scaleutils

package
v0.4.3 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Mar 12, 2024 License: MPL-2.0 Imports: 15 Imported by: 4

Documentation

Index

Constants

View Source 
const (
	// Deprecated. Please use defaultNodeDrainDeadline.
	//
	// DefaultDrainDeadline is the drainSpec deadline used if one is not
	// specified by an operator.
	DefaultDrainDeadline = 15 * time.Minute

	// Deprecated. Please use defaultNodeIgnoreSystemJobs.
	DefaultIgnoreSystemJobs = false
)
View Source 
const (
	XNodeFilterOptionIgnoreInit  = "node_filter_ignore_init"
	XNodeFilterOptionIgnoreDrain = "node_filter_ignore_drain"
)

EXPERIMENTAL Node filter options are experimental features and should not be used.

Variables

This section is empty.

Functions

func FilterNodes added in v0.3.0 

func FilterNodes(n []*api.NodeListStub, idFn func(*api.NodeListStub) bool) ([]*api.NodeListStub, error)

FilterNodes returns a filtered list of nodes which are active in the cluster and where they pass the match performed by the idFn. In the event that nodes are found within the pool in an unstable state, and thus indicating there is change occurring; an error will be returned.

func FilterNodesWithOptions added in v0.4.1 

func FilterNodesWithOptions(n []*api.NodeListStub, idFn func(*api.NodeListStub) bool, opts *NodeFilterOptions) ([]*api.NodeListStub, error)

FilterNodesWithOptions is an experimental function. Use FilterNodes instead.

Types

type ClusterNodeIDLookupFunc added in v0.3.0 

type ClusterNodeIDLookupFunc func(*api.Node) (string, error)

ClusterNodeIDLookupFunc is the callback function signature used to identify a nodes remote ID from the api.Node object. This allows the function to be defined by external plugins.

type ClusterScaleInNodeIDStrategy added in v0.3.0 

type ClusterScaleInNodeIDStrategy string

ClusterScaleInNodeIDStrategy identifies the method in which nodes are selected from the node pool for removal during scale in actions. 

const (
	// NewestCreateIndexClusterScaleInNodeIDStrategy uses the Nomad
	// Nodes().List() output in the order it is presented. This means we do not
	// need additional sorting and thus it is fastest. In an environment that
	// uses bin-packing this may also be preferable as nodes with older create
	// indexes are expected to be most packed.
	NewestCreateIndexClusterScaleInNodeIDStrategy ClusterScaleInNodeIDStrategy = "newest_create_index"
)

type ClusterScaleUtils added in v0.3.0 

type ClusterScaleUtils struct {

	// ClusterNodeIDLookupFunc is the callback function used to translate a
	// Nomad nodes ID to the remote resource ID used by the target platform.
	ClusterNodeIDLookupFunc ClusterNodeIDLookupFunc
	// contains filtered or unexported fields
}

ClusterScaleUtils provides common functionality when performing horizontal cluster scaling evaluations and actions.

func NewClusterScaleUtils added in v0.3.0 

func NewClusterScaleUtils(cfg *api.Config, log hclog.Logger) (*ClusterScaleUtils, error)

NewClusterScaleUtils instantiates a new ClusterScaleUtils object for use.

func (*ClusterScaleUtils) DrainNodes added in v0.3.1 

func (c *ClusterScaleUtils) DrainNodes(ctx context.Context, cfg map[string]string, nodes []NodeResourceID) error

DrainNodes iterates the provided nodeID list and performs a drain on each one. Each node drain is monitored and events logged until the context is closed or all drains reach a terminal state.

func (*ClusterScaleUtils) IdentifyScaleInNodes added in v0.3.1 

func (c *ClusterScaleUtils) IdentifyScaleInNodes(cfg map[string]string, num int) ([]*api.NodeListStub, error)

func (*ClusterScaleUtils) IdentifyScaleInRemoteIDs added in v0.3.1 

func (c *ClusterScaleUtils) IdentifyScaleInRemoteIDs(nodes []*api.NodeListStub) ([]NodeResourceID, error)

func (*ClusterScaleUtils) IsPoolReady added in v0.3.0 

func (c *ClusterScaleUtils) IsPoolReady(cfg map[string]string) (bool, error)

IsPoolReady provides a method for understanding whether the node pool is in a state that allows it to be safely scaled. This should be used by target plugins when providing their status response. A non-nil error indicates there was a problem performing the check.

func (*ClusterScaleUtils) RunPostScaleInTasks added in v0.3.0 

func (c *ClusterScaleUtils) RunPostScaleInTasks(_ context.Context, cfg map[string]string, ids []NodeResourceID) error

RunPostScaleInTasks triggers any tasks which should occur after the nodes have been terminated within the remote provider.

The context is currently ignored on purpose, pending investigation into plugging this into the Nomad API query meta.

func (*ClusterScaleUtils) RunPostScaleInTasksOnFailure added in v0.3.1 

func (c *ClusterScaleUtils) RunPostScaleInTasksOnFailure(nodes []NodeResourceID) error

RunPostScaleInTasksOnFailure is a utility that runs tasks on nodes which were unable to be successfully terminated. The current tasks are:

  • modify node eligibility to true

func (*ClusterScaleUtils) RunPreScaleInTasks added in v0.3.0 

func (c *ClusterScaleUtils) RunPreScaleInTasks(ctx context.Context, cfg map[string]string, num int) ([]NodeResourceID, error)

RunPreScaleInTasks triggers all the tasks, including node identification and draining, required before terminating the nodes in the remote provider. COMPAT(0.4)

func (*ClusterScaleUtils) RunPreScaleInTasksWithRemoteCheck added in v0.3.4 

func (c *ClusterScaleUtils) RunPreScaleInTasksWithRemoteCheck(ctx context.Context, cfg map[string]string, remoteIDs []string, num int) ([]NodeResourceID, error)

RunPreScaleInTasksWithRemoteCheck triggers all the tasks, including node identification, filtering by remote ID, and draining, required before terminating the nodes in the remote provider.

func (*ClusterScaleUtils) SelectScaleInNodes added in v0.3.4 

func (c *ClusterScaleUtils) SelectScaleInNodes(nodes []*api.NodeListStub, cfg map[string]string, num int) ([]*api.NodeListStub, error)

type IdentifierKey 

type IdentifierKey string

Deprecated. Please use nodepool.ClusterNodePoolIdentifier.

IdentifierKey is the identifier to group nodes into a pool of resource and thus forms the scalable object. 

const IdentifierKeyClass IdentifierKey = "class"

IdentifierKeyClass uses the Node.Class field to identify nodes into pools of resource. This is the default.

type NodeFilterOptions added in v0.4.1 

type NodeFilterOptions struct {
	// contains filtered or unexported fields
}

func NewNodeFilterOptions added in v0.4.1 

func NewNodeFilterOptions(cfg map[string]string) (*NodeFilterOptions, error)

func (*NodeFilterOptions) IgnoreDrain added in v0.4.1 

func (n *NodeFilterOptions) IgnoreDrain() bool

func (*NodeFilterOptions) IgnoreInit added in v0.4.1 

func (n *NodeFilterOptions) IgnoreInit() bool

type NodeID 

type NodeID struct {
	NomadID, RemoteID string
}

Deprecated. Please use NodeResourceID.

NodeID provides a mapping between the Nomad ID of a node and its remote infrastructure provider specific ID.

type NodeIDStrategy 

type NodeIDStrategy string

Deprecated. Please use ClusterScaleInNodeIDStrategy.

NodeIDStrategy is the strategy used to identify nodes for removal as part of scaling in. 

const IDStrategyNewestCreateIndex NodeIDStrategy = "newest_create_index"

IDStrategyNewestCreateIndex uses the Nomad Nodes().List() output in the order it is presented. This means we do not need additional sorting and thus it is fastest. In an environment that uses bin-packing this may also be preferable as nodes with older create indexes are expected to be most packed.

type NodeResourceID added in v0.3.0 

type NodeResourceID struct {

	// NomadNodeID is the ID as seen within the Nomad api.Node object.
	NomadNodeID string

	// RemoteResourceID is the remote resource ID of the server/instance such
	// as the AWS EC2 instance ID.
	RemoteResourceID string
}

NodeResourceID maps a Nomad node ID to a remote resource ID.

type PoolIdentifier 

type PoolIdentifier struct {
	IdentifierKey IdentifierKey
	Value         string
}

Deprecated. Please use nodepool.ClusterNodePoolIdentifier.

PoolIdentifier is the information used to identify nodes into pools of resources. This then forms our scalable unit.

func (*PoolIdentifier) IdentifyNodes 

func (p *PoolIdentifier) IdentifyNodes(n []*api.NodeListStub) ([]*api.NodeListStub, error)

IdentifyNodes filters the supplied node list based on the PoolIdentifier params.

func (*PoolIdentifier) Validate added in v0.3.0 

func (p *PoolIdentifier) Validate() error

Validate is used to check the validation of the PoolIdentifier object.

type RemoteProvider 

type RemoteProvider string

Deprecated. Pleas use ClusterNodeIDLookupFunc.

RemoteProvider is infrastructure provider which hosts and therefore manages the Nomad client instances. This is used to understand how to translate the Nomad NodeID to an ID that the provider understands. 

const RemoteProviderAWSInstanceID RemoteProvider = "aws_instance_id"

RemoteProviderAWSInstanceID is the Amazon Web Services remote provider for EC2 instances. This provider will use the node attribute as defined by nodeAttrAWSInstanceID to perform ID translation. 

const RemoteProviderAzureInstanceID RemoteProvider = "azure_instance_id"

RemoteProviderAzureInstanceID is the Azure remote provider for VM instances. This provider will use the node attribute as defined by nodeAttrAzureInstanceID to perform ID translation. 

const RemoteProviderGCEInstanceID RemoteProvider = "gce_instance_id"

type ScaleIn 

type ScaleIn struct {
	// contains filtered or unexported fields
}

Deprecated. Please use ClusterScaleUtils.

func NewScaleInUtils 

func NewScaleInUtils(cfg *api.Config, log hclog.Logger) (*ScaleIn, error)

Deprecated. Please use NewClusterScaleUtils.

NewScaleInUtils returns a new ScaleIn implementation which provides helper functions for performing scaling in operations.

func (*ScaleIn) Ready added in v0.3.0 

func (si *ScaleIn) Ready(id PoolIdentifier) (bool, error)

Ready provides a method for understanding whether the node pool is in a state that allows it to be safely scaled. This should be used by target plugins when providing their status response. A non-nil error indicates there was a problem performing the check.

func (*ScaleIn) RunPostScaleInTasks 

func (si *ScaleIn) RunPostScaleInTasks(cfg map[string]string, nodes []NodeID) error

RunPostScaleInTasks runs any tasks that need to occur after a remote node provider has completed its work. It handles any users configuration so that the plugin does not need to perform this work.

func (*ScaleIn) RunPreScaleInTasks 

func (si *ScaleIn) RunPreScaleInTasks(ctx context.Context, req *ScaleInReq) ([]NodeID, error)

RunPreScaleInTasks helps tie together all the tasks required prior to scaling in Nomad nodes, and thus terminating the server in the remote provider.

type ScaleInReq 

type ScaleInReq struct {

	// Num is the number of nodes we should select and prepare for termination.
	Num int

	// DrainDeadline is the deadline used within the DrainSpec when performing
	// Nomad Node drain.
	DrainDeadline    time.Duration
	IgnoreSystemJobs bool

	PoolIdentifier *PoolIdentifier
	RemoteProvider RemoteProvider
	NodeIDStrategy NodeIDStrategy
}

Deprecated. Please use NewClusterScaleUtils.

ScaleInReq represents an individual cluster scaling request and encompasses all the information needed to perform the pre-termination tasks.

Source Files

View all Source files

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.