// Copyright (C) 2018 spdfg
//
// This file is part of Elektron.
//
// Elektron is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// Elektron is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with Elektron. If not, see .
//
package schedulers
import (
"bytes"
"fmt"
"sync"
"time"
"github.com/golang/protobuf/proto"
mesos "github.com/mesos/mesos-go/api/v0/mesosproto"
"github.com/mesos/mesos-go/api/v0/mesosutil"
sched "github.com/mesos/mesos-go/api/v0/scheduler"
elekLog "github.com/sirupsen/logrus"
"github.com/spdfg/elektron/def"
"github.com/spdfg/elektron/elektronLogging"
elekLogTypes "github.com/spdfg/elektron/elektronLogging/types"
"github.com/spdfg/elektron/utilities"
"github.com/spdfg/elektron/utilities/schedUtils"
)
type BaseScheduler struct {
ElectronScheduler
SchedPolicyContext
// Current scheduling policy used for resource offer consumption.
curSchedPolicy SchedPolicyState
tasksCreated int
tasksRunning int
tasks []def.Task
metrics map[string]def.Metric
Running map[string]map[string]bool
wattsAsAResource bool
classMapWatts bool
TasksRunningMutex sync.Mutex
HostNameToSlaveID map[string]string
totalResourceAvailabilityRecorded bool
// First set of PCP values are garbage values, signal to logger to start recording when we're
// about to schedule a new task
RecordPCP *bool
// This channel is closed when the program receives an interrupt,
// signalling that the program should shut down.
Shutdown chan struct{}
// This channel is closed after shutdown is closed, and only when all
// outstanding tasks have been cleaned up.
Done chan struct{}
// Controls when to shutdown pcp logging.
PCPLog chan struct{}
schedTrace *elekLog.Logger
mutex sync.Mutex
// Whether switching of scheduling policies at runtime has been enabled
schedPolSwitchEnabled bool
// Name of the first scheduling policy to be deployed, if provided.
// This scheduling policy would be deployed first regardless of the distribution of tasks in the TaskQueue.
// Note: Scheduling policy switching needs to be enabled.
nameOfFstSchedPolToDeploy string
// Scheduling policy switching criteria.
schedPolSwitchCriteria string
// Size of window of tasks that can be scheduled in the next offer cycle.
// The window size can be adjusted to make the most use of every resource offer.
// By default, the schedWindowSize would correspond to the number of remaining tasks that haven't yet
// been scheduled.
schedWindowSize int
// Number of tasks in the window
numTasksInSchedWindow int
// Whether the scheduling window needs to be fixed.
toFixSchedWindow bool // If yes, then schedWindowSize is initialized and kept unchanged.
// Strategy to resize the schedulingWindow.
schedWindowResStrategy schedUtils.SchedWindowResizingStrategy
// Indicate whether the any resource offers from mesos have been received.
hasReceivedResourceOffers bool
}
func (s *BaseScheduler) init(opts ...SchedulerOptions) {
for _, opt := range opts {
// applying options
if err := opt(s); err != nil {
elekLog.Fatal(err)
}
}
s.TasksRunningMutex.Lock()
s.Running = make(map[string]map[string]bool)
s.TasksRunningMutex.Unlock()
s.HostNameToSlaveID = make(map[string]string)
s.mutex = sync.Mutex{}
s.schedWindowResStrategy = schedUtils.SchedWindowResizingCritToStrategy["fillNextOfferCycle"]
// Initially no resource offers would have been received.
s.hasReceivedResourceOffers = false
}
func (s *BaseScheduler) SwitchSchedPol(newSchedPol SchedPolicyState) {
s.curSchedPolicy = newSchedPol
}
func (s *BaseScheduler) newTask(offer *mesos.Offer, task def.Task) *mesos.TaskInfo {
taskName := fmt.Sprintf("%s-%d", task.Name, *task.Instances)
s.tasksCreated++
if !*s.RecordPCP {
// Turn on elecLogDef
*s.RecordPCP = true
time.Sleep(1 * time.Second) // Make sure we're recording by the time the first task starts
}
resources := []*mesos.Resource{
mesosutil.NewScalarResource("cpus", task.CPU),
mesosutil.NewScalarResource("mem", task.RAM),
}
if s.wattsAsAResource {
if wattsToConsider, err := def.WattsToConsider(task, s.classMapWatts, offer); err == nil {
s.LogTaskWattsConsideration(task, *offer.Hostname, wattsToConsider)
resources = append(resources, mesosutil.NewScalarResource("watts", wattsToConsider))
} else {
// Error in determining wattsConsideration
s.LogElectronError(err)
}
}
return &mesos.TaskInfo{
Name: proto.String(taskName),
TaskId: &mesos.TaskID{
Value: proto.String("electron-" + taskName),
},
SlaveId: offer.SlaveId,
Resources: resources,
Command: &mesos.CommandInfo{
Value: proto.String(task.CMD),
},
Container: &mesos.ContainerInfo{
Type: mesos.ContainerInfo_DOCKER.Enum(),
Docker: &mesos.ContainerInfo_DockerInfo{
Image: proto.String(task.Image),
Network: mesos.ContainerInfo_DockerInfo_BRIDGE.Enum(), // Run everything isolated
},
},
}
}
func (s *BaseScheduler) OfferRescinded(_ sched.SchedulerDriver, offerID *mesos.OfferID) {
s.LogOfferRescinded(offerID)
}
func (s *BaseScheduler) SlaveLost(_ sched.SchedulerDriver, slaveID *mesos.SlaveID) {
s.LogSlaveLost(slaveID)
}
func (s *BaseScheduler) ExecutorLost(_ sched.SchedulerDriver, executorID *mesos.ExecutorID,
slaveID *mesos.SlaveID, status int) {
s.LogExecutorLost(executorID, slaveID)
}
func (s *BaseScheduler) Error(_ sched.SchedulerDriver, err string) {
s.LogMesosError(err)
}
func (s *BaseScheduler) FrameworkMessage(
driver sched.SchedulerDriver,
executorID *mesos.ExecutorID,
slaveID *mesos.SlaveID,
message string) {
s.LogFrameworkMessage(executorID, slaveID, message)
}
func (s *BaseScheduler) Registered(
_ sched.SchedulerDriver,
frameworkID *mesos.FrameworkID,
masterInfo *mesos.MasterInfo) {
s.LogFrameworkRegistered(frameworkID, masterInfo)
}
func (s *BaseScheduler) Reregistered(_ sched.SchedulerDriver, masterInfo *mesos.MasterInfo) {
s.LogFrameworkReregistered(masterInfo)
}
func (s *BaseScheduler) Disconnected(sched.SchedulerDriver) {
s.LogDisconnected()
}
func (s *BaseScheduler) ResourceOffers(driver sched.SchedulerDriver, offers []*mesos.Offer) {
// Recording the total amount of resources available across the cluster.
utilities.RecordTotalResourceAvailability(offers)
for _, offer := range offers {
if _, ok := s.HostNameToSlaveID[offer.GetHostname()]; !ok {
s.HostNameToSlaveID[offer.GetHostname()] = *offer.SlaveId.Value
}
}
// Switch just before consuming the resource offers.
s.curSchedPolicy.SwitchIfNecessary(s)
// s.Log(elecLogDef.GENERAL, fmt.Sprintf("SchedWindowSize[%d], #TasksInWindow[%d]",
// s.schedWindowSize, s.numTasksInSchedWindow))
s.curSchedPolicy.ConsumeOffers(s, driver, offers)
s.hasReceivedResourceOffers = true
}
func (s *BaseScheduler) StatusUpdate(driver sched.SchedulerDriver, status *mesos.TaskStatus) {
s.LogTaskStatusUpdate(status)
if *status.State == mesos.TaskState_TASK_RUNNING {
// If this is our first time running into this Agent
s.TasksRunningMutex.Lock()
if _, ok := s.Running[*status.SlaveId.Value]; !ok {
s.Running[*status.SlaveId.Value] = make(map[string]bool)
}
// Add task to list of tasks running on node
s.Running[*status.SlaveId.Value][*status.TaskId.Value] = true
s.tasksRunning++
s.TasksRunningMutex.Unlock()
} else if IsTerminal(status.State) {
// Update resource availability.
utilities.ResourceAvailabilityUpdate("ON_TASK_TERMINAL_STATE",
*status.TaskId, *status.SlaveId)
s.TasksRunningMutex.Lock()
delete(s.Running[*status.SlaveId.Value], *status.TaskId.Value)
s.tasksRunning--
s.TasksRunningMutex.Unlock()
if s.tasksRunning == 0 {
select {
case <-s.Shutdown:
close(s.Done)
default:
}
}
}
}
func (s *BaseScheduler) LogTaskStarting(ts *def.Task, offer *mesos.Offer) {
lmt := elekLogTypes.GENERAL
if ts == nil {
elektronLogging.ElektronLog.Log(lmt, elekLog.InfoLevel,
elekLog.Fields{"host": fmt.Sprintf("%s", offer.GetHostname())}, "TASKS STARTING...")
} else {
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"task": fmt.Sprintf("%s", ts.Name),
"Instance": fmt.Sprintf("%d", *ts.Instances), "host": fmt.Sprintf("%s", offer.GetHostname())},
"TASK STARTING... ")
}
}
func (s *BaseScheduler) LogTaskWattsConsideration(ts def.Task, host string, wattsToConsider float64) {
lmt := elekLogTypes.GENERAL
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"task": ts.Name, "host": host, "Watts": fmt.Sprintf("%f", wattsToConsider)}, "Watts considered for ")
}
func (s *BaseScheduler) LogOffersReceived(offers []*mesos.Offer) {
lmt := elekLogTypes.GENERAL
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"Resource offers received": fmt.Sprintf("%d", len(offers))}, "")
}
func (s *BaseScheduler) LogNoPendingTasksDeclineOffers(offer *mesos.Offer) {
lmt := elekLogTypes.WARNING
elektronLogging.ElektronLog.Log(lmt,
elekLog.WarnLevel,
elekLog.Fields{"DECLINING OFFER for host": fmt.Sprintf("%s", offer.GetHostname())}, "No tasks left to schedule ")
}
func (s *BaseScheduler) LogNumberOfRunningTasks() {
lmt := elekLogTypes.GENERAL
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"Number of tasks still Running": fmt.Sprintf("%d", s.tasksRunning)}, "")
}
func (s *BaseScheduler) LogCoLocatedTasks(slaveID string) {
lmt := elekLogTypes.GENERAL
buffer := bytes.Buffer{}
s.TasksRunningMutex.Lock()
for taskName := range s.Running[slaveID] {
buffer.WriteString(fmt.Sprintln(taskName))
}
s.TasksRunningMutex.Unlock()
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"Colocated with": fmt.Sprintf("%s", buffer.String())}, "")
}
func (s *BaseScheduler) LogSchedTrace(taskToSchedule *mesos.TaskInfo, offer *mesos.Offer) {
elektronLogging.ElektronLog.Log(elekLogTypes.SCHED_TRACE,
elekLog.InfoLevel,
elekLog.Fields{offer.GetHostname(): fmt.Sprintf("%s", taskToSchedule.GetTaskId().GetValue())}, "")
}
func (s *BaseScheduler) LogTerminateScheduler() {
lmt := elekLogTypes.GENERAL
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{}, "Done scheduling all tasks!")
}
func (s *BaseScheduler) LogInsufficientResourcesDeclineOffer(offer *mesos.Offer,
offerResources ...interface{}) {
lmt := elekLogTypes.WARNING
buffer := bytes.Buffer{}
buffer.WriteString(fmt.Sprintf("", offerResources...))
elektronLogging.ElektronLog.Log(lmt,
elekLog.WarnLevel,
elekLog.Fields{"Offer Resources": fmt.Sprintf("%s", buffer.String())}, "DECLINING OFFER... Offer has insufficient resources to launch a task")
}
func (s *BaseScheduler) LogOfferRescinded(offerID *mesos.OfferID) {
lmt := elekLogTypes.ERROR
elektronLogging.ElektronLog.Log(lmt,
elekLog.ErrorLevel,
elekLog.Fields{"OfferID": fmt.Sprintf("%s", offerID)}, "OFFER RESCINDED")
}
func (s *BaseScheduler) LogSlaveLost(slaveID *mesos.SlaveID) {
lmt := elekLogTypes.ERROR
elektronLogging.ElektronLog.Log(lmt,
elekLog.ErrorLevel,
elekLog.Fields{"SlaveID": fmt.Sprintf("%s", slaveID)}, "SLAVE LOST")
}
func (s *BaseScheduler) LogExecutorLost(executorID *mesos.ExecutorID, slaveID *mesos.SlaveID) {
lmt := elekLogTypes.ERROR
elektronLogging.ElektronLog.Log(lmt,
elekLog.ErrorLevel,
elekLog.Fields{"ExecutorID": fmt.Sprintf("%s", executorID), "SlaveID": fmt.Sprintf("%s", slaveID)}, "EXECUTOR LOST")
}
func (s *BaseScheduler) LogFrameworkMessage(executorID *mesos.ExecutorID,
slaveID *mesos.SlaveID, message string) {
lmt := elekLogTypes.GENERAL
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"Received Framework message from executor": executorID}, message)
}
func (s *BaseScheduler) LogMesosError(err string) {
lmt := elekLogTypes.ERROR
elektronLogging.ElektronLog.Log(lmt,
elekLog.ErrorLevel,
elekLog.Fields{"MESOS ERROR": fmt.Sprintf("%v", err)}, "")
}
func (s *BaseScheduler) LogElectronError(err error) {
lmt := elekLogTypes.ERROR
elektronLogging.ElektronLog.Log(lmt,
elekLog.ErrorLevel,
elekLog.Fields{"ELECTRON ERROR": fmt.Sprintf("%v", err)}, "")
}
func (s *BaseScheduler) LogFrameworkRegistered(frameworkID *mesos.FrameworkID,
masterInfo *mesos.MasterInfo) {
lmt := elekLogTypes.SUCCESS
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"frameworkID": fmt.Sprintf("%s", frameworkID), "master": fmt.Sprintf("%s", masterInfo)}, "FRAMEWORK REGISTERED!")
}
func (s *BaseScheduler) LogFrameworkReregistered(masterInfo *mesos.MasterInfo) {
lmt := elekLogTypes.GENERAL
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"master": fmt.Sprintf("%s", masterInfo)}, "Framework re-registered")
}
func (s *BaseScheduler) LogDisconnected() {
lmt := elekLogTypes.WARNING
elektronLogging.ElektronLog.Log(lmt,
elekLog.WarnLevel,
elekLog.Fields{}, "Framework disconnected with master")
}
func (s *BaseScheduler) LogTaskStatusUpdate(status *mesos.TaskStatus) {
lmt := elekLogTypes.GENERAL
switch *status.State {
case mesos.TaskState_TASK_ERROR, mesos.TaskState_TASK_FAILED,
mesos.TaskState_TASK_KILLED, mesos.TaskState_TASK_LOST:
lmt = elekLogTypes.ERROR
case mesos.TaskState_TASK_FINISHED:
lmt = elekLogTypes.SUCCESS
default:
lmt = elekLogTypes.GENERAL
}
elektronLogging.ElektronLog.Log(lmt,
elekLog.InfoLevel,
elekLog.Fields{"task": fmt.Sprintf("%s", *status.TaskId.Value), "state": NameFor(status.State)}, "Task Status received")
}
func (s *BaseScheduler) LogSchedPolicySwitch(name string, nextPolicy SchedPolicyState) {
logSPS := func() {
elektronLogging.ElektronLog.Log(elekLogTypes.SPS,
elekLog.InfoLevel,
elekLog.Fields{"Name": name}, "")
}
if s.hasReceivedResourceOffers && (s.curSchedPolicy != nextPolicy) {
logSPS()
} else if !s.hasReceivedResourceOffers {
logSPS()
}
// Logging the size of the scheduling window and the scheduling policy
// that is going to schedule the tasks in the scheduling window.
elektronLogging.ElektronLog.Log(elekLogTypes.SCHED_WINDOW,
elekLog.InfoLevel,
elekLog.Fields{"Window size": fmt.Sprintf("%d", s.schedWindowSize), "Name": name}, "")
}
func (s *BaseScheduler) LogClsfnAndTaskDistOverhead(overhead time.Duration) {
// Logging the overhead in microseconds.
elektronLogging.ElektronLog.Log(elekLogTypes.CLSFN_TASKDIST_OVERHEAD,
elekLog.InfoLevel,
elekLog.Fields{"Overhead in microseconds": fmt.Sprintf("%f", float64(overhead.Nanoseconds())/1000.0)}, "")
}