Managing Pipelines Node Pools

Overview

Node Poolsprovide a convenient way to logically group build nodesin a JFrog Pipelines environment. This enables you to run steps simultaneously on different nodes.

Administrators may create any number of node pools.For SaaS deployments, a node pool nameddefault-dynamic-nodepoolis available by default. For on-premise deployments,at least one node pool must beaddedand designated as the default node pool, and assigned at least one build node for pipelines to execute in the selected Project.

A node pool must be designated to contain build nodes of one of these types:

Dynamic Nodes- Build nodes that are spun up and down on-demand by Pipelines from a cloud orKubernetesservice. This can help scale operations, and help manage costs by not incurring cloud service charges to run idle nodes.
Static Nodes– Build nodes (virtual machines) that are persistently available through a fixed IP address.
For information about the minimum requirements for a static node that can be assigned to JFrog Pipelines and to learn how static nodes can be added, seeManaging Pipelines Static Nodes.

Page Contents

Administering Dynamic Node Pools

Dynamic node pools spin build nodes up and down on-demand from a cloud or Kubernetes service.To create a dynamic node pool, you must configure and provide anAdmin integrationfor one of the following cloud or Kubernetes service:

A dynamic node isprovisioned when a step is triggered, then either cached or destroyed after a preset amount of idle time if no new step is triggered during that interval. When used on a cloud service, this can help manage costs by not incurring charges to run idle nodes.

The following actions can only be performed byadministrator usersto configure Pipelines dynamic node pools for operation.

Adding a Dynamic Node Pool

To create a new dynamic node pool for on-demand build nodes, go toAdministration |管道|节点池, clickAdd Node Pool, and then selectDynamic.

When a user-defined integration is selected, SaaS users can create custom dynamic nodes and customize all the relevant settings. Users can select any cloud provider with no restrictions on settings, such as machine type, region, and others.

Fields that are not required may be left blank. For example, you may not wish to set the fields to limit node count, disk usage, or timeout. When finished, clickSave Node Pool.

General Settings
Name	An alphanumeric friendly name for the node pool. When adding a node pool within a specificProject, by default,the name is prefixed with the Project Key. After a node pool is saved, its name cannot be edited.
Set as Default Node Pool	When checked, sets the node pool as the default node pool for all pipelines in the selected Project. All steps that do not explicitly define which node pool to use will run on nodes of the default node pool.
Select an Integration/Select a Build Image
Integration	Create or select from the availableAdmin integrationsfor one of the followingcloud or Kubernetes service: AWS Keys Azure Keys Google Cloud Kubernetes
Build Image	In theBuild Plane Imageswindow: Select from the available build images. For example, an Amazon Machine Image (AMI) on AWS. or ClickAdd Build Image to add your own custom VM image. For more information, seeCreating Custom VM Images.
Additional Settings
Machine Type	Select from the available machine (instance) types for the service. For example: c4.large on AWS.
Fetch Machine Types	Click to fetch all the supported machine types (in addition to the static list of machine types) from the cloud provider for the selected region/zone. The button is available only when a user-defined integration and valid zone (in some cases) is selected.
Node Limits	The maximum number of nodes that can created on-demand for this node pool.
Disk Size	Optional. Configure dynamic node disk size. The minimum limit is: AWS (Windows): 120 GB Azure (Windows): 128 GB Others: 60 GB This option is available for Pipelines on-premise versions only. If the disk size is not specified, the node will use the image’s disk size by default.
Timeout	Number of seconds for connection timeout. Default is 3600 seconds. For more information, seeSetting Step Timeouts.
Node Idle Interval	Number of minutes a node can be idle before it is destroyed.
Insecure Docker Registries	A comma-separated list of allowedinsecure docker registries.
Kubernetes Settings(seeCreating Dynamic Nodes on Kubernetesfor configuration procedures)
Storage Class	Thestorage classavailable from the service to use. Run`kubectl get storageclass`to list the StorageClasses in your cluster.
CPU	The CPU limitresource settingfor every pod.
Labels	Optional labels fornode selector.
Memory	The memory limitresource settingfor every pod.
Namespace	Thenamespacefor nodes.
AWS Settings(seeAWS Keys Integrationfor configuration requirements)
SSH Key Pair	The SSH encrypted access key for the cloud service. Name of theSSH Key pair created on AWS.
Security Group ID	An ID for an optional security group for all created nodes.
Subnet Id	Optional VPC subnet ID
GCP Settings(seeGoogle Cloud Integrationfor configuration requirements)
Zone	Zonewithin the hosting GCP region
Network Tags	OptionalNetwork tags(text attributes) for the VM instances
Subnet Name	Optional subnet name
Assign Public IP	选中时,节点会be created with public IP addresses
Azure Settings(seeAzure Keys Integrationfor configuration requirements)
Subscription ID	GUID that uniquely identifies yoursubscriptionto useAzureservices
Subnet Name	Vnet subnet name
Vnet Name	Azure Virtual Network name (e.g.,myVirtualNetwork)
Resource Group Name	Name of the resource group (e.g.,myResourceGroup)
Security Group Name	Optional NIC network security group
Public SSH Key	Public Secure Shell (SSH) key for Linux VMs
Admin Username	Username for RDP (Windows nodes only, optional)
Admin Password	Password for RDP (Windows nodes only, optional)
Enable running nodes with non-root users	When checked, a non-root user named`pipelinesRootless`is automatically created and all nodes in the node pool will use this non-root user. Limitations Running a node as a non-root user has certain limitations. For more information, seeInitializing Nodes as a Non-Root User.
Cache Settings
Enable Cache	When checked, enablesnode caching.
Reset Cache	Select the type of interval to reset the cache:On Day of WeekorBy Number of Days. Available only when node caching is enabled.
Assign Pipelines to this Node Pool
Any Pipeline Source	When selected, pipelines loaded from any pipeline source can use this node pool. Unselect this option torestrict node pool use to a set of pipeline sources.

Node Duration

When a step is triggered, a dynamic node is spun up to execute the step. After the step is complete, the node is available to execute subsequent steps until it is terminated.

When node caching is not enabled, Pipelines will check the node at each node idle interval after the node was first spun up. If the node is idle Pipelines terminates the node. This ensures that your build infrastructure is managed in an efficient manner, to help you control computing costs.

Dynamic Node Caching

When enabled, node caching helps speed up execution of your builds by avoiding having to repeat time-consuming processes, such as installing or loading large amounts of dependency data. This feature is available only on dynamic node pools.

When node caching is enabled for a dynamic node pool, Pipelines pauses your node after executing a step, instead of terminating it after the specified idle time. When the next step executes, the paused node is restarted, retaining any Docker images or dependencies that were installed as part of the previous step. This greatly speeds up builds that pull or build Docker images as part of their workflow.

In addition, the node is available to execute the job much sooner, since restarting the node from a paused state is much faster than provisioning a new node. So if your build takes just a few minutes to execute, you save on the node provisioning time for every build, which adds up quickly.

Administering Static Node Pools

Sttatic node pools are collections of persistently available build nodes (virtual machines). Static nodes enable you to execute pipeline steps on your own infrastructure. For more information, seeManaging Pipelines Static Nodes.

The following actions can only be performed byadministrator usersto configure Pipelines static node pools.

For information about troubleshooting node initialization issues, seeTroubleshooting - Node Initialization

Adding a Static Node Pool

To create a new node pool of static build nodes, go toAdministration |管道|节点池, clickAdd Node Pool, and then selectStatic.

Fields that are not required may be left blank. For example, you may not wish to set the fields to limit node count, disk usage, or timeout. When finished, clickSave Node Pool.

General Settings
Name	An alphanumeric friendly name for the node pool.
Set as Default Node Pool	When checked, sets the node pool as the default node pool for all pipelines. All steps that do not explicitly define which node pool to use will run on nodes of the default node pool.
Additional Settings
Architecture	Currently limited tox86_64
OS	Select from the list of operating systems supported for the selected architecture. For a list of supported OS and architecture combinations, refer toRuntime Images.
Runtime Version	The default runtime image that will be used to run pipeline steps. Choose a specific runtime to make sure the build nodes have the versions you need for languages/services/CLIs/etc. To learn more, seeManaging Runtimes.
Node Limits	Limits the maximum number of nodes that can be added to the node pool.
Max Disk Usage %	Limits the maximum disk usage. A build node that exceeds this maximum will be marked as failed.
Timeout	The maximum time (in seconds) after which you want your pipeline steps to timeout. This timeout is used when your step does not have any timeout specified. Default value for timeout is 3600 seconds. For more information, seeSetting Step Timeouts.
Insecure Docker Registries	A comma-separated list of allowedinsecure docker registries.
Enable running nodes with non-root users	Select to enable running the nodes as a non-root user. Ensure that the prerequisites mentioned in theInitializing Static Nodes as a Non-Root Userpage are available to run the build node as a non-root user. Limitations Running a node as a non-root user has certain limitations. For more information, seeInitializing Nodes as a Non-Root User. enforceNonRootNodes Global Setting Pipelines admins can use the`enforceNonRootNodes`system-level setting to enforce use of only those node pools that are configured with non-root. When thisglobal setting is set as`true`, it takes precedence over the non-root user setting configured in the UI.Note that this is currently supported only in Ubuntu 18 and Ubuntu 20. Depending on the type of Pipelines installation, the`enforceNonRootNodes`setting can be configured as follows: Public chart: If the installation was using the public chart, then change the setting inthevalues.ymlfile. Docker: If the installation was using Docker, then change the setting in theRuntime Configurationsection in thePipelines System YAML.
Assign Pipelines to this Node Pool
Any Pipeline Source	When selected, pipelines loaded from any pipeline source can use this node pool. Unselect this option torestrict node pool use to a set of pipeline sources.

Once you have created a static node pool, you canadd static nodes.

Administering All Node Pools

The following actions can only be performed byadministrator usersto configure Pipelines node pools (both static and dynamic) for operation.

Restricting Node Pool Use

如果池是一个节点在一个项目中,它可能是一个ssigned to one or more environments, and pipelines in pipeline sources that share at least one environment with the node pool may use the node pool. By default, node pools are in theDEV环境。

Node pools that are not in a Project are available to pipelines in pipeline sources that are not in a Project. In addition, further restrictions may be specified to restrict use of particular pipeline sources andusers that have been givenpermissionsto use those pipeline sources.

You can establish those usage restrictions when you add or modify a node pool.

UnderAssign Pipeines to this Node Pool, uncheck the option forAny Pipeline Source.
In the revealed interface, select from the list of available pipeline sources and move the row(s) (by click-and-drag or arrow button) to the list of included pipeline sources.
When finished, clickSave Node Pool.

Viewing Available Node Pools

To view thethe list of node pools already added and available for use,go toAdministration |管道|节点池orApplication |管道|节点池.

Non-administrator users will only be shown node pools that they are permitted to use.

Max Limit Exceeded

If the number of nodes in a node pool exceeds its maximum limit, it is listed with an exclamation mark in theMax Limitcolumn.

Node Provision Status

In the Node Pools list view, theProvision Statuscolumn provides a color representation of the provision status for each node. Every color represents one of the stages in the lifecycle of a node.

Color	Provision Status	Description
	Not provisioned	The node has not been requested to be provisioned.
	Waiting	When a node is provisioned, it first goes into waiting state. This means that the cloud provider is initializing the node and/or has not reported the status back to Pipelines.
	Processing	The cloud provider reported the node's status back to Pipelines and Pipelines is setting up the agent. or 节点运行构建.
	Success	The node successfully initialized and is waiting for a Pipelines build.
	Failure	The node failed. This is usually caused by an out of memory issue. However, there could be several other reasons, which are reflected in the node console.
	Stopped	When a node is deleted, its provisioning is first stopped and then it is deleted.
	Cached	The node is stopped and not running, but the actual instance is still available for instant use. Caching must be enabledfor a node to be cached.

Viewing Node Pool Details

To view the details of a node pool, click the name of the node pool in the list of available node pools.

Modifying or Removing a Node Pool

Cloud only: The default node pooldefault-dynamic-nodepoolis managed by JFrog and certain settings, such as Machine Type, cannot be modified.

To edit the node pool settings:

View the node poolas shown above.
ClickEdit.
Settings that may not be changed have gray fields.
When finished changing settings, clickSave Node Pool.

To delete a node pool:

View the node poolas shown above.
Click删除.
Confirm you wish to delete by clickingConfirm.

Sharing Node Pools across Multiple Projects

Projectadmins can share node pools across multiple projects to allow members in more than one project to use them. A classic use case for sharing a node pool is the need to share computing resources when running pipelines frommultipleprojects. A node pool can be shared with a single project or across multiple projects. After a node pool is shared, it is added to the target projects and displayed on the node pools page.

Shared Node Pool Rules and Guidelines

A node pool can be shared only after it has been created.
A node pool can be shared only if it has a globally unique name.
Only the owner of the shared node pool can delete the shared node pool.

To share a node pool:

Select the relevant Project and in theAdministrationmodule, go to管道|节点池.
Hover over the node pool you want to share and click theShareicon.
You can share the selected Integration:
- Across all the projects, by selecting theAuto share with all projectsfield.
- Only with specific projects, by selecting the name of the Project from the projects list.
ClickSave.

Content

Space Tools

Managing Pipelines Node Pools

Overview

Administering Dynamic Node Pools

Adding a Dynamic Node Pool

Name

Set as Default Node Pool

Integration

Build Image

Machine Type

Fetch Machine Types

Node Limits

Disk Size

Timeout

Node Idle Interval

Insecure Docker Registries

Storage Class

CPU

Labels

Memory

Namespace

SSH Key Pair

Security Group ID

Subnet Id

Zone

Network Tags

Subnet Name

Assign Public IP

Subscription ID

Subnet Name

Vnet Name

Resource Group Name

Security Group Name

Public SSH Key

Admin Username

Admin Password

Enable running nodes with non-root users

Enable Cache

Reset Cache

Any Pipeline Source