k8s| SchedulingLatency 指标查看

1.1. metrics.go指标定义

pkg/scheduler/metrics/metrics.go

SchedulingLatency = prometheus.NewSummaryVec( prometheus.SummaryOpts{ Subsystem: SchedulerSubsystem, Name: SchedulingLatencyName, Help: "Scheduling latency in seconds split by sub-parts of the scheduling operation", // Make the sliding window of 5h. // TODO: The value for this should be based on some SLI definition (long term). MaxAge: 5 * time.Hour, }, []string{OperationLabel},

)

第64行：Scheduling latency in seconds split by sub-parts of the scheduling operation

一次调度被分割为子组件的操作时间的和

1.1. scheduler.go调度入口

1.1.1. pkg/scheduler/scheduler.go

// Run begins watching and scheduling. It waits for cache to be synced, then starts a goroutine and returns immediately. func (sched *Scheduler) Run() { if !sched.config.WaitForCacheSync() { return } go wait.Until(sched.scheduleOne, 0, sched.config.StopEverything) }

1.1.1. 核心调度逻辑顺序

1. sched.schedule

2. sched.assumeVolumes

3. sched.assume

4. sched.bindVolumes

5. sched.Cache

6. sched.bind

1.1. generic_scheduler.go调度算法实现

1.1.1. pkg/scheduler/core/generic_scheduler.go