source: src/Fragmentation/Automation/FragmentScheduler.cpp@ 668b55

/*
 * Project: MoleCuilder
 * Description: creates and alters molecular systems
 * Copyright (C)  2011 University of Bonn. All rights reserved.
 * Please see the LICENSE file or "Copyright notice" in builder.cpp for details.
 */

/*
 * \file FragmentScheduler.cpp
 *
 * This file strongly follows the Serialization example from the boost::asio
 * library (see server.cpp)
 *
 *  Created on: Oct 19, 2011
 *      Author: heber
 */

// include config.h
#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

// boost asio needs specific operator new
#include <boost/asio.hpp>

#include "CodePatterns/MemDebug.hpp"

#include <algorithm>
#include <boost/bind.hpp>
#include <boost/lambda/lambda.hpp>
#include <boost/lexical_cast.hpp>
#include <iostream>
#include <vector>
#include "Connection.hpp" // Must come before boost/serialization headers.
#include <boost/serialization/vector.hpp>
#include "CodePatterns/Info.hpp"
#include "CodePatterns/Log.hpp"
#include "CodePatterns/Observer/Notification.hpp"
#include "ControllerChoices.hpp"
#include "Operations/Servers/SendJobToWorkerOperation.hpp"
#include "Operations/Workers/EnrollInPoolOperation.hpp"
#include "Jobs/MPQCCommandJob.hpp"
#include "Jobs/SystemCommandJob.hpp"
#include "JobId.hpp"

#include "FragmentScheduler.hpp"

/** Helper function to enforce binding of FragmentWorker to possible derived
 * FragmentJob classes.
 */
void dummyInit() {
  SystemCommandJob("/bin/false", "something", JobId::IllegalJob);
  MPQCCommandJob("nofile", JobId::IllegalJob);
}

/** Constructor of class FragmentScheduler.
 *
 * We setup both acceptors to accept connections from workers and Controller.
 *
 * \param io_service io_service of the asynchronous communications
 * \param workerport port to listen for worker connections
 * \param controllerport port to listen for controller connections.
 */
FragmentScheduler::FragmentScheduler(boost::asio::io_service& _io_service, unsigned short workerport, unsigned short controllerport) :
    Observer("FragmentScheduler"),
    io_service(_io_service),
    WorkerListener(_io_service, workerport, JobsQueue, pool,
        boost::bind(&FragmentScheduler::sendJobToWorker, boost::ref(*this), _1, _2)),
    ControllerListener(_io_service, controllerport, JobsQueue,
        boost::bind(&FragmentScheduler::removeAllWorkers, boost::ref(*this)),
        boost::bind(&FragmentScheduler::shutdown, boost::ref(*this))),
    connection(_io_service)
{
  Info info(__FUNCTION__);

  // sign on to idle workers and present jobs
  pool.signOn(this, WorkerPool::WorkerIdle);
  JobsQueue.signOn(this, FragmentQueue::JobAdded);

  // listen for controller
  ControllerListener.initiateSocket();

  // listen for workers
  WorkerListener.initiateSocket();
}

FragmentScheduler::~FragmentScheduler()
{
  // sign off
  pool.signOff(this, WorkerPool::WorkerIdle);
  JobsQueue.signOff(this, FragmentQueue::JobAdded);
}

/** Handle a new worker connection.
 *
 * We store the given address in the pool.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::WorkerListener_t::handle_Accept(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  if (!e)
  {
    // Successfully accepted a new connection.
    // read address
    conn->async_read(address,
      boost::bind(&FragmentScheduler::WorkerListener_t::handle_ReadAddress, this,
      boost::asio::placeholders::error, conn));
  }
  else
  {
    // An error occurred. Log it and return. Since we are not starting a new
    // accept operation the io_service will run out of work to do and the
    // server will exit.
    Exitflag = ErrorFlag;
    ELOG(0, e.message());
  }
}

/** Handle having received Worker's address
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::WorkerListener_t::handle_ReadAddress(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  if (!e)
  {
    // Successfully accepted a new connection.
    // read address
    conn->async_read(choice,
      boost::bind(&FragmentScheduler::WorkerListener_t::handle_ReadChoice, this,
      boost::asio::placeholders::error, conn));
  }
  else
  {
    // An error occurred. Log it and return. Since we are not starting a new
    // accept operation the io_service will run out of work to do and the
    // server will exit.
    Exitflag = ErrorFlag;
    ELOG(0, e.message());
  }
}

/** Controller callback function to read the choice for next operation.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::WorkerListener_t::handle_ReadChoice(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  if (!e)
  {
    LOG(1, "INFO: Received request for operation " << choice << ".");
    // switch over the desired choice read previously
    switch(choice) {
      case NoWorkerOperation:
      {
        ELOG(1, "WorkerListener_t::handle_ReadChoice() - called with NoOperation.");
        break;
      }
      case EnrollInPool:
      {
        if (pool.presentInPool(address)) {
          ELOG(1, "INFO: worker "+toString(address)+" is already contained in pool.");
          enum EnrollInPoolOperation::EnrollFlag flag = EnrollInPoolOperation::Fail;
          conn->async_write(flag,
            boost::bind(&FragmentScheduler::WorkerListener_t::handle_enrolled, this,
            boost::asio::placeholders::error, conn));
        } else {
          // insert as its new worker
          LOG(1, "INFO: Adding " << address << " to pool ...");
          pool.addWorker(address);
          enum EnrollInPoolOperation::EnrollFlag flag = EnrollInPoolOperation::Success;
          conn->async_write(flag,
            boost::bind(&FragmentScheduler::WorkerListener_t::handle_enrolled, this,
            boost::asio::placeholders::error, conn));
        break;
      }
      case SendResult:
      {
        if (pool.presentInPool(address)) {
          // check whether its priority is busy_priority
          if (pool.isWorkerBusy(address)) {
            conn->async_read(result,
              boost::bind(&FragmentScheduler::WorkerListener_t::handle_ReceiveResultFromWorker, this,
              boost::asio::placeholders::error, conn));
          } else {
            ELOG(1, "Worker " << address << " trying to send result who is not marked as busy.");
            conn->async_read(result,
              boost::bind(&FragmentScheduler::WorkerListener_t::handle_RejectResultFromWorker, this,
              boost::asio::placeholders::error, conn));
          }
        } else {
          ELOG(1, "Worker " << address << " trying to send result who is not in pool.");
          conn->async_read(result,
            boost::bind(&FragmentScheduler::WorkerListener_t::handle_RejectResultFromWorker, this,
            boost::asio::placeholders::error, conn));
        }
        break;
      }
      case RemoveFromPool:
      {
        if (pool.presentInPool(address)) {
          // removing present worker
          pool.removeWorker(address);
        } else {
          ELOG(1, "Shutting down Worker " << address << " not contained in pool.");
        }
        break;
      }
      default:
        Exitflag = ErrorFlag;
        ELOG(1, "WorkerListener_t::handle_ReadChoice() - called with no valid choice.");
        break;
      }
    }
    // restore NoOperation choice such that choice is not read twice
    choice = NoWorkerOperation;

    initiateSocket();
  }
  else
  {
    // An error occurred. Log it and return. Since we are not starting a new
    // accept operation the io_service will run out of work to do and the
    // server will exit.
    Exitflag = ErrorFlag;
    ELOG(0, e.message());
  }
}


/** Callback function when new worker has enrolled.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::WorkerListener_t::handle_enrolled(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  if (e)
  {
    // An error occurred. Log it and return. Since we are not starting a new
    // accept operation the io_service will run out of work to do and the
    // server will exit.
    Exitflag = ErrorFlag;
    ELOG(0, e.message());
  }
}

/** Callback function when result has been received.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::WorkerListener_t::handle_ReceiveResultFromWorker(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  LOG(1, "INFO: Received result for job #" << result->getId() << " ...");

  // and push into queue
  ASSERT(result->getId() != (JobId_t)JobId::NoJob,
      "WorkerListener_t::handle_ReceiveResultFromWorker() - result received has NoJob id.");
  ASSERT(result->getId() != (JobId_t)JobId::IllegalJob,
      "WorkerListener_t::handle_ReceiveResultFromWorker() - result received has IllegalJob id.");
  // place id into expected
  if ((result->getId() != (JobId_t)JobId::NoJob) && (result->getId() != (JobId_t)JobId::IllegalJob))
    JobsQueue.pushResult(result);

  // mark as idle
  pool.unmarkWorkerBusy(address);

  // erase result
  result.reset();
  LOG(1, "INFO: JobsQueue has " << JobsQueue.getDoneJobs() << " results.");
}

/** Callback function when result has been received.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::WorkerListener_t::handle_RejectResultFromWorker(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  // nothing to do
  LOG(1, "INFO: Rejecting result for job #" << result->getId() << ", placing back into queue.");

  JobsQueue.resubmitJob(result->getId());

  LOG(1, "INFO: JobsQueue has " << JobsQueue.getDoneJobs() << " results.");
}


/** Handle a new controller connection.
 *
 * \sa handle_ReceiveJobs()
 * \sa handle_CheckResultState()
 * \sa handle_SendResults()
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::ControllerListener_t::handle_Accept(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  if (!e)
  {
    conn->async_read(choice,
      boost::bind(&FragmentScheduler::ControllerListener_t::handle_ReadChoice, this,
      boost::asio::placeholders::error, conn));
  }
  else
  {
    // An error occurred. Log it and return. Since we are not starting a new
    // accept operation the io_service will run out of work to do and the
    // server will exit.
    Exitflag = ErrorFlag;
    ELOG(0, e.message());
  }
}

/** Controller callback function to read the choice for next operation.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::ControllerListener_t::handle_ReadChoice(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  if (!e)
  {
    bool LaunchNewAcceptor = true;
    LOG(1, "INFO: Received request for operation " << choice << ".");
    // switch over the desired choice read previously
    switch(choice) {
    case NoControllerOperation:
    {
      ELOG(1, "ControllerListener_t::handle_ReadChoice() - called with NoOperation.");
      break;
    }
    case GetNextJobId:
    {
      LOG(1, "INFO: Receiving number of desired job ids from controller ...");
      conn->async_read(NumberIds,
        boost::bind(&FragmentScheduler::ControllerListener_t::handle_GetNextJobIdState, this,
        boost::asio::placeholders::error, conn));
      break;
    }
    case SendJobs:
    {
      // The connection::async_write() function will automatically
      // serialize the data structure for us.
      LOG(1, "INFO: Receiving bunch of jobs from a controller ...");
      conn->async_read(jobs,
        boost::bind(&FragmentScheduler::ControllerListener_t::handle_ReceiveJobs, this,
        boost::asio::placeholders::error, conn));
      break;
    }
    case CheckState:
    {
      // first update number
      jobInfo[0] = JobsQueue.getPresentJobs();
      jobInfo[1] = JobsQueue.getDoneJobs();
      // now we accept connections to check for state of calculations
      LOG(1, "INFO: Sending state that "+toString(jobInfo[0])+" jobs are present and "+toString(jobInfo[1])+" jobs are done to controller ...");
      conn->async_write(jobInfo,
        boost::bind(&FragmentScheduler::ControllerListener_t::handle_CheckResultState, this,
        boost::asio::placeholders::error, conn));
      break;
    }
    case RemoveAll:
    {
      removeallWorkers();
      break;
    }
    case ReceiveResults:
    {
      const std::vector<FragmentResult::ptr> results = JobsQueue.getAllResults();
      // ... or we give the results
      LOG(1, "INFO: Sending "+toString(results.size())+" results to controller ...");
      conn->async_write(results,
        boost::bind(&FragmentScheduler::ControllerListener_t::handle_SendResults, this,
        boost::asio::placeholders::error, conn));
      break;
    }
    case ShutdownControllerSocket:
    {
      LOG(1, "INFO: Received shutdown from controller ...");
      // only allow for shutdown when there are no more jobs in the queue
      if (!JobsQueue.isJobPresent()) {
        // we shutdown? Hence, also shutdown controller
        LaunchNewAcceptor = !shutdownAllSockets();
      } else {
        ELOG(2, "There are still jobs waiting in the queue.");
      }
      break;
    }
    default:
      Exitflag = ErrorFlag;
      ELOG(1, "ControllerListener_t::handle_ReadChoice() - called with no valid choice.");
      break;
    }
    // restore NoControllerOperation choice such that choice is not read twice
    choice = NoControllerOperation;

    if (LaunchNewAcceptor) {
      LOG(1, "Launching new acceptor on socket.");
      // Start an accept operation for a new Connection.
      initiateSocket();
    }
  }
  else
  {
    // An error occurred. Log it and return. Since we are not starting a new
    // accept operation the io_service will run out of work to do and the
    // server will exit.
    Exitflag = ErrorFlag;
    ELOG(0, e.message());
  }
}

/** Controller callback function when job has been sent.
 *
 * We check here whether the worker socket is accepting, if there
 * have been no jobs we re-activate it, as it is shut down after
 * last job.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::ControllerListener_t::handle_ReceiveJobs(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  // jobs are received, hence place in JobsQueue
  if (!jobs.empty()) {
    LOG(1, "INFO: Pushing " << jobs.size() << " jobs into queue.");
    JobsQueue.pushJobs(jobs);
  }
  jobs.clear();
}

/** Controller callback function when checking on state of results.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::ControllerListener_t::handle_CheckResultState(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  // do nothing
  LOG(1, "INFO: Sent that " << jobInfo << " jobs are (scheduled, done).");
}

/** Controller callback function when checking on state of results.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::ControllerListener_t::handle_GetNextJobIdState(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);

  std::vector<JobId_t> nextids( NumberIds, JobId::IllegalJob);
  std::generate(nextids.begin(), nextids.end(),
      boost::bind(&GlobalJobId::getNextId, boost::ref(globalId)));
  LOG(1, "INFO: Sending next available job ids " << nextids << " to controller ...");
  conn->async_write(nextids,
    boost::bind(&FragmentScheduler::ControllerListener_t::handle_SendIds, this,
    boost::asio::placeholders::error, conn));
}

/** Controller callback function when free job ids have been sent.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::ControllerListener_t::handle_SendIds(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  // do nothing
  LOG(1, "INFO: Ids have been sent.");
}

/** Controller callback function when result has been received.
 *
 * \param e error code if something went wrong
 * \param conn reference with the connection
 */
void FragmentScheduler::ControllerListener_t::handle_SendResults(const boost::system::error_code& e, connection_ptr conn)
{
  Info info(__FUNCTION__);
  // do nothing
  LOG(1, "INFO: Results have been sent.");
}


/** Helper function to send a job to worker.
 *
 * Note that we do not set the worker as busy. We simply send it the job.
 *
 * @param address address of worker
 * @param job job to send
 */
void FragmentScheduler::sendJobToWorker(const WorkerAddress &address, FragmentJob::ptr &job)
{
  ASSERT( pool.isWorkerBusy(address),
      "FragmentScheduler::sendJobToWorker() - Worker "+toString(address)+" is not marked as busy.");
  LOG(1, "INFO: Sending job " << job->getId() << " to worker " << address << ".");

  // create op, sign on, and hand over to queue
  AsyncOperation *sendJobOp = new SendJobToWorkerOperation(connection,job);
  OpQueue.push_back(sendJobOp, address);
}

/** Helper function to shutdown a single worker.
 *
 * We send NoJob to indicate shutdown
 *
 * @param address of worker to shutdown
 */
void FragmentScheduler::shutdownWorker(const WorkerAddress &address)
{
  ASSERT( !pool.isWorkerBusy(address),
      "FragmentScheduler::sendJobToWorker() - Worker "+toString(address)+" is already busy.");
  LOG(2, "INFO: Shutting down worker " << address << "...");
  AsyncOperation *shutdownWorkerOp = new ShutdownWorkerOperation(connection);
  OpQueue.push_back(shutdownWorkerOp, address);
}

/** Sends shutdown to all current workers in the pool.
 *
 */
void FragmentScheduler::removeAllWorkers()
{
  // first, sign off such that no new jobs are given to workers
  pool.signOff(this, WorkerPool::WorkerIdle);

  LOG(2, "DEBUG: Waiting for busy workers to finish ...");
  while (pool.hasBusyWorkers())
    ;

  LOG(2, "INFO: Shutting down workers ...");
  // iterate until there are no more idle workers
  do {
    // get list of all idle workers
    typedef std::vector<std::pair<std::string, std::string> > WorkerList_t;
    WorkerList_t WorkerList = pool.getListOfIdleWorkers();

    // give all workers shutdown signal
    for (WorkerList_t::const_iterator iter = WorkerList.begin(); iter != WorkerList.end(); ++iter)
      shutdownWorker(WorkerAddress(iter->first, iter->second));

    // wait for pending shutdown operations
    while (!OpQueue.empty())
      ;
  } while (pool.presentIdleWorkers());
  pool.removeAllWorkers();
}

/** Helper function to shutdown the server properly.
 *
 * \todo one should idle here until all workers have returned from
 * calculating stuff (or workers need to still listen while they are
 * calculating which is probably better).
 *
 * \note We only shutdown when there are no workers left
 *
 * @return true - doing shutdown, false - precondition not met, not shutting down
 */
bool FragmentScheduler::shutdown()
{
  if (!pool.presentIdleWorkers() && !pool.hasBusyWorkers()) {
    LOG(1, "INFO: Shutting all down ...");

    /// close the worker listener's socket
    WorkerListener.closeSocket();

    /// close the controller listener's socket
    ControllerListener.closeSocket();

    /// finally, stop the io_service
    io_service.stop();
    return true;
  } else {
    ELOG(2, "There are still idle or busy workers present.");
    return false;
  }
}

/** Internal helper to send the next available job to the next idle worker.
 *
 */
void FragmentScheduler::sendAvailableJobToNextIdleWorker()
{
  const WorkerAddress address = pool.getNextIdleWorker();
  FragmentJob::ptr job = JobsQueue.popJob();
  sendJobToWorker(address, job);
}

void FragmentScheduler::update(Observable *publisher)
{
  ASSERT(0, "FragmentScheduler::update() - we are not signed on for global updates.");
}

void FragmentScheduler::recieveNotification(Observable *publisher, Notification_ptr notification)
{
  if ((publisher == &pool) && (notification->getChannelNo() == WorkerPool::WorkerIdle)) {
    // we have an idle worker
    LOG(1, "INFO: We are notified of an idle worker.");
    // are jobs available?
    if (JobsQueue.isJobPresent()) {
      sendAvailableJobToNextIdleWorker();
    }
  } else if ((publisher == &JobsQueue) && (notification->getChannelNo() == FragmentQueue::JobAdded)) {
    // we have new jobs
    LOG(1, "INFO: We are notified of a new job.");
    // check for idle workers
    if (pool.presentIdleWorkers()) {
      sendAvailableJobToNextIdleWorker();
    }
  } else {
    ASSERT(0, "FragmentScheduler::recieveNotification() - we are not signed on for updates in channel "
        +toString(notification->getChannelNo())+".");
  }
}

void FragmentScheduler::subjectKilled(Observable *publisher)
{}