kube-scheduler Configuration (v1beta3)

MinCandidateNodesPercentage is the minimum number of candidates to shortlist when dry running preemption as a percentage of number of nodes. Must be in the range [0, 100]. Defaults to 10% of the cluster size if unspecified.

MinCandidateNodesAbsolute is the absolute minimum number of candidates to shortlist. The likely number of candidates enumerated for dry running preemption is given by the formula: numCandidates = max(numNodes * minCandidateNodesPercentage, minCandidateNodesAbsolute) We say "likely" because there are other factors such as PDB violations that play a role in the number of candidates shortlisted. Must be at least 0 nodes. Defaults to 100 nodes if unspecified.

HardPodAffinityWeight is the scoring weight for existing pods with a matching hard affinity to the incoming pod.

IgnorePreferredTermsOfExistingPods configures the scheduler to ignore existing pods' preferred affinity rules when scoring candidate nodes, unless the incoming pod has inter-pod affinities.

Parallelism defines the amount of parallelism in algorithms for scheduling a Pods. Must be greater than 0. Defaults to 16

LeaderElection defines the configuration of leader election client.

ClientConnection specifies the kubeconfig file and client connection settings for the proxy server to use when communicating with the apiserver.

DebuggingConfiguration holds configuration for Debugging related features TODO: We might wanna make this a substruct like Debugging componentbaseconfigv1alpha1.DebuggingConfiguration

PercentageOfNodesToScore is the percentage of all nodes that once found feasible for running a pod, the scheduler stops its search for more feasible nodes in the cluster. This helps improve scheduler's performance. Scheduler always tries to find at least "minFeasibleNodesToFind" feasible nodes no matter what the value of this flag is. Example: if the cluster size is 500 nodes and the value of this flag is 30, then scheduler stops finding further feasible nodes once it finds 150 feasible ones. When the value is 0, default percentage (5%--50% based on the size of the cluster) of the nodes will be scored.

PodInitialBackoffSeconds is the initial backoff for unschedulable pods. If specified, it must be greater than 0. If this value is null, the default value (1s) will be used.

PodMaxBackoffSeconds is the max backoff for unschedulable pods. If specified, it must be greater than podInitialBackoffSeconds. If this value is null, the default value (10s) will be used.

Profiles are scheduling profiles that kube-scheduler supports. Pods can choose to be scheduled under a particular profile by setting its associated scheduler name. Pods that don't specify any scheduler name are scheduled with the "default-scheduler" profile, if present here.

Extenders are the list of scheduler extenders, each holding the values of how to communicate with the extender. These extenders are shared by all scheduler profiles.

AddedAffinity is applied to all Pods additionally to the NodeAffinity specified in the PodSpec. That is, Nodes need to satisfy AddedAffinity AND .spec.NodeAffinity. AddedAffinity is empty by default (all Nodes match). When AddedAffinity is used, some Pods with affinity requirements that match a specific Node (such as Daemonset Pods) might remain unschedulable.

Resources to be managed, the default is "cpu" and "memory" if not specified.

IgnoredResources is the list of resources that NodeResources fit filter should ignore. This doesn't apply to scoring.

IgnoredResourceGroups defines the list of resource groups that NodeResources fit filter should ignore. e.g. if group is ["example.com"], it will ignore all resource names that begin with "example.com", such as "example.com/aaa" and "example.com/bbb". A resource group name can't contain '/'. This doesn't apply to scoring.

ScoringStrategy selects the node resource scoring strategy. The default strategy is LeastAllocated with an equal "cpu" and "memory" weight.

DefaultConstraints defines topology spread constraints to be applied to Pods that don't define any in pod.spec.topologySpreadConstraints. .defaultConstraints[*].labelSelectors must be empty, as they are deduced from the Pod's membership to Services, ReplicationControllers, ReplicaSets or StatefulSets. When not empty, .defaultingType must be "List".

DefaultingType determines how .defaultConstraints are deduced. Can be one of "System" or "List".

• "System": Use kubernetes defined constraints that spread Pods among Nodes and Zones.
• "List": Use constraints defined in .defaultConstraints.

Defaults to "System".

BindTimeoutSeconds is the timeout in seconds in volume binding operation. Value must be non-negative integer. The value zero indicates no waiting. If this value is nil, the default value (600) will be used.

Shape specifies the points defining the score function shape, which is used to score nodes based on the utilization of statically provisioned PVs. The utilization is calculated by dividing the total requested storage of the pod by the total capacity of feasible PVs on each node. Each point contains utilization (ranges from 0 to 100) and its associated score (ranges from 0 to 10). You can turn the priority by specifying different scores for different utilization numbers. The default shape points are:

1. 0 for 0 utilization
2. 10 for 100 utilization All points must be sorted in increasing order by utilization.

URLPrefix at which the extender is available

Verb for the filter call, empty if not supported. This verb is appended to the URLPrefix when issuing the filter call to extender.

Verb for the preempt call, empty if not supported. This verb is appended to the URLPrefix when issuing the preempt call to extender.

Verb for the prioritize call, empty if not supported. This verb is appended to the URLPrefix when issuing the prioritize call to extender.

The numeric multiplier for the node scores that the prioritize call generates. The weight should be a positive integer

Verb for the bind call, empty if not supported. This verb is appended to the URLPrefix when issuing the bind call to extender. If this method is implemented by the extender, it is the extender's responsibility to bind the pod to apiserver. Only one extender can implement this function.

EnableHTTPS specifies whether https should be used to communicate with the extender

TLSConfig specifies the transport layer security config

HTTPTimeout specifies the timeout duration for a call to the extender. Filter timeout fails the scheduling of the pod. Prioritize timeout is ignored, k8s/other extenders priorities are used to select the node.

NodeCacheCapable specifies that the extender is capable of caching node information, so the scheduler should only send minimal information about the eligible nodes assuming that the extender already cached full details of all nodes in the cluster

ManagedResources is a list of extended resources that are managed by this extender.

• A pod will be sent to the extender on the Filter, Prioritize and Bind (if the extender is the binder) phases iff the pod requests at least one of the extended resources in this list. If empty or unspecified, all pods will be sent to this extender.
• If IgnoredByScheduler is set to true for a resource, kube-scheduler will skip checking the resource in predicates.

Ignorable specifies if the extender is ignorable, i.e. scheduling should not fail when the extender returns an error or is not reachable.

Name is the extended resource name.

IgnoredByScheduler indicates whether kube-scheduler should ignore this resource when applying predicates.

Server should be accessed without verifying the TLS certificate. For testing only.

ServerName is passed to the server for SNI and is used in the client to check server certificates against. If ServerName is empty, the hostname used to contact the server is used.

Server requires TLS client certificate authentication

Server requires TLS client certificate authentication

Trusted root certificates for server

CertData holds PEM-encoded bytes (typically read from a client certificate file). CertData takes precedence over CertFile

KeyData holds PEM-encoded bytes (typically read from a client certificate key file). KeyData takes precedence over KeyFile

CAData holds PEM-encoded bytes (typically read from a root certificates bundle). CAData takes precedence over CAFile

SchedulerName is the name of the scheduler associated to this profile. If SchedulerName matches with the pod's "spec.schedulerName", then the pod is scheduled with this profile.

Plugins specify the set of plugins that should be enabled or disabled. Enabled plugins are the ones that should be enabled in addition to the default plugins. Disabled plugins are any of the default plugins that should be disabled. When no enabled or disabled plugin is specified for an extension point, default plugins for that extension point will be used if there is any. If a QueueSort plugin is specified, the same QueueSort Plugin and PluginConfig must be specified for all profiles.

PluginConfig is an optional set of custom plugin arguments for each plugin. Omitting config args for a plugin is equivalent to using the default config for that plugin.

Name defines the name of plugin

Weight defines the weight of plugin, only used for Score plugins.

Name defines the name of plugin being configured

Args defines the arguments passed to the plugins at the time of initialization. Args can have arbitrary structure.

Enabled specifies plugins that should be enabled in addition to default plugins. If the default plugin is also configured in the scheduler config file, the weight of plugin will be overridden accordingly. These are called after default plugins and in the same order specified here.

Disabled specifies default plugins that should be disabled. When all default plugins need to be disabled, an array containing only one "*" should be provided.

PreEnqueue is a list of plugins that should be invoked before adding pods to the scheduling queue.

QueueSort is a list of plugins that should be invoked when sorting pods in the scheduling queue.

PreFilter is a list of plugins that should be invoked at "PreFilter" extension point of the scheduling framework.

Filter is a list of plugins that should be invoked when filtering out nodes that cannot run the Pod.

PostFilter is a list of plugins that are invoked after filtering phase, but only when no feasible nodes were found for the pod.

PreScore is a list of plugins that are invoked before scoring.

Score is a list of plugins that should be invoked when ranking nodes that have passed the filtering phase.

Reserve is a list of plugins invoked when reserving/unreserving resources after a node is assigned to run the pod.

Permit is a list of plugins that control binding of a Pod. These plugins can prevent or delay binding of a Pod.

PreBind is a list of plugins that should be invoked before a pod is bound.

Bind is a list of plugins that should be invoked at "Bind" extension point of the scheduling framework. The scheduler call these plugins in order. Scheduler skips the rest of these plugins as soon as one returns success.

PostBind is a list of plugins that should be invoked after a pod is successfully bound.

MultiPoint is a simplified config section to enable plugins for all valid extension points. Plugins enabled through MultiPoint will automatically register for every individual extension point the plugin has implemented. Disabling a plugin through MultiPoint disables that behavior. The same is true for disabling "*" through MultiPoint (no default plugins will be automatically registered). Plugins can still be disabled through their individual extension points.

In terms of precedence, plugin config follows this basic hierarchy

1. Specific extension points
2. Explicitly configured MultiPoint plugins
3. The set of default plugins, as MultiPoint plugins This implies that a higher precedence plugin will run first and overwrite any settings within MultiPoint. Explicitly user-configured plugins also take a higher precedence over default plugins. Within this hierarchy, an Enabled setting takes precedence over Disabled. For example, if a plugin is set in both multiPoint.Enabled and multiPoint.Disabled, the plugin will be enabled. Similarly, including multiPoint.Disabled = '*' and multiPoint.Enabled = pluginA will still register that specific plugin through MultiPoint. This follows the same behavior as all other extension point configurations.

Shape is a list of points defining the scoring function shape.

Name of the resource.

Weight of the resource.

Type selects which strategy to run.

Resources to consider when scoring. The default resource set includes "cpu" and "memory" with an equal weight. Allowed weights go from 1 to 100. Weight defaults to 1 if not specified or explicitly set to 0.

Arguments specific to RequestedToCapacityRatio strategy.

Utilization (x axis). Valid values are 0 to 100. Fully utilized node maps to 100.

Score assigned to given utilization (y axis). Valid values are 0 to 10.

kubeconfig is the path to a KubeConfig file.

acceptContentTypes defines the Accept header sent by clients when connecting to a server, overriding the default value of 'application/json'. This field will control all connections to the server used by a particular client.

contentType is the content type used when sending data to the server from this client.

qps controls the number of queries per second allowed for this connection.

burst allows extra queries to accumulate when a client is exceeding its rate.

enableProfiling enables profiling via web interface host:port/debug/pprof/

enableContentionProfiling enables block profiling, if enableProfiling is true.

leaderElect enables a leader election client to gain leadership before executing the main loop. Enable this when running replicated components for high availability.

leaseDuration is the duration that non-leader candidates will wait after observing a leadership renewal until attempting to acquire leadership of a led but unrenewed leader slot. This is effectively the maximum duration that a leader can be stopped before it is replaced by another candidate. This is only applicable if leader election is enabled.

renewDeadline is the interval between attempts by the acting master to renew a leadership slot before it stops leading. This must be less than or equal to the lease duration. This is only applicable if leader election is enabled.

retryPeriod is the duration the clients should wait between attempting acquisition and renewal of a leadership. This is only applicable if leader election is enabled.

resourceLock indicates the resource object type that will be used to lock during leader election cycles.

resourceName indicates the name of resource object that will be used to lock during leader election cycles.

resourceName indicates the namespace of resource object that will be used to lock during leader election cycles.