func main() {
crawlQueue := list.New() // list of string
renderQueue := list.New() // list of string
processedURLs := list.New() // list of string
crawlResults := list.New() // list of CrawlEdge
renderResults := make(map[string]string)
seedUrl := flag.String("seed", "http://mesosphere.io", "The first URL to crawl")
master := flag.String("master", "127.0.1.1:5050", "Location of leading Mesos master")
localMode := flag.Bool("local", true, "If true, saves rendered web pages on local disk")
// TODO(nnielsen): Add flag for artifacts.
flag.Parse()
crawlQueue.PushBack(*seedUrl)
tasksCreated := 0
tasksRunning := 0
shuttingDown := false
crawlCommand := "python crawl_executor.py"
renderCommand := "python render_executor.py"
if *localMode {
renderCommand += " --local"
}
// TODO(nnielsen): In local mode, verify artifact locations.
rendlerArtifacts := executorURIs()
crawlExecutor := &mesos.ExecutorInfo{
ExecutorId: &mesos.ExecutorID{Value: proto.String("crawl-executor")},
Command: &mesos.CommandInfo{
Value: proto.String(crawlCommand),
Uris: rendlerArtifacts,
},
Name: proto.String("Crawler"),
}
renderExecutor := &mesos.ExecutorInfo{
ExecutorId: &mesos.ExecutorID{Value: proto.String("render-executor")},
Command: &mesos.CommandInfo{
Value: proto.String(renderCommand),
Uris: rendlerArtifacts,
},
Name: proto.String("Renderer"),
}
makeTaskPrototype := func(offer mesos.Offer) *mesos.TaskInfo {
taskId := tasksCreated
tasksCreated++
return &mesos.TaskInfo{
TaskId: &mesos.TaskID{
Value: proto.String(fmt.Sprintf("RENDLER-%d", taskId)),
},
SlaveId: offer.SlaveId,
Resources: []*mesos.Resource{
mesos.ScalarResource("cpus", TASK_CPUS),
mesos.ScalarResource("mem", TASK_MEM),
},
}
}
makeCrawlTask := func(url string, offer mesos.Offer) *mesos.TaskInfo {
task := makeTaskPrototype(offer)
task.Name = proto.String("CRAWL_" + *task.TaskId.Value)
task.Executor = crawlExecutor
task.Data = []byte(url)
return task
}
makeRenderTask := func(url string, offer mesos.Offer) *mesos.TaskInfo {
task := makeTaskPrototype(offer)
task.Name = proto.String("RENDER_" + *task.TaskId.Value)
task.Executor = renderExecutor
task.Data = []byte(url)
return task
}
maxTasksForOffer := func(offer mesos.Offer) int {
// TODO(nnielsen): Parse offer resources.
count := 0
var cpus float64 = 0
var mem float64 = 0
for _, resource := range offer.Resources {
if resource.GetName() == "cpus" {
cpus = *resource.GetScalar().Value
}
if resource.GetName() == "mem" {
mem = *resource.GetScalar().Value
}
}
for cpus >= TASK_CPUS && mem >= TASK_MEM {
count++
cpus -= TASK_CPUS
mem -= TASK_MEM
}
return count
}
printQueueStatistics := func() {
// TODO(nnielsen): Print queue lengths.
}
driver := mesos.SchedulerDriver{
Master: *master,
Framework: mesos.FrameworkInfo{
Name: proto.String("RENDLER"),
User: proto.String(""),
},
Scheduler: &mesos.Scheduler{
Registered: func(
driver *mesos.SchedulerDriver,
frameworkId mesos.FrameworkID,
masterInfo mesos.MasterInfo) {
log.Printf("Registered")
},
ResourceOffers: func(driver *mesos.SchedulerDriver, offers []mesos.Offer) {
printQueueStatistics()
for _, offer := range offers {
if shuttingDown {
fmt.Println("Shutting down: declining offer on [", offer.Hostname, "]")
driver.DeclineOffer(offer.Id)
continue
}
tasks := []mesos.TaskInfo{}
for i := 0; i < maxTasksForOffer(offer)/2; i++ {
if crawlQueue.Front() != nil {
url := crawlQueue.Front().Value.(string)
crawlQueue.Remove(crawlQueue.Front())
task := makeCrawlTask(url, offer)
tasks = append(tasks, *task)
}
if renderQueue.Front() != nil {
url := renderQueue.Front().Value.(string)
renderQueue.Remove(renderQueue.Front())
task := makeRenderTask(url, offer)
tasks = append(tasks, *task)
}
}
if len(tasks) == 0 {
driver.DeclineOffer(offer.Id)
} else {
driver.LaunchTasks(offer.Id, tasks)
}
}
},
StatusUpdate: func(driver *mesos.SchedulerDriver, status mesos.TaskStatus) {
log.Printf("Received task status [%s] for task [%s]", rendler.NameFor(status.State), *status.TaskId.Value)
if *status.State == mesos.TaskState_TASK_RUNNING {
tasksRunning++
} else if rendler.IsTerminal(status.State) {
tasksRunning--
}
},
FrameworkMessage: func(
driver *mesos.SchedulerDriver,
executorId mesos.ExecutorID,
slaveId mesos.SlaveID,
message string) {
switch *executorId.Value {
case *crawlExecutor.ExecutorId.Value:
log.Print("Received framework message from crawler")
var result rendler.CrawlResult
err := json.Unmarshal([]byte(message), &result)
if err != nil {
log.Printf("Error deserializing CrawlResult: [%s]", err)
} else {
for _, link := range result.Links {
edge := rendler.Edge{From: result.URL, To: link}
log.Printf("Appending [%s] to crawl results", edge)
crawlResults.PushBack(edge)
alreadyProcessed := false
for e := processedURLs.Front(); e != nil && !alreadyProcessed; e = e.Next() {
processedURL := e.Value.(string)
if link == processedURL {
alreadyProcessed = true
}
}
if !alreadyProcessed {
log.Printf("Enqueueing [%s]", link)
crawlQueue.PushBack(link)
renderQueue.PushBack(link)
processedURLs.PushBack(link)
}
}
}
case *renderExecutor.ExecutorId.Value:
log.Printf("Received framework message from renderer")
var result rendler.RenderResult
err := json.Unmarshal([]byte(message), &result)
if err != nil {
log.Printf("Error deserializing RenderResult: [%s]", err)
} else {
log.Printf(
"Appending [%s] to render results",
rendler.Edge{From: result.URL, To: result.ImageURL})
renderResults[result.URL] = result.ImageURL
}
default:
log.Printf("Received a framework message from some unknown source: %s", *executorId.Value)
}
},
},
}
c := make(chan os.Signal, 1)
signal.Notify(c, os.Interrupt, os.Kill)
go func(c chan os.Signal) {
s := <-c
fmt.Println("Got signal:", s)
if s == os.Interrupt {
fmt.Println("RENDLER is shutting down")
shuttingDown = true
wait_started := time.Now()
for tasksRunning > 0 && SHUTDOWN_TIMEOUT > int(time.Since(wait_started).Seconds()) {
time.Sleep(time.Second)
}
if tasksRunning > 0 {
fmt.Println("Shutdown by timeout,", tasksRunning, "task(s) have not completed")
}
driver.Stop(false)
}
}(c)
driver.Init()
defer driver.Destroy()
driver.Start()
driver.Join()
driver.Stop(false)
rendler.WriteDOTFile(crawlResults, renderResults)
os.Exit(0)
}
func main() {
taskId := 0
localExecutor, _ := executorPath()
goPath := os.Getenv("GOPATH") + "/"
master := flag.String("master", "localhost:5050", "Location of leading Mesos master")
executorPath := flag.String("executor-uri", localExecutor, "Path to executor executable")
address := flag.String("address", "localhost", "Hostname to serve artifacts from")
angstromPath := flag.String("angstrom-path", goPath+"src/github.com/nqn/angstrom", "Path to angstrom checkout")
flag.Parse()
serveExecutorArtifact := func(path string) string {
serveFile := func(pattern string, filename string) {
http.HandleFunc(pattern, func(w http.ResponseWriter, r *http.Request) {
http.ServeFile(w, r, filename)
})
}
// Create base path (http://foobar:5000/<base>)
pathSplit := strings.Split(path, "/")
var base string
if len(pathSplit) > 0 {
base = pathSplit[len(pathSplit)-1]
} else {
base = path
}
serveFile("/"+base, path)
hostURI := fmt.Sprintf("http://%s:%d/%s", *address, defaultPort, base)
glog.V(2).Infof("Serving '%s'\n", hostURI)
return hostURI
}
executorURI := serveExecutorArtifact(*executorPath)
executable := true
executor := &mesos.ExecutorInfo{
ExecutorId: &mesos.ExecutorID{Value: proto.String("default")},
Command: &mesos.CommandInfo{
Value: proto.String("./angstrom-executor"),
Uris: []*mesos.CommandInfo_URI{
&mesos.CommandInfo_URI{Value: &executorURI, Executable: &executable},
},
},
Name: proto.String("Angstrom Executor"),
Source: proto.String("angstrom"),
}
cluster := acluster.NewCluster(*master)
cluster.Update()
// Keep updating cluster state
go func() {
for {
cluster.Update()
time.Sleep(updateInterval)
}
}()
slaves := list.New()
for _, slave := range cluster.Sample.Slaves {
slaveHostname := slave.Hostname + ":" + strconv.Itoa(slave.Port)
slaves.PushBack(slaveHostname)
}
// TODO(nnielsen): Move this and callbacks to dedicated scheduler package / struct.
taskToSlave := make(map[string]string)
scheduleTask := func(offer mesos.Offer) *mesos.TaskInfo {
slave := slaves.Front()
if slave == nil {
return nil
}
glog.V(2).Infof("Scheduling slave '%s'", slave.Value.(string))
slaves.Remove(slave)
task := "angstrom-task-" + strconv.Itoa(taskId)
taskToSlave[task] = slave.Value.(string)
return &mesos.TaskInfo{
Name: proto.String("angstrom-task"),
TaskId: &mesos.TaskID{
Value: proto.String(task),
},
SlaveId: offer.SlaveId,
Executor: executor,
Data: []byte("{\"slave\": \"" + slave.Value.(string) + "\"}"),
Resources: []*mesos.Resource{
mesos.ScalarResource("cpus", 0.5),
mesos.ScalarResource("mem", 32),
},
}
}
driver := mesos.SchedulerDriver{
Master: *master,
Framework: mesos.FrameworkInfo{
Name: proto.String("Angstrom metrics"),
User: proto.String(""),
},
Scheduler: &mesos.Scheduler{
ResourceOffers: func(driver *mesos.SchedulerDriver, offers []mesos.Offer) {
for _, offer := range offers {
taskId++
tasks := make([]mesos.TaskInfo, 0)
task := scheduleTask(offer)
if task != nil {
glog.V(2).Infof("Launching on offer %v", offer.Id)
tasks = append(tasks, *task)
driver.LaunchTasks(offer.Id, tasks)
} else {
glog.V(2).Infof("Declining offer %v", offer.Id)
driver.DeclineOffer(offer.Id)
}
}
},
FrameworkMessage: func(driver *mesos.SchedulerDriver, _executorId mesos.ExecutorID, slaveId mesos.SlaveID, data string) {
var target []payload.StatisticsInfo
err := json.Unmarshal([]byte(data), &target)
if err != nil {
return
}
cluster.AddSlaveSamples(slaveId, target)
},
StatusUpdate: func(driver *mesos.SchedulerDriver, status mesos.TaskStatus) {
// TODO(nnielsen): Readd slave task to queue in case of any terminal state.
if *status.State == mesos.TaskState_TASK_RUNNING {
} else if *status.State == mesos.TaskState_TASK_FINISHED {
}
},
},
}
driver.Init()
defer driver.Destroy()
driver.Start()
endpoints.Initialize(defaultPort, *angstromPath, cluster)
glog.V(2).Infof("Waiting for threads to join")
driver.Join()
}