Documentation
      ¶
    
    
  
    
  
    Index ¶
Constants ¶
This section is empty.
Variables ¶
      View Source
      
  
    var ( ErrNotFoundJobs = errors.New(`not found jobs under the assigned conditions.`) ErrTooManyJobs = errors.New(`found jobs more than one,please use --version or --type to filter.`) )
      View Source
      
  
    var KubeConfig string
    
      View Source
      
  
var SERVING_CHARTS = map[string]string{
	"tensorflow-serving-0.2.0":        "Tensorflow",
	"tensorrt-inference-server-0.0.1": "TensorRT",
}
    Functions ¶
func DisplayGPUShareDetails ¶ added in v0.3.1
func DisplayGPUShareDetails(nodeInfos []*GPUShareNodeInfo)
func DisplayGPUShareSummary ¶ added in v0.3.1
func DisplayGPUShareSummary(nodeInfos []*GPUShareNodeInfo)
func GetGPUMemoryInPod ¶ added in v0.3.1
Types ¶
type DeviceInfo ¶ added in v0.3.1
type DeviceInfo struct {
	Pods        []v1.Pod
	UsedGPUMem  int
	TotalGPUMem int
	// contains filtered or unexported fields
}
    func (*DeviceInfo) String ¶ added in v0.3.1
func (d *DeviceInfo) String() string
type GPUShareNodeInfo ¶ added in v0.3.1
type GPUShareNodeInfo struct {
	// contains filtered or unexported fields
}
    func BuildAllGPUShareNodeInfos ¶ added in v0.3.1
For all GPUShare nodes,decide whether the memory of GPU is measured by MiB or GiB
func BuildGPUShareNodeInfo ¶ added in v0.3.1
For one GPUShare node,decide whether the memory of GPU is measured by MiB or GiB
type Instance ¶
type Instance struct {
	// the status of of instance
	Status string `json:"status"`
	// the name of instance
	Name string `json:"name"`
	// the age of instance
	Age string `json:"age"`
	// the node instance runs on
	Node string `json:"node"`
	// the instance is chief or not
	IsChief bool `json:"chief" yaml:"chief"`
}
    type JobInfo ¶
type JobInfo struct {
	// The name of the training job
	Name string `json:"name"`
	// The namespace of the training job
	Namespace string `json:"namespace"`
	// The time of the training job
	Duration string `json:"duration"`
	// The status of the training Job
	Status JobStatus `json:"status"`
	// The training type of the training job
	Trainer string `json:"trainer"`
	// The tensorboard of the training job
	Tensorboard string `json:"tensorboard,omitempty"`
	// The name of the chief Instance
	ChiefName string `json:"chiefName" yaml:"chiefName"`
	// The instances under the training job
	Instances []Instance `json:"instances"`
	// The priority of the training job
	Priority string `json:"priority"`
}
    type JobStatus ¶
type JobStatus string
all the kinds of JobStatus
const ( // JobPending means the job is pending JobPending JobStatus = "PENDING" // JobRunning means the job is running JobRunning JobStatus = "RUNNING" // JobSucceeded means the job is Succeeded JobSucceeded JobStatus = "SUCCEEDED" // JobFailed means the job is failed JobFailed JobStatus = "FAILED" )
type ServingType ¶ added in v0.3.0
type ServingType string
const ( // tensorflow ServingTF ServingType = "TENSORFLOW" // tensorrt ServingTRT ServingType = "TENSORRT" // custom ServingCustom ServingType = "CUSTOM" //kfserving KFServing ServingType = "KFSERVING" )
three serving types.
type TrainingJobInfo ¶
 Click to show internal directories. 
   Click to hide internal directories.