Пример #1
0
def createFirstJob(command, config, memory=None, cpu=None, time=sys.maxint):
    """Adds the first job to to the jobtree.
    """
    logger.info("Adding the first job")
    if memory == None:
        memory = config.attrib["default_memory"]
    if cpu == None:
        cpu = config.attrib["default_cpu"]
    job = createJob({ "command":command, "memory":str(int(memory)), "cpu":str(int(cpu)), "time":str(float(time)) }, None, config)
    writeJobs([job])
    logger.info("Added the first job")
Пример #2
0
def main():
    sys.path.append(sys.argv[1])
    sys.argv.remove(sys.argv[1])
    
    #Now we can import all the stuff..
    from sonLib.bioio import getBasicOptionParser
    from sonLib.bioio import parseBasicOptions
    from sonLib.bioio import logger
    from sonLib.bioio import addLoggingFileHandler
    from sonLib.bioio import setLogLevel
    from sonLib.bioio import getTotalCpuTime
    from sonLib.bioio import getTempDirectory
    from jobTree.src.master import writeJobs
    from jobTree.src.master import readJob
    from sonLib.bioio import system
    
    ##########################################
    #Construct the arguments.
    ##########################################
    
    parser = getBasicOptionParser("usage: %prog [options]", "%prog 0.1")
    
    parser.add_option("--job", dest="jobFile", 
                      help="Job file containing command to run",
                      default="None")
    
    options, args = parseBasicOptions(parser)
    assert len(args) == 0

    ##########################################
    #Parse the job.
    ##########################################
    
    job = readJob(options.jobFile)
    
    ##########################################
    #Setup the logging
    ##########################################
    
    #Setup the logging
    setLogLevel(job.attrib["log_level"])
    addLoggingFileHandler(job.attrib["slave_log_file"], rotatingLogging=False)
    logger.info("Parsed arguments and set up logging")
    
    ##########################################
    #Setup the stats, if requested
    ##########################################
    
    if job.attrib.has_key("stats"):
        startTime = time.time()
        startClock = getTotalCpuTime()
        stats = ET.Element("slave")
    else:
        stats = None
    
    ##########################################
    #Load the environment for the job
    ##########################################
    
    #First load the environment for the job.
    fileHandle = open(job.attrib["environment_file"], 'r')
    environment = cPickle.load(fileHandle)
    fileHandle.close()
    for i in environment:
        if i not in ("TMPDIR", "TMP", "HOSTNAME", "HOSTTYPE"):
            os.environ[i] = environment[i]
    # sys.path is used by __import__ to find modules
    if "PYTHONPATH" in environment:
        for e in environment["PYTHONPATH"].split(':'):
            if e != '':
                sys.path.append(e)
    #os.environ = environment
    #os.putenv(key, value)
    logger.info("Loaded the environment for the process")
        
    ##########################################
    #Setup the temporary directories.
    ##########################################
        
    #Dir to put all the temp files in.
    localSlaveTempDir = getTempDirectory()
    localTempDir = os.path.join(localSlaveTempDir, "localTempDir") 
    os.mkdir(localTempDir)
    os.chmod(localTempDir, 0777)
    
    ##########################################
    #Run the script.
    ##########################################
    
    maxTime = float(job.attrib["job_time"])
    assert maxTime > 0.0
    assert maxTime < sys.maxint
    jobToRun = job.find("followOns").findall("followOn")[-1]
    memoryAvailable = int(jobToRun.attrib["memory"])
    cpuAvailable = int(jobToRun.attrib["cpu"])
    startTime = time.time()
    while True:
        tempLogFile = processJob(job, jobToRun, memoryAvailable, cpuAvailable, stats, environment, localSlaveTempDir, localTempDir)
        
        if job.attrib["colour"] != "black": 
            logger.critical("Exiting the slave because of a failed job")
            system("mv %s %s" % (tempLogFile, job.attrib["log_file"])) #Copy back the job log file, because we saw failure
            break
        elif job.attrib.has_key("reportAllJobLogFiles"):
            logger.info("Exiting because we've been asked to report all logs, and this involves returning to the master")
            #Copy across the log file
            system("mv %s %s" % (tempLogFile, job.attrib["log_file"]))
            break
   
        totalRuntime = float(job.attrib["total_time"])  #This is the estimate runtime of the jobs on the followon stack
        
        childrenNode = job.find("children")
        childrenList = childrenNode.findall("child")
        #childRuntime = sum([ float(child.attrib["time"]) for child in childrenList ])
            
        if len(childrenList) >= 2: # or totalRuntime + childRuntime > maxTime: #We are going to have to return to the parent
            logger.info("No more jobs can run in series by this slave, its got %i children" % len(childrenList))
            break
        
        if time.time() - startTime > maxTime:
            logger.info("We are breaking because the maximum time the job should run for has been exceeded")
            break
        
        followOns = job.find("followOns")
        while len(childrenList) > 0:
            child = childrenList.pop()
            childrenNode.remove(child)
            totalRuntime += float(child.attrib["time"])
            ET.SubElement(followOns, "followOn", child.attrib.copy())
        #assert totalRuntime <= maxTime + 1 #The plus one second to avoid unimportant rounding errors
        job.attrib["total_time"] = str(totalRuntime)
        assert len(childrenNode.findall("child")) == 0
        
        if len(followOns.findall("followOn")) == 0:
            logger.info("No more jobs can run by this slave as we have exhausted the follow ons")
            break
        
        #Get the next job and see if we have enough cpu and memory to run it..
        jobToRun = job.find("followOns").findall("followOn")[-1]
        if int(jobToRun.attrib["memory"]) > memoryAvailable:
            logger.info("We need more memory for the next job, so finishing")
            break
        if int(jobToRun.attrib["cpu"]) > cpuAvailable:
            logger.info("We need more cpus for the next job, so finishing")
            break
        
        ##Updated the job so we can start the next loop cycle
        job.attrib["colour"] = "grey"
        writeJobs([ job ])
        logger.info("Updated the status of the job to grey and starting the next job")
    
    #Write back the job file with the updated jobs, using the checkpoint method.
    writeJobs([ job ])
    logger.info("Written out an updated job file")
    
    logger.info("Finished running the chain of jobs on this node, we ran for a total of %f seconds" % (time.time() - startTime))
    
    ##########################################
    #Cleanup the temporary directory
    ##########################################
    
    system("rm -rf %s" % localSlaveTempDir)
    
    ##########################################
    #Finish up the stats
    ##########################################
    
    if stats != None:
        stats.attrib["time"] = str(time.time() - startTime)
        stats.attrib["clock"] = str(getTotalCpuTime() - startClock)
        fileHandle = open(job.attrib["stats"], 'w')
        ET.ElementTree(stats).write(fileHandle)
        fileHandle.close()
Пример #3
0
def main():
    sys.path +=  [ sys.argv[1] ]
    sys.argv.remove(sys.argv[1])
    
    #Now we can import all the stuff..
    from jobTree.src.bioio import getBasicOptionParser
    from jobTree.src.bioio import parseBasicOptions
    from jobTree.src.bioio import logger
    from jobTree.src.bioio import addLoggingFileHandler
    from jobTree.src.bioio import setLogLevel
    from jobTree.src.bioio import getTotalCpuTime
    
    from jobTree.src.master import writeJobs
    
    ##########################################
    #Construct the arguments.
    ##########################################
    
    parser = getBasicOptionParser("usage: %prog [options]", "%prog 0.1")
    
    parser.add_option("--job", dest="jobFile", 
                      help="Job file containing command to run",
                      default="None")
    
    options, args = parseBasicOptions(parser)
    assert len(args) == 0

    ##########################################
    #Parse the job.
    ##########################################
    
    job = ET.parse(options.jobFile).getroot()
    
    ##########################################
    #Setup the logging
    ##########################################
    
    #Setup the logging
    setLogLevel(job.attrib["log_level"])
    addLoggingFileHandler(job.attrib["slave_log_file"], rotatingLogging=False)
    logger.info("Parsed arguments and set up logging")
    
    ##########################################
    #Setup the stats, if requested
    ##########################################
    
    if job.attrib.has_key("stats"):
        startTime = time.time()
        startClock = time.clock()
        stats = ET.Element("slave")
    else:
        stats = None
    
    ##########################################
    #Run the script.
    ##########################################
    
    maxTime = float(job.attrib["job_time"])
    assert maxTime > 0.0
    assert maxTime < sys.maxint
    jobToRun = job.find("followOns").findall("followOn")[-1]
    memoryAvailable = int(jobToRun.attrib["memory"])
    cpuAvailable = int(jobToRun.attrib["cpu"])
    while True:
        processJob(job, jobToRun, memoryAvailable, cpuAvailable, stats)
        
        if job.attrib["colour"] != "black":
            logger.info("Exiting the slave because of a failed job")
            break
   
        totalRuntime = float(job.attrib["total_time"])  #This is the estimate runtime of the jobs on the followon stack
        
        childrenNode = job.find("children")
        childrenList = childrenNode.findall("child")
        #childRuntime = sum([ float(child.attrib["time"]) for child in childrenList ])
            
        if len(childrenList) >= 2: # or totalRuntime + childRuntime > maxTime: #We are going to have to return to the parent
            logger.info("No more jobs can run in series by this slave, its got %i children" % len(childrenList))
            break
        
        followOns = job.find("followOns")
        while len(childrenList) > 0:
            child = childrenList.pop()
            childrenNode.remove(child)
            totalRuntime += float(child.attrib["time"])
            ET.SubElement(followOns, "followOn", child.attrib.copy())
        #assert totalRuntime <= maxTime + 1 #The plus one second to avoid unimportant rounding errors
        job.attrib["total_time"] = str(totalRuntime)
        assert len(childrenNode.findall("child")) == 0
        
        if len(followOns.findall("followOn")) == 0:
            logger.info("No more jobs can run by this slave as we have exhausted the follow ons")
            break
        
        #Get the next job and see if we have enough cpu and memory to run it..
        jobToRun = job.find("followOns").findall("followOn")[-1]
        if int(jobToRun.attrib["memory"]) > memoryAvailable:
            logger.info("We need more memory for the next job, so finishing")
            break
        if int(jobToRun.attrib["cpu"]) > cpuAvailable:
            logger.info("We need more cpus for the next job, so finishing")
            break
        
        ##Updated the job so we can start the next loop cycle
        job.attrib["colour"] = "grey"
        writeJobs([ job ])
        logger.info("Updated the status of the job to grey and starting the next job")
    
    #Write back the job file with the updated jobs, using the checkpoint method.
    writeJobs([ job ])
    logger.info("Written out an updated job file")
    
    logger.info("Finished running the chain of jobs on this node")
    
    ##########################################
    #Finish up the stats
    ##########################################
    
    if stats != None:
        stats.attrib["time"] = str(time.time() - startTime)
        stats.attrib["clock"] = str(getTotalCpuTime() - startClock)
        fileHandle = open(job.attrib["stats"], 'w')
        ET.ElementTree(stats).write(fileHandle)
        fileHandle.close()