rescheduling

const (
	// RePropertyName name specifying re-scheduler cm
	RePropertyName = "re-scheduling"
	// ReschedulingReasonKey is used to record the reason of rescheduling
	ReschedulingReasonKey = "rescheduling-reason"
	// CmName Name of ReSchedulerConfigmap
	CmName = "vcjob-fault-npu-cm"
	// CmNameSpace Namespace of ReSchedulerConfigmap
	CmNameSpace = "volcano-system"
	// RescheduleReasonCmName Name of RescheduleReasonConfigmap
	RescheduleReasonCmName = "job-reschedule-reason"
	// RescheduleReasonCmNamespace Namespace of RescheduleReasonConfigmap
	RescheduleReasonCmNamespace = "mindx-dl"

	// JobRescheduleLabelKey key word of re-scheduling configuration
	JobRescheduleLabelKey = "fault-scheduling"
	// JobGraceRescheduleLabelValue Grace delete reschedule job, possible value of re-scheduling configuration
	JobGraceRescheduleLabelValue = "grace"
	// JobForceRescheduleLabelValue Force delete reschedule job, possible value of re-scheduling configuration
	JobForceRescheduleLabelValue = "force"
	// JobOffRescheduleLabelValue not delete reschedule job, possible value of re-scheduling configuration
	JobOffRescheduleLabelValue = "off"
	// GraceOverTimeKey for GraceOverTime config by user
	GraceOverTimeKey = "grace-over-time"
	// ElasticSchedulingKey for distinguishing whether a job is enabled with elastic scheduling
	ElasticSchedulingKey = "elastic-scheduling"
	// JobOnElasticScheduling job enabled with elastic scheduling
	JobOnElasticScheduling = "on"
	// JobOffElasticScheduling job not enabled with elastic scheduling
	JobOffElasticScheduling = "off"

	// CmFaultNodeKind key in configmap which saves the FaultNode cache
	CmFaultNodeKind = "fault-node"
	// CmFaultJob910bx2Kind key in configmap which saves the 910bx2 FaultJob cache
	CmFaultJob910bx2Kind = "fault-job-910bx2"
	// CmFaultJob910x8Kind key in configmap which saves the 910x8 FaultJob cache
	CmFaultJob910x8Kind = "fault-job-910x8"
	// CmJobRemainRetryTimes key in configmap which saves remain retry times of job
	CmJobRemainRetryTimes = "remain-retry-times"
	// MaxRescheduleRecordsNum the upper limit of the cm kept reschedule records, oldest record will be deleted
	// if record more than MaxRescheduleRecordsNum records
	MaxRescheduleRecordsNum = 10
	// MaxKbOfRescheduleRecords the upper limit words of the cm kept reschedule records
	MaxKbOfRescheduleRecords = 950 * 1024
	// CmJobRescheduleReasonsKey keeping recent MaxRescheduleRecordsNum records of rescheduling
	CmJobRescheduleReasonsKey = "recent-reschedule-records"
	// CmNodeRankTimeMapKind record map jobUID rankIndex node and times of occurrence
	CmNodeRankTimeMapKind = "node-rankIndex-Occurrence"
	// CmCheckCode Check code key
	CmCheckCode = "checkCode"

	// CmFaultJob key in configmap which saves the FaultJob cache
	CmFaultJob = "fault-job"

	// DefaultGraceOverTime time interval for grace delete
	DefaultGraceOverTime = 900

	// PublicFaultType represents a PublicFault fault type
	PublicFaultType = "PublicFault"
	// CardHealthy represents a healthy card
	CardHealthy = "Healthy"
	// CardUnhealthy represents an unhealthy card
	CardUnhealthy = "Unhealthy"
	// CardNetworkUnhealthy represents a network unhealthy card
	CardNetworkUnhealthy = "NetworkUnhealthy"
	// NodeHealthy represents node is available for scheduling
	NodeHealthy = "Healthy"
	// NodeUnhealthy represents node is unhealthy
	NodeUnhealthy = "NodeUnhealthy"
	// NodeCardUnhealthy represents node is unhealthy because of the card is unhealthy
	NodeCardUnhealthy = "CardUnhealthy"
	// NodeCardNetworkUnhealthy represents node is unhealthy because of card is network unhealthy
	NodeCardNetworkUnhealthy = "CardNetworkUnhealthy"
	// NoFaultJobsErr none fault jobs
	NoFaultJobsErr = "none fault jobs to be restarted in cache"
	// JobRecovery Name of cm for recovery
	JobRecovery = "job-recovery"
	// DeviceFaultCmKeySuffix the key of DeviceFault info
	DeviceFaultCmKeySuffix = "-Fault"
	// PodFailed the state of failed pod
	PodFailed = "pod-failed"
	// PodHealthy the state of healthy pod
	PodHealthy = "pod-healthy"

	// FaultRetryTimesKey key of fault-retry-times label
	FaultRetryTimesKey = "fault-retry-times"
)

const (
	// PreSeparateNPU fault type waiting user check
	PreSeparateNPU = "PreSeparateNPU"
	// NotHandleFault fault type not handle
	NotHandleFault = "NotHandleFault"
	// NodeFaultCode fault type nodeUnhealthy
	NodeFaultCode = "heartbeatTimeOut"
	// SubHealthFault subHealth code
	SubHealthFault = "SubHealthFault"
)

const (

	// SuperPodAnnoKey annotation key of super pod
	SuperPodAnnoKey = "sp-block"
)