to confirm a runtime supports the workload you intend to run on the cluster.

Compute

Compute* is its own required field (it is not part of the instance settings). Click Select Compute to open the Select Compute modal, or + to define a new compute shape.

The modal header states the architecture type, for example: Select Compute ( Architecture type : x86_64 ).

Every column is sortable and has its own filter input:

Column	Filter placeholder
Name	Search By Name
Cores	Min cores
Memory	e.g., 256 GiB
Scratch Disk Size	Scratch Disk / Size
No Of Disks	—
Total Disks Size	—
YCU/hr	Max YCU
GPUs	Min GPUs
GPU Memory	e.g., 48 GiB

Rows are grouped under category headers. The categories depend on the selected cloud environment — on an OnPrem environment the groups are CUSTOM COMPUTE and GENERAL PURPOSE.

Select a single row with its radio button, then click Select.

Instances

Field	Description
Min Instances*	Minimum number of instances kept running.
Max Instances*	Upper bound the cluster can scale to.
Number of Parallel Execution*	Maximum number of concurrent executions per instance.

Note: there is no Workers per Node field in v2.10.1 — Number of Parallel Execution is the concurrency control on this form.

Instance Settings

Auto Shutdown

Reads "Stop Each Instance After <n> Minutes Of Idle Time".
Each instance is stopped once it has been idle for the configured number of minutes.

Labels

Add labels as key / value rows using the + button; remove a row with the trash icon.
Placeholders: e.g., team (key) and e.g., data-engineering (value).

Automatically Added Labels

A collapsible panel lists the labels Yeedu attaches to every cluster without any input from you:

Label	Value
`resource`	`yeedu`
`vm`	`yeedu_node`
`tenant_id`	The tenant UUID
`created_by`	The creating user's email address
`yeedu_env`	The Yeedu environment name

Metastore

Select Metastore dropdown, with + to create a metastore inline.
Metastore types available in v2.10.1: AWS GLUE, DATABRICKS UNITY, HIVE.
The selected metastore type is shown as a badge in the Metastore column of the Clusters list.

Additional Configurations

Mounts

A collapsible section. When nothing is attached it shows the empty state "No Mounts Selected" with two buttons:

+ New – create a new mount.
Attach – attach an existing mount.

Mounts

Spark Config

A collapsible section with 10 tabs:

Tab	Purpose
Configs	Spark configuration key/value pairs.
Config Secrets	Configuration values sourced from secrets.
Packages	Maven/Spark packages.
Jars	JAR dependencies.
Archives	Archive files.
Repositories	Package repositories.
Py-files	Python file dependencies.
Files	Additional files.
Env Variables	Environment variables.
Env Secrets	Environment variables sourced from secrets.

The Configs tab has a Bulk Edit toggle, plus Key/Value rows with + (add) and trash (delete).

Spark Configs

Advanced Options

A collapsible section with 3 tabs: Init Script, Disks Config and Others.

Advanced Options

Init Script

A code editor with line numbers. Enter a script to run during node bootstrap.

Disks Config

Field	Notes
Disk Type	Dropdown. The available values are provider-specific — on an OnPrem cloud environment the only value is `local-lvm`.
Size (GiB)	Entered as `size` × `quantity`, with the computed total shown alongside. Helper text: "Enter size and quantity above".
IOPS	Provider-specific; not shown for OnPrem environments.
Throughput (MBps)	Provider-specific; not shown for OnPrem environments.

Disks config

Others

Termination Policy

"Terminate Cluster After <n> Minutes Of Inactivity." (the form ships with 240).
Keep Scratch Disk – checkbox, enabled by default.

Manage And SSH Permissions

Enable Public IP – checkbox.

Spot Instances (Experimental)

Provider-dependent: this section is not rendered for an OnPrem cloud environment. Availability on other providers depends on the selected cloud environment.

Uses discounted VMs (70–90% cheaper).
Best for stateless, retryable or batch workloads.
Can be interrupted at any time — unsuitable for long-running jobs.

CLUSTER DETAILS Panel

The right-hand column shows a live preview of the cluster as you configure it:

Cloud Provider
Cluster Type
Cloud Environment
Compute Type
Acceleration
Disk Volume
Runtime Version (with its capability badges)
Labels

The same panel component appears on the Create Job and Create Notebook forms, where it summarises the cluster you have selected.

Actions

Button	Behaviour
Reset	Clears all fields on the form.
Create	Creates the cluster without starting it.
Create & Start	Creates the cluster and immediately begins bootstrapping it.

Backend bootstrap time: ~6–7 minutes Ready to use in: ~1 minute after creation

Accessing Cluster Creation​

Field Order​

Basic Details​

Cluster Name​

Description​

Cluster Type​

Cloud Environment​

Supported cloud providers​

Cloud environment fields​

Credential types​

Cloud Permissions Required by Each Credential​

OCI: OCI API Key Pair​

OnPrem: Proxmox Basic Auth​

OnPrem: Custom S3a Credentials​

GCP: Google Service Account​

AWS: AWS Access Secret Key Pair​

Azure: Azure Service Principal​

Acceleration​

Dependency Repository​

Run Time Version​

Compute​

Select Compute modal​

Instances​

Auto Shutdown​

Labels​

Automatically Added Labels​

Metastore​

Mounts​

Spark Config​

Advanced Options​

Init Script​

Disks Config​

Others​

Spot Instances (Experimental)​

CLUSTER DETAILS Panel​

Actions​

Accessing Cluster Creation

Field Order

Basic Details

Cluster Name

Description

Cluster Type

Cloud Environment

Supported cloud providers

Cloud environment fields

Credential types

Cloud Permissions Required by Each Credential

OCI: OCI API Key Pair

OnPrem: Proxmox Basic Auth

OnPrem: Custom S3a Credentials

GCP: Google Service Account

AWS: AWS Access Secret Key Pair

Azure: Azure Service Principal

Acceleration

Dependency Repository

Run Time Version

Compute

Select Compute modal

Instances

Auto Shutdown

Labels

Automatically Added Labels

Metastore

Mounts

Spark Config

Advanced Options

Init Script

Disks Config

Others

Spot Instances (Experimental)

CLUSTER DETAILS Panel

Actions