def test_hello_world(self):
dir = os.path.dirname(os.path.abspath(__file__))
subprocess.check_call("python {dir}/helloWorld.py "
"--batchSystem=mesos "
"--logLevel={logLevel}".format(
dir=dir, logLevel=getLogLevelString()),
shell=True)
def __init__(self):
#Core options
self.jobStore = os.path.abspath("./toil")
self.logLevel = getLogLevelString()
self.workDir = None
self.stats = False
self.clean = "never"
#Restarting the workflow options
self.restart = False
#Batch system options
self.batchSystem = "singleMachine"
self.scale = 1
self.masterIP = '127.0.0.1:5050'
self.parasolCommand = "parasol"
#Resource requirements
self.defaultMemory = 2147483648
self.defaultCores = 1
self.defaultDisk = 2147483648
self.maxCores = sys.maxint
self.maxMemory = sys.maxint
self.maxDisk = sys.maxint
#Retrying/rescuing jobs
self.retryCount = 0
self.maxJobDuration = sys.maxint
self.rescueJobsFrequency = 3600
#Misc
self.maxLogFileSize = 50120
self.sseKey = None
def createConfig(options):
"""
Creates a config object from the options object.
TODO: Make the config object a proper class
"""
logger.info("Starting to create the toil setup for the first time")
config = ET.Element("config")
config.attrib["log_level"] = getLogLevelString()
config.attrib["master_ip"] = options.masterIP
config.attrib["job_store"] = os.path.abspath(
options.toil) if options.toil.startswith('.') else options.toil
config.attrib["parasol_command"] = options.parasolCommand
config.attrib["try_count"] = str(int(options.retryCount) + 1)
config.attrib["max_job_duration"] = str(float(options.maxJobDuration))
config.attrib["batch_system"] = options.batchSystem
config.attrib["job_time"] = str(float(options.jobTime))
config.attrib["max_log_file_size"] = str(int(options.maxLogFileSize))
config.attrib["default_memory"] = str(int(options.defaultMemory))
config.attrib["default_cpu"] = str(int(options.defaultCpu))
config.attrib["default_disk"] = str(int(options.defaultDisk))
config.attrib["max_cpus"] = str(int(options.maxCpus))
config.attrib["max_memory"] = str(int(options.maxMemory))
config.attrib["max_disk"] = str(int(options.maxDisk))
config.attrib["scale"] = str(float(options.scale))
if options.bigBatchSystem is not None:
config.attrib["big_batch_system"] = options.bigBatchSystem
config.attrib["big_memory_threshold"] = str(
int(options.bigMemoryThreshold))
config.attrib["big_cpu_threshold"] = str(int(options.bigCpuThreshold))
config.attrib["big_max_cpus"] = str(int(options.bigMaxCpus))
config.attrib["big_max_memory"] = str(int(options.bigMaxMemory))
if options.stats:
config.attrib["stats"] = ""
return config
def createConfig(options):
"""
Creates a config object from the options object.
TODO: Make the config object a proper class
"""
logger.info("Starting to create the toil setup for the first time")
config = ET.Element("config")
config.attrib["log_level"] = getLogLevelString()
config.attrib["master_ip"] = options.masterIP
config.attrib["job_store"] = os.path.abspath(options.toil) if options.toil.startswith('.') else options.toil
config.attrib["parasol_command"] = options.parasolCommand
config.attrib["try_count"] = str(int(options.retryCount) + 1)
config.attrib["max_job_duration"] = str(float(options.maxJobDuration))
config.attrib["batch_system"] = options.batchSystem
config.attrib["job_time"] = str(float(options.jobTime))
config.attrib["max_log_file_size"] = str(int(options.maxLogFileSize))
config.attrib["default_memory"] = str(int(options.defaultMemory))
config.attrib["default_cpu"] = str(int(options.defaultCpu))
config.attrib["default_disk"] = str(int(options.defaultDisk))
config.attrib["max_cpus"] = str(int(options.maxCpus))
config.attrib["max_memory"] = str(int(options.maxMemory))
config.attrib["max_disk"] = str(int(options.maxDisk))
config.attrib["scale"] = str(float(options.scale))
if options.bigBatchSystem is not None:
config.attrib["big_batch_system"] = options.bigBatchSystem
config.attrib["big_memory_threshold"] = str(int(options.bigMemoryThreshold))
config.attrib["big_cpu_threshold"] = str(int(options.bigCpuThreshold))
config.attrib["big_max_cpus"] = str(int(options.bigMaxCpus))
config.attrib["big_max_memory"] = str(int(options.bigMaxMemory))
if options.stats:
config.attrib["stats"] = ""
return config
def test_hello_world( self ):
dir = os.path.dirname( os.path.abspath( __file__ ) )
subprocess.check_call( "python {dir}/helloWorld.py "
"--batchSystem=mesos "
"--logLevel={logLevel}".format( dir=dir,
logLevel=getLogLevelString( ) ),
shell=True )
def test_hello_world(self):
system([sys.executable,
'-m', helloWorld.__name__,
'file:./toilTest',
'--batchSystem', 'mesos',
'--mesosMaster', 'localhost:5050',
'--logLevel', getLogLevelString()])
def __init__(self):
#Core options
self.jobStore = os.path.abspath("./toil")
self.logLevel = getLogLevelString()
self.workDir = None
self.stats = False
self.clean = "never"
#Restarting the workflow options
self.restart = False
#Batch system options
self.batchSystem = "singleMachine"
self.scale = 1
self.masterIP = '127.0.0.1:5050'
self.parasolCommand = "parasol"
#Resource requirements
self.defaultMemory = 2147483648
self.defaultCores = 1
self.defaultDisk = 2147483648
self.maxCores = sys.maxint
self.maxMemory = sys.maxint
self.maxDisk = sys.maxint
#Retrying/rescuing jobs
self.retryCount = 0
self.maxJobDuration = sys.maxint
self.rescueJobsFrequency = 3600
#Misc
self.maxLogFileSize=50120
self.sseKey = None
def test_hello_world(self):
system(
[
sys.executable,
"-m",
helloWorld.__name__,
"./toilTest",
"--batchSystem=mesos",
"--logLevel",
getLogLevelString(),
]
)
def toilSortTest(self,
testNo=1,
batchSystem="singleMachine",
jobStore='file',
lines=10000,
maxLineLength=10,
N=10000):
"""
Tests toil by sorting a file in parallel.
"""
for test in xrange(testNo):
options = Job.Runner.getDefaultOptions()
#toil
if jobStore == 'file':
options.toil = self.toilDir
else:
options.toil = jobStore
# Specify options
options.logLevel = getLogLevelString()
options.retryCount = 32
options.batchSystem = batchSystem
#Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile,
lines=lines,
maxLineLength=maxLineLength)
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Run the toil
firstJob = Job.wrapJobFn(
setup,
tempSortFile,
N,
memory=5000,
)
i = Job.Runner.startToil(firstJob, options)
#There should be no failed jobs
self.assertEquals(i, 0)
# Now check the file is properly sorted..
# Now get the sorted file
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
checkEqual(l, l2)
def __init__(self):
# Core options
self.workflowID = None
"""This attribute uniquely identifies the job store and therefore the workflow. It is
necessary in order to distinguish between two consequitive workflows for which
self.jobStore is the same, e.g. when a job store name is reused after a previous run has
finished sucessfully and its job store has been clean up."""
self.jobStore = os.path.abspath("./toil")
self.logLevel = getLogLevelString()
self.workDir = None
self.stats = False
# Because the stats option needs the jobStore to persist past the end of the run,
# the clean default value depends the specified stats option and is determined in setOptions
self.clean = None
self.cleanWorkDir = None
#Restarting the workflow options
self.restart = False
#Batch system options
self.batchSystem = "singleMachine"
self.scale = 1
self.mesosMasterAddress = 'localhost:5050'
self.parasolCommand = "parasol"
self.parasolMaxBatches = 10000
self.environment = {}
#Resource requirements
self.defaultMemory = 2147483648
self.defaultCores = 1
self.defaultDisk = 2147483648
self.defaultCache = self.defaultDisk
self.maxCores = sys.maxint
self.maxMemory = sys.maxint
self.maxDisk = sys.maxint
#Retrying/rescuing jobs
self.retryCount = 0
self.maxJobDuration = sys.maxint
self.rescueJobsFrequency = 3600
#Misc
self.maxLogFileSize=50120
self.sseKey = None
self.cseKey = None
self.servicePollingInterval = 60
#Debug options
self.badWorker = 0.0
self.badWorkerFailInterval = 0.01
def __init__(self):
# Core options
self.workflowID = None
"""This attribute uniquely identifies the job store and therefore the workflow. It is
necessary in order to distinguish between two consequitive workflows for which
self.jobStore is the same, e.g. when a job store name is reused after a previous run has
finished sucessfully and its job store has been clean up."""
self.jobStore = os.path.abspath("./toil")
self.logLevel = getLogLevelString()
self.workDir = None
self.stats = False
# Because the stats option needs the jobStore to persist past the end of the run,
# the clean default value depends the specified stats option and is determined in setOptions
self.clean = None
#Restarting the workflow options
self.restart = False
#Batch system options
self.batchSystem = "singleMachine"
self.scale = 1
self.mesosMasterAddress = 'localhost:5050'
self.parasolCommand = "parasol"
self.parasolMaxBatches = 10000
self.environment = {}
#Resource requirements
self.defaultMemory = 2147483648
self.defaultCores = 1
self.defaultDisk = 2147483648
self.defaultCache = self.defaultDisk
self.maxCores = sys.maxint
self.maxMemory = sys.maxint
self.maxDisk = sys.maxint
#Retrying/rescuing jobs
self.retryCount = 0
self.maxJobDuration = sys.maxint
self.rescueJobsFrequency = 3600
#Misc
self.maxLogFileSize = 50120
self.sseKey = None
self.cseKey = None
#Debug options
self.badWorker = 0.0
self.badWorkerFailInterval = 0.01
def __init__(self):
# Core options
self.jobStore = os.path.abspath("./toil")
self.logLevel = getLogLevelString()
self.workDir = None
self.stats = False
# Because the stats option needs the jobStore to persist past the end of the run,
# the clean default value depends the specified stats option and is determined in setOptions
self.clean = None
# Restarting the workflow options
self.restart = False
# Batch system options
self.batchSystem = "singleMachine"
self.scale = 1
self.mesosMasterAddress = "localhost:5050"
self.parasolCommand = "parasol"
self.parasolMaxBatches = 10000
self.environment = {}
# Resource requirements
self.defaultMemory = 2147483648
self.defaultCores = 1
self.defaultDisk = 2147483648
self.defaultCache = self.defaultDisk
self.maxCores = sys.maxint
self.maxMemory = sys.maxint
self.maxDisk = sys.maxint
# Retrying/rescuing jobs
self.retryCount = 0
self.maxJobDuration = sys.maxint
self.rescueJobsFrequency = 3600
# Misc
self.maxLogFileSize = 50120
self.sseKey = None
self.cseKey = None
# Debug options
self.badWorker = 0.0
self.badWorkerFailInterval = 0.01
def toilSortTest(self, testNo=1, batchSystem="singleMachine", jobStore='file',
lines=10000, maxLineLength=10, N=10000):
"""
Tests toil by sorting a file in parallel.
"""
for test in xrange(testNo):
options = Job.Runner.getDefaultOptions()
#toil
if jobStore == 'file':
options.toil = self.toilDir
else:
options.toil = jobStore
# Specify options
options.logLevel = getLogLevelString()
options.retryCount = 32
options.batchSystem = batchSystem
#Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile, lines=lines, maxLineLength=maxLineLength)
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Run the toil
firstJob = Job.wrapJobFn(setup, tempSortFile, N, memory=5000, )
i = Job.Runner.startToil(firstJob, options)
#There should be no failed jobs
self.assertEquals(i, 0)
# Now check the file is properly sorted..
# Now get the sorted file
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
checkEqual(l, l2)
def _toilSort(self, jobStore, batchSystem,
lines=defaultLines, N=defaultN, testNo=1, lineLen=defaultLineLen):
"""
Generate a file consisting of the given number of random lines, each line of the given
length. Sort the file with Toil by splitting the file recursively until each part is less
than the given number of bytes, sorting each part and merging them back together. Then
verify the result.
:param jobStore: a job store string
:param batchSystem: the name of the batch system
:param lines: the number of random lines to generate
:param N: the size in bytes of each split
:param testNo: the number of repeats of this test
:param lineLen: the length of each random line in the file
"""
for test in xrange(testNo):
try:
# Specify options
options = Job.Runner.getDefaultOptions(jobStore)
options.logLevel = getLogLevelString()
options.retryCount = 2
options.batchSystem = batchSystem
options.clean = "never"
options.badWorker = 0.5
options.badWorkerFailInterval = 0.05
# Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile, lines=lines, lineLen=lineLen)
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Make the first job
firstJob = Job.wrapJobFn(setup, tempSortFile, N, memory=sortMemory)
# Check we get an exception if we try to restart a workflow that doesn't exist
options.restart = True
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = False
# Now actually run the workflow
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Check we get an exception if we try to run without restart on an existing store
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = True
# This loop tests the restart behavior
totalTrys = 1
while i != 0:
options.useExistingOptions = random.random() > 0.5
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
if totalTrys > 16: #p(fail after this many restarts) = 0.5**32
self.fail() #Exceeded a reasonable number of restarts
totalTrys += 1
# Now check that if you try to restart from here it will raise an exception
# indicating that there are no jobs remaining in the workflow.
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobException:
pass
# Now check the file is properly sorted..
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
self.assertEquals(l, l2)
finally:
subprocess.check_call([resolveEntryPoint('toil'), 'clean', jobStore])
def test_hello_world(self):
system([sys.executable,
'-m', helloWorld.__name__,
'./toilTest',
'--batchSystem=mesos',
'--logLevel', getLogLevelString()])
def _toilSort(self, jobStoreLocator, batchSystem,
lines=defaultLines, N=defaultN, testNo=1, lineLen=defaultLineLen,
retryCount=2, badWorker=0.5, downCheckpoints=False, disableCaching=False):
"""
Generate a file consisting of the given number of random lines, each line of the given
length. Sort the file with Toil by splitting the file recursively until each part is less
than the given number of bytes, sorting each part and merging them back together. Then
verify the result.
:param jobStoreLocator: The location of the job store.
:param batchSystem: the name of the batch system
:param lines: the number of random lines to generate
:param N: the size in bytes of each split
:param testNo: the number of repeats of this test
:param lineLen: the length of each random line in the file
"""
for test in xrange(testNo):
try:
# Specify options
options = Job.Runner.getDefaultOptions(jobStoreLocator)
options.logLevel = getLogLevelString()
options.retryCount = retryCount
options.batchSystem = batchSystem
options.clean = "never"
options.badWorker = badWorker
options.badWorkerFailInterval = 0.05
options.disableCaching = disableCaching # FIXME maybe this line should be deleted
options.downCheckpoints = downCheckpoints
options.N = N
# Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile, lines=lines, lineLen=lineLen)
options.fileToSort = tempSortFile
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Check we get an exception if we try to restart a workflow that doesn't exist
options.restart = True
with self.assertRaises(NoSuchJobStoreException):
main(options)
options.restart = False
# Now actually run the workflow
try:
main(options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Check we get an exception if we try to run without restart on an existing store
with self.assertRaises(JobStoreExistsException):
main(options)
options.restart = True
# This loop tests the restart behavior
totalTrys = 1
while i != 0:
options.useExistingOptions = random.random() > 0.5
try:
main(options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
if totalTrys > 32: # p(fail after this many restarts) = 0.5**32
self.fail('Exceeded a reasonable number of restarts')
totalTrys += 1
# Now check that if you try to restart from here it will raise an exception
# indicating that there are no jobs remaining in the workflow.
with self.assertRaises(JobException):
main(options)
# Now check the file is properly sorted..
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
self.assertEquals(l, l2)
finally:
subprocess.check_call([resolveEntryPoint('toil'), 'clean', jobStoreLocator])
def __init__(self):
# Core options
self.workflowID = None
"""This attribute uniquely identifies the job store and therefore the workflow. It is
necessary in order to distinguish between two consequitive workflows for which
self.jobStore is the same, e.g. when a job store name is reused after a previous run has
finished sucessfully and its job store has been clean up."""
self.workflowAttemptNumber = None
self.jobStore = None
self.logLevel = getLogLevelString()
self.workDir = None
self.stats = False
# Because the stats option needs the jobStore to persist past the end of the run,
# the clean default value depends the specified stats option and is determined in setOptions
self.clean = None
self.cleanWorkDir = None
#Restarting the workflow options
self.restart = False
#Batch system options
self.batchSystem = "singleMachine"
self.scale = 1
self.mesosMasterAddress = 'localhost:5050'
self.parasolCommand = "parasol"
self.parasolMaxBatches = 10000
self.environment = {}
#Autoscaling options
self.provisioner = None
self.nodeType = None
self.nodeOptions = None
self.minNodes = 0
self.maxNodes = 10
self.preemptableNodeType = None
self.preemptableNodeOptions = None
self.minPreemptableNodes = 0
self.maxPreemptableNodes = 0
self.alphaPacking = 0.8
self.betaInertia = 1.2
self.scaleInterval = 10
self.preemptableCompensation = 0.0
#Resource requirements
self.defaultMemory = 2147483648
self.defaultCores = 1
self.defaultDisk = 2147483648
self.readGlobalFileMutableByDefault = False
self.defaultPreemptable = False
self.maxCores = sys.maxint
self.maxMemory = sys.maxint
self.maxDisk = sys.maxint
#Retrying/rescuing jobs
self.retryCount = 0
self.maxJobDuration = sys.maxint
self.rescueJobsFrequency = 3600
#Misc
self.disableCaching = False
self.maxLogFileSize = 50120
self.sseKey = None
self.cseKey = None
self.servicePollingInterval = 60
self.useAsync = True
#Debug options
self.badWorker = 0.0
self.badWorkerFailInterval = 0.01
def __init__(self):
# Core options
self.workflowID = None
"""This attribute uniquely identifies the job store and therefore the workflow. It is
necessary in order to distinguish between two consequitive workflows for which
self.jobStore is the same, e.g. when a job store name is reused after a previous run has
finished sucessfully and its job store has been clean up."""
self.workflowAttemptNumber = None
self.jobStore = None
self.logLevel = getLogLevelString()
self.workDir = None
self.stats = False
# Because the stats option needs the jobStore to persist past the end of the run,
# the clean default value depends the specified stats option and is determined in setOptions
self.clean = None
self.cleanWorkDir = None
#Restarting the workflow options
self.restart = False
#Batch system options
self.batchSystem = "singleMachine"
self.scale = 1
self.mesosMasterAddress = 'localhost:5050'
self.parasolCommand = "parasol"
self.parasolMaxBatches = 10000
self.environment = {}
#Autoscaling options
self.provisioner = None
self.nodeType = None
self.nodeOptions = None
self.minNodes = 0
self.maxNodes = 10
self.preemptableNodeType = None
self.preemptableNodeOptions = None
self.minPreemptableNodes = 0
self.maxPreemptableNodes = 0
self.alphaPacking = 0.8
self.betaInertia = 1.2
self.scaleInterval = 10
self.preemptableCompensation = 0.0
# Parameters to limit service jobs, so preventing deadlock scheduling scenarios
self.maxPreemptableServiceJobs = sys.maxint
self.maxServiceJobs = sys.maxint
self.deadlockWait = 60 # Wait one minute before declaring a deadlock
#Resource requirements
self.defaultMemory = 2147483648
self.defaultCores = 1
self.defaultDisk = 2147483648
self.readGlobalFileMutableByDefault = False
self.defaultPreemptable = False
self.maxCores = sys.maxint
self.maxMemory = sys.maxint
self.maxDisk = sys.maxint
#Retrying/rescuing jobs
self.retryCount = 0
self.maxJobDuration = sys.maxint
self.rescueJobsFrequency = 3600
#Misc
self.disableCaching = False
self.maxLogFileSize=50120
self.sseKey = None
self.cseKey = None
self.servicePollingInterval = 60
self.useAsync = True
#Debug options
self.badWorker = 0.0
self.badWorkerFailInterval = 0.01
def _toilSort(self,
jobStoreLocator,
batchSystem,
lines=defaultLines,
N=defaultN,
testNo=1,
lineLen=defaultLineLen,
retryCount=2,
badWorker=0.5,
downCheckpoints=False,
disableCaching=False):
"""
Generate a file consisting of the given number of random lines, each line of the given
length. Sort the file with Toil by splitting the file recursively until each part is less
than the given number of bytes, sorting each part and merging them back together. Then
verify the result.
:param jobStoreLocator: The location of the job store.
:param batchSystem: the name of the batch system
:param lines: the number of random lines to generate
:param N: the size in bytes of each split
:param testNo: the number of repeats of this test
:param lineLen: the length of each random line in the file
"""
for test in range(testNo):
try:
# Specify options
options = Job.Runner.getDefaultOptions(jobStoreLocator)
options.logLevel = getLogLevelString()
options.retryCount = retryCount
options.batchSystem = batchSystem
options.clean = "never"
options.badWorker = badWorker
options.badWorkerFailInterval = 0.05
options.disableCaching = disableCaching
# This is required because mesosMasterAddress now defaults to the IP of the machine
# that is starting the workflow while the mesos *tests* run locally.
if batchSystem == 'mesos':
options.mesosMasterAddress = 'localhost:5050'
options.downCheckpoints = downCheckpoints
options.N = N
options.outputFile = self.outputFile
options.fileToSort = self.inputFile
options.overwriteOutput = True
options.realTimeLogging = True
# Make the file to sort
makeFileToSort(options.fileToSort,
lines=lines,
lineLen=lineLen)
# First make our own sorted version
with open(options.fileToSort, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Check we get an exception if we try to restart a workflow that doesn't exist
options.restart = True
with self.assertRaises(NoSuchJobStoreException):
with runMain(options):
# Now check the file is properly sorted..
with open(options.outputFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
self.assertEquals(l, l2)
options.restart = False
# Now actually run the workflow
try:
with runMain(options):
pass
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Check we get an exception if we try to run without restart on an existing store
with self.assertRaises(JobStoreExistsException):
with runMain(options):
pass
options.restart = True
# This loop tests the restart behavior
totalTrys = 1
while i != 0:
options.useExistingOptions = random.random() > 0.5
try:
with runMain(options):
pass
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
if totalTrys > 32: # p(fail after this many restarts) = 0.5**32
self.fail(
'Exceeded a reasonable number of restarts')
totalTrys += 1
finally:
subprocess.check_call(
[resolveEntryPoint('toil'), 'clean', jobStoreLocator])
# final test to make sure the jobStore was actually deleted
self.assertRaises(NoSuchJobStoreException, Toil.resumeJobStore,
jobStoreLocator)
def _toilSort(self,
jobStore,
batchSystem,
lines=defaultLines,
N=defaultN,
testNo=1,
lineLen=defaultLineLen):
"""
Generate a file consisting of the given number of random lines, each line of the given
length. Sort the file with Toil by splitting the file recursively until each part is less
than the given number of bytes, sorting each part and merging them back together. Then
verify the result.
:param jobStore: a job store string
:param batchSystem: the name of the batch system
:param lines: the number of random lines to generate
:param N: the size in bytes of each split
:param testNo: the number of repeats of this test
:param lineLen: the length of each random line in the file
"""
for test in xrange(testNo):
try:
# Specify options
options = Job.Runner.getDefaultOptions(jobStore)
options.logLevel = getLogLevelString()
options.retryCount = 2
options.batchSystem = batchSystem
options.clean = "never"
options.badWorker = 0.5
options.badWorkerFailInterval = 0.05
# Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile, lines=lines, lineLen=lineLen)
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Make the first job
firstJob = Job.wrapJobFn(setup,
tempSortFile,
N,
memory=sortMemory)
# Check we get an exception if we try to restart a workflow that doesn't exist
options.restart = True
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = False
# Now actually run the workflow
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Check we get an exception if we try to run without restart on an existing store
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = True
# This loop tests the restart behavior
totalTrys = 1
while i != 0:
options.useExistingOptions = random.random() > 0.5
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
if totalTrys > 16: #p(fail after this many restarts) = 0.5**32
self.fail(
) #Exceeded a reasonable number of restarts
totalTrys += 1
# Now check that if you try to restart from here it will raise an exception
# indicating that there are no jobs remaining in the workflow.
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobException:
pass
# Now check the file is properly sorted..
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
self.assertEquals(l, l2)
finally:
subprocess.check_call(
[resolveEntryPoint('toil'), 'clean', jobStore])
def toilSortTest(self,
jobStore,
batchSystem,
lines,
N,
testNo=1,
maxLineLength=10):
"""
Tests toil by sorting a file in parallel.
"""
for test in xrange(testNo):
# Specify options
options = Job.Runner.getDefaultOptions()
options.jobStore = jobStore
options.logLevel = getLogLevelString()
options.retryCount = 2
options.batchSystem = batchSystem
# Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile,
lines=lines,
maxLineLength=maxLineLength)
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Make the first job
firstJob = Job.wrapJobFn(setup, tempSortFile, N, memory=5000)
# Check we get an exception if we try to restart a workflow that doesn't exist
options.restart = True
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = False
# Now actually run the workflow
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Check we get an exception if we try to run without restart on an existing job store
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = True
# This loop tests the restart behavior
while i != 0:
options.useExistingOptions = random.random() > 0.5
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Now check that if you try to restart from here it will raise an exception
# indicating that there are no jobs remaining in the workflow.
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobException:
pass
# self.assertTrue(e.message.endswith('left in toil workflow (workflow has finished successfully?)'))
# Now check the file is properly sorted..
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
checkEqual(l, l2)
def _toilSort(self,
jobStoreLocator,
batchSystem,
lines=defaultLines,
N=defaultN,
testNo=1,
lineLen=defaultLineLen,
retryCount=2,
badWorker=0.5,
downCheckpoints=False,
disableCaching=False):
"""
Generate a file consisting of the given number of random lines, each line of the given
length. Sort the file with Toil by splitting the file recursively until each part is less
than the given number of bytes, sorting each part and merging them back together. Then
verify the result.
:param jobStoreLocator: The location of the job store.
:param batchSystem: the name of the batch system
:param lines: the number of random lines to generate
:param N: the size in bytes of each split
:param testNo: the number of repeats of this test
:param lineLen: the length of each random line in the file
"""
for test in xrange(testNo):
try:
# Specify options
options = Job.Runner.getDefaultOptions(jobStoreLocator)
options.logLevel = getLogLevelString()
options.retryCount = retryCount
options.batchSystem = batchSystem
options.clean = "never"
options.badWorker = badWorker
options.badWorkerFailInterval = 0.05
options.disableCaching = disableCaching # FIXME maybe this line should be deleted
options.downCheckpoints = downCheckpoints
options.N = N
# Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile, lines=lines, lineLen=lineLen)
options.fileToSort = tempSortFile
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Check we get an exception if we try to restart a workflow that doesn't exist
options.restart = True
with self.assertRaises(NoSuchJobStoreException):
main(options)
options.restart = False
# Now actually run the workflow
try:
main(options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Check we get an exception if we try to run without restart on an existing store
with self.assertRaises(JobStoreExistsException):
main(options)
options.restart = True
# This loop tests the restart behavior
totalTrys = 1
while i != 0:
options.useExistingOptions = random.random() > 0.5
try:
main(options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
if totalTrys > 32: # p(fail after this many restarts) = 0.5**32
self.fail(
'Exceeded a reasonable number of restarts')
totalTrys += 1
# Now check that if you try to restart from here it will raise an exception
# indicating that there are no jobs remaining in the workflow.
with self.assertRaises(JobException):
main(options)
# Now check the file is properly sorted..
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
self.assertEquals(l, l2)
finally:
subprocess.check_call(
[resolveEntryPoint('toil'), 'clean', jobStoreLocator])
def toilSortTest(self, jobStore, batchSystem, lines, N, testNo=1, maxLineLength=10):
"""
Tests toil by sorting a file in parallel.
"""
for test in xrange(testNo):
# Specify options
options = Job.Runner.getDefaultOptions()
options.jobStore = jobStore
options.logLevel = getLogLevelString()
options.retryCount = 2
options.batchSystem = batchSystem
# Make the file to sort
tempSortFile = os.path.join(self.tempDir, "fileToSort.txt")
makeFileToSort(tempSortFile, lines=lines, maxLineLength=maxLineLength)
# First make our own sorted version
with open(tempSortFile, 'r') as fileHandle:
l = fileHandle.readlines()
l.sort()
# Make the first job
firstJob = Job.wrapJobFn(setup, tempSortFile, N, memory=5000)
# Check we get an exception if we try to restart a workflow that doesn't exist
options.restart = True
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = False
# Now actually run the workflow
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Check we get an exception if we try to run without restart on an existing job store
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobStoreCreationException:
pass
options.restart = True
# This loop tests the restart behavior
while i != 0:
options.useExistingOptions = random.random() > 0.5
try:
Job.Runner.startToil(firstJob, options)
i = 0
except FailedJobsException as e:
i = e.numberOfFailedJobs
# Now check that if you try to restart from here it will raise an exception
# indicating that there are no jobs remaining in the workflow.
try:
Job.Runner.startToil(firstJob, options)
self.fail()
except JobException:
pass
# self.assertTrue(e.message.endswith('left in toil workflow (workflow has finished successfully?)'))
# Now check the file is properly sorted..
with open(tempSortFile, 'r') as fileHandle:
l2 = fileHandle.readlines()
checkEqual(l, l2)
def test_hello_world(self):
system([sys.executable,
'-m', helloWorld.__name__,
'./toilTest',
'--batchSystem=mesos',
'--logLevel', getLogLevelString()])
