云计算下高效资源调度算法Python 开源云计算调度框架

转载

mob6454cc667b1d 2024-05-07 22:32:38

文章标签 云计算下高效资源调度算法Python 开源 java 服务器 System 文章分类 Python 后端开发

项目介绍

PowerJob（原OhMyScheduler）是全新一代分布式调度与计算框架，能让您轻松完成作业的调度与繁杂任务的分布式计算。

云计算下高效资源调度算法Python 开源云计算调度框架_开源

项目地址

源码：https://gitee.com/KFCFans/PowerJob
官网：http://www.powerjob.tech/index.html

快速入门

https://www.yuque.com/powerjob/guidence/nyio9g 云计算下高效资源调度算法Python 开源云计算调度框架_云计算下高效资源调度算法Python_02

源码解析

服务端启动

PowerJobServerApplication启动。

public static void main(String[] args) {

        pre();

        AkkaStarter.init();
        VertXStarter.init();

        // Start SpringBoot application.
        try {
            SpringApplication.run(PowerJobServerApplication.class, args);
        } catch (Throwable t) {
            log.error(TIPS);
            throw t;
        }
    }

AkkaStarter.init();，启动actorSystem，用FriendRequestHandler作为消息的处理器。加载配置oms-server.akka.conf。服务端口号设置为10086。

public static void init() {

        Stopwatch stopwatch = Stopwatch.createStarted();
        log.info("[PowerJob] PowerJob's akka system start to bootstrap...");

        // 忽略了一个问题，机器是没办法访问外网的，除非架设自己的NTP服务器
        // TimeUtils.check();

        // 解析配置文件
        Config akkaFinalConfig = parseConfig();
        actorSystem = ActorSystem.create(RemoteConstant.SERVER_ACTOR_SYSTEM_NAME, akkaFinalConfig);
        actorSystem.actorOf(FriendRequestHandler.defaultProps(), RemoteConstant.SERVER_FRIEND_ACTOR_NAME);
        log.info("[PowerJob] PowerJob's akka system started successfully, using time {}.", stopwatch);
    }

VertXStarter.init();主要是新建了Vertx对象，Vertx.vertx();，端口号设置成10010。
在Initializer#initHandler设置了Vertx处理器WorkerRequestHttpHandler和Akka的消息处理器。

@Component
@ConditionalOnExpression("'${execution.env}'!='test'")
public class Initializer {

    @PostConstruct
    public void initHandler() {
        // init akka
        AkkaStarter.actorSystem.actorOf(WorkerRequestAkkaHandler.defaultProps(), RemoteConstant.SERVER_ACTOR_NAME);
        // init vert.x
        VertXStarter.vertx.deployVerticle(new WorkerRequestHttpHandler());
    }
}

WorkerRequestHttpHandler是创建了一个HttpServer，并且设置路由。

@Slf4j
public class WorkerRequestHttpHandler extends AbstractVerticle {

    @Override
    public void start() throws Exception {

        Properties properties = PropertyUtils.getProperties();
        int port = Integer.parseInt(properties.getProperty(PowerJobServerConfigKey.HTTP_PORT, String.valueOf(OmsConstant.SERVER_DEFAULT_HTTP_PORT)));

        HttpServerOptions options = new HttpServerOptions();
        HttpServer server = vertx.createHttpServer(options);

        Router router = Router.router(vertx);
        router.route().handler(BodyHandler.create());
        router.post(ProtocolConstant.SERVER_PATH_HEARTBEAT)
                .handler(ctx -> {
                    WorkerHeartbeat heartbeat = ctx.getBodyAsJson().mapTo(WorkerHeartbeat.class);
                    fetchWorkerRequestHandler().processWorkerHeartbeat(heartbeat);
                    success(ctx);
                });
        router.post(ProtocolConstant.SERVER_PATH_STATUS_REPORT)
                .blockingHandler(ctx -> {
                    TaskTrackerReportInstanceStatusReq req = ctx.getBodyAsJson().mapTo(TaskTrackerReportInstanceStatusReq.class);
                    try {
                        fetchWorkerRequestHandler().processTaskTrackerReportInstanceStatus(req);
                        out(ctx, AskResponse.succeed(null));
                    } catch (Exception e) {
                        log.error("[WorkerRequestHttpHandler] update instance status failed for request: {}.", req, e);
                        out(ctx, AskResponse.failed(ExceptionUtils.getMessage(e)));
                    }
                });
        router.post(ProtocolConstant.SERVER_PATH_LOG_REPORT)
                .blockingHandler(ctx -> {
                    WorkerLogReportReq req = ctx.getBodyAsJson().mapTo(WorkerLogReportReq.class);
                    fetchWorkerRequestHandler().processWorkerLogReport(req);
                    success(ctx);
                });
        server.requestHandler(router).listen(port);
    }

    private static void out(RoutingContext ctx, Object msg) {
        ctx.response()
                .putHeader(OmsConstant.HTTP_HEADER_CONTENT_TYPE, OmsConstant.JSON_MEDIA_TYPE)
                .end(JsonObject.mapFrom(msg).encode());
    }

    private static void success(RoutingContext ctx) {
        out(ctx, ResultDTO.success(null));
    }
}

客户端启动

PowerJobWorker实现了InitializingBean，执行方法PowerJobWorker#init。该方法中会连接服务器

ServerDiscoveryService serverDiscoveryService = new ServerDiscoveryService(workerRuntime.getAppId(), workerRuntime.getWorkerConfig());
            serverDiscoveryService.start(timingPool);

启动服务，ServerDiscoveryService#start。进行了服务的发现，在ServerDiscoveryService#acquire方法中调用http://127.0.0.1:7700/server/acquire?appId=2&currentServer=null&protocol=AKKA，找到服务器地址10.132.17.10:10086。
客户端会也会启动akka服务，加载oms-worker.akka.conf的配置，设置端口号27777。

客户端-服务发现

客户端发起http://127.0.0.1:7700/server/acquire?appId=2&currentServer=null&protocol=AKKA接口获取当前akka的server的地址。
服务端响应，ServerController#acquireServer。服务端设置Ping请求信息，访问Ping接口，path地址：akka://oms-server@10.132.17.10:10086/user/friend_actor，调用成功则说明地址是ok的。

private String activeAddress(String serverAddress, Set<String> downServerCache, String protocol) {

        if (downServerCache.contains(serverAddress)) {
            return null;
        }
        if (StringUtils.isEmpty(serverAddress)) {
            return null;
        }

        Ping ping = new Ping();
        ping.setCurrentTime(System.currentTimeMillis());

        ActorSelection serverActor = AkkaStarter.getFriendActor(serverAddress);
        try {
            CompletionStage<Object> askCS = Patterns.ask(serverActor, ping, Duration.ofMillis(PING_TIMEOUT_MS));
            AskResponse response = (AskResponse) askCS.toCompletableFuture().get(PING_TIMEOUT_MS, TimeUnit.MILLISECONDS);
            downServerCache.remove(serverAddress);
            if (response.isSuccess()) {
                return JsonUtils.parseObject(response.getData(), JSONObject.class).getString(protocol);
            }
        }catch (Exception e) {
            log.warn("[ServerElection] server({}) was down.", serverAddress);
        }
        downServerCache.add(serverAddress);
        return null;
    }

public static ActorSelection getFriendActor(String address) {
        String path = String.format(AKKA_PATH, RemoteConstant.SERVER_ACTOR_SYSTEM_NAME, address, RemoteConstant.SERVER_FRIEND_ACTOR_NAME);
        return actorSystem.actorSelection(path);
    }

FriendRequestHandler#onReceivePing会处理传到FriendRequestHandler里面的Ping请求。

private void onReceivePing(Ping ping) {
        getSender().tell(AskResponse.succeed(TransportService.getAllAddress()), getSelf());
    }

服务端-执行任务

入口是JobController#runImmediately，先过切面DesignateServerAspect#execute，执行InstanceService#create进行任务实例创建。经过切面UseCacheLockAspect#execute进行分发DispatchService#dispatch。
如果可以找到合适的worker线程，则构造请求实体，发送请求；如果找不到worker，则InstanceManager#processFinishedInstance，完成或者失败，都需要一些处理。比如日志打印，告警。

// 获取当前最合适的 worker 列表
        List<WorkerInfo> suitableWorkers = workerClusterQueryService.getSuitableWorkers(jobInfo);

        if (CollectionUtils.isEmpty(suitableWorkers)) {
            log.warn("[Dispatcher-{}|{}] cancel dispatch job due to no worker available", jobId, instanceId);
            instanceInfoRepository.update4TriggerFailed(instanceId, FAILED.getV(), current, current, RemoteConstant.EMPTY_ADDRESS, SystemInstanceResult.NO_WORKER_AVAILABLE, now);

            instanceManager.processFinishedInstance(instanceId, instanceInfo.getWfInstanceId(), FAILED, SystemInstanceResult.NO_WORKER_AVAILABLE);
            return;
        }
		List<String> workerIpList = suitableWorkers.stream().map(WorkerInfo::getAddress).collect(Collectors.toList());

        // 构造任务调度请求
        ServerScheduleJobReq req = constructServerScheduleJobReq(jobInfo, instanceInfo, workerIpList);

        // 发送请求（不可靠，需要一个后台线程定期轮询状态）
        WorkerInfo taskTracker = suitableWorkers.get(0);
        String taskTrackerAddress = taskTracker.getAddress();

        transportService.tell(Protocol.of(taskTracker.getProtocol()), taskTrackerAddress, req);
        log.info("[Dispatcher-{}|{}] send schedule request to TaskTracker[protocol:{},address:{}] successfully: {}.", jobId, instanceId, taskTracker.getProtocol(), taskTrackerAddress, req);

        // 修改状态
        instanceInfoRepository.update4TriggerSucceed(instanceId, WAITING_WORKER_RECEIVE.getV(), current, taskTrackerAddress, now);

        // 装载缓存
        instanceMetadataService.loadJobInfo(instanceId, jobInfo);

因为是akka协议，所以用到了AkkaTransporter。拼接客户端地址akka://oms@10.132.17.10:27777/user/worker，调用服务。构造的对象是ServerScheduleJobReq。

public void tell(String address, PowerSerializable object) {
        ActorSelection taskTrackerActor = AkkaStarter.getWorkerActor(address);
        taskTrackerActor.tell(object, null);
    }

客户端接受到请求，TaskTrackerActor#onReceiveServerScheduleJobReq，会初始化CommonTaskTracker#initTaskTracker，开启定时任务执行Dispatcher。分发任务，会TaskTracker#dispatchTask，用到了akka框架，发送TaskTrackerStartTaskReq对象，用的处理器是processor_tracker。
ProcessorTrackerActor#onReceiveTaskTrackerStartTaskReq，会将任务信息封装成ProcessorRunnable，找到对应的BasicProcessor进行处理。

ClassLoader classLoader = omsContainer == null ? getClass().getClassLoader() : omsContainer.getContainerClassLoader();
        ProcessorRunnable processorRunnable = new ProcessorRunnable(instanceInfo, taskTrackerActorRef, newTask, processor, omsLogger, classLoader, statusReportRetryQueue, workerRuntime);
        try {
            threadPool.submit(processorRunnable);
            success = true;
        } catch (RejectedExecutionException ignore) {
            log.warn("[ProcessorTracker-{}] submit task(taskId={},taskName={}) to ThreadPool failed due to ThreadPool has too much task waiting to process, this task will dispatch to other ProcessorTracker.",
                    instanceId, newTask.getTaskId(), newTask.getTaskName());
        } catch (Exception e) {
            log.error("[ProcessorTracker-{}] submit task(taskId={},taskName={}) to ThreadPool failed.", instanceId, newTask.getTaskId(), newTask.getTaskName(), e);
        }

调用成功之后，会发送ProcessorReportTaskStatusReq请求。

// 2. 回复接收成功
        if (success) {
            ProcessorReportTaskStatusReq reportReq = new ProcessorReportTaskStatusReq();
            reportReq.setInstanceId(instanceId);
            reportReq.setSubInstanceId(newTask.getSubInstanceId());
            reportReq.setTaskId(newTask.getTaskId());
            reportReq.setStatus(TaskStatus.WORKER_RECEIVED.getValue());
            reportReq.setReportTime(System.currentTimeMillis());

            taskTrackerActorRef.tell(reportReq, null);

            log.debug("[ProcessorTracker-{}] submit task(taskId={}, taskName={}) success, current queue size: {}.",
                    instanceId, newTask.getTaskId(), newTask.getTaskName(), threadPool.getQueue().size());
        }

客户端-发送心跳

PowerJobWorker实现了InitializingBean，在init方法中会执行timingPool.scheduleAtFixedRate(new WorkerHealthReporter(workerRuntime), 0, 15, TimeUnit.SECONDS);。获取服务端的接口地址，akka://oms-server@192.168.1.5:10086/user/server_actor，发送心跳

// 发送请求
        String serverPath = AkkaUtils.getServerActorPath(currentServer);
        if (StringUtils.isEmpty(serverPath)) {
            return;
        }
        ActorSelection actorSelection = workerRuntime.getActorSystem().actorSelection(serverPath);
        actorSelection.tell(heartbeat, null);

服务端AbWorkerRequestHandler#processWorkerHeartbeat，用来处理心跳请求。Server在接收到心跳信息后会进行状态的更新，ClusterStatusHolder#updateStatus。

public void updateStatus(WorkerHeartbeat heartbeat) {

        String workerAddress = heartbeat.getWorkerAddress();
        long heartbeatTime = heartbeat.getHeartbeatTime();

        WorkerInfo workerInfo = address2WorkerInfo.computeIfAbsent(workerAddress, ignore -> {
            WorkerInfo wf = new WorkerInfo();
            wf.refresh(heartbeat);
            return wf;
        });
        long oldTime = workerInfo.getLastActiveTime();
        if (heartbeatTime < oldTime) {
            log.warn("[ClusterStatusHolder-{}] receive the expired heartbeat from {}, serverTime: {}, heartTime: {}", appName, heartbeat.getWorkerAddress(), System.currentTimeMillis(), heartbeat.getHeartbeatTime());
            return;
        }

        workerInfo.refresh(heartbeat);

        List<DeployedContainerInfo> containerInfos = heartbeat.getContainerInfos();
        if (!CollectionUtils.isEmpty(containerInfos)) {
            containerInfos.forEach(containerInfo -> {
                Map<String, DeployedContainerInfo> infos = containerId2Infos.computeIfAbsent(containerInfo.getContainerId(), ignore -> Maps.newConcurrentMap());
                infos.put(workerAddress, containerInfo);
            });
        }
    }

总结

akka是一个说是比较简单的框架，但是是scala写的，不同版本差异很大，不熟悉scala的很难用的好。
vertx是比较简单的框架。
目前只整理了如上5个功能的源码，其实我对任务调度框架的理解，就是服务端告诉客户端你该执行了。作者还是做了很多检查任务状态的处理，保证了任务可以顺利执行。更加全面，也更加复杂了。

本文章为转载内容，我们尊重原作者对文章享有的著作权。如有内容错误或侵权问题，欢迎原作者联系我们进行内容更正或删除文章。