def main():
parser = parser_with_common_options()
options = parser.parse_args()
set_logging_from_options(options)
config = Config()
config.setOptions(options)
config.jobStore = config.jobStore[5:] if config.jobStore.startswith('file:') else config.jobStore
# ':' means an aws/google jobstore; use the old (broken?) method
if ':' in config.jobStore:
jobStore = Toil.resumeJobStore(config.jobStore)
logger.info("Starting routine to kill running jobs in the toil workflow: %s", config.jobStore)
# TODO: This behaviour is now broken: https://github.com/DataBiosphere/toil/commit/a3d65fc8925712221e4cda116d1825d4a1e963a1
batchSystem = Toil.createBatchSystem(jobStore.config) # Should automatically kill existing jobs, so we're good.
for jobID in batchSystem.getIssuedBatchJobIDs(): # Just in case we do it again.
batchSystem.killBatchJobs(jobID)
logger.info("All jobs SHOULD have been killed")
# otherwise, kill the pid recorded in the jobstore
else:
pid_log = os.path.join(os.path.abspath(config.jobStore), 'pid.log')
with open(pid_log, 'r') as f:
pid2kill = f.read().strip()
try:
os.kill(int(pid2kill), signal.SIGKILL)
logger.info("Toil process %s successfully terminated." % str(pid2kill))
except OSError:
logger.error("Toil process %s could not be terminated." % str(pid2kill))
raise
def main():
parser = parser_with_common_options(jobstore_option=True)
parser.add_argument(
"jobID",
nargs=1,
help=
"The job store id of a job within the provided jobstore to run by itself."
)
parser.add_argument(
"--printJobInfo",
nargs=1,
help=
"Return information about this job to the user including preceding jobs, "
"inputs, outputs, and runtime from the last known run.")
options = parser.parse_args()
set_logging_from_options(options)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
if options.printJobInfo:
printContentsOfJobStore(jobStorePath=config.jobStore,
nameOfJob=options.printJobInfo)
# TODO: Option to print list of successor jobs
# TODO: Option to run job within python debugger, allowing step through of arguments
# idea would be to have option to import pdb and set breakpoint at the start of the user's code
jobID = options.jobID[0]
logger.debug(f"Running the following job locally: {jobID}")
workerScript(jobStore, config, jobID, jobID, redirectOutputToLogFile=False)
logger.debug(f"Finished running: {jobID}")
def main():
parser = getBasicOptionParser()
parser.add_argument(
"jobStore",
type=str,
help=
"The location of the job store used by the workflow whose jobs should "
"be killed." + jobStoreLocatorHelp)
parser.add_argument("--version", action='version', version=version)
options = parseBasicOptions(parser)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
logger.info(
"Starting routine to kill running jobs in the toil workflow: %s",
config.jobStore)
####This behaviour is now broken
batchSystem = Toil.createBatchSystem(
jobStore.config
) #This should automatically kill the existing jobs.. so we're good.
for jobID in batchSystem.getIssuedBatchJobIDs(
): #Just in case we do it again.
batchSystem.killBatchJobs(jobID)
logger.info("All jobs SHOULD have been killed")
def main():
parser = getBasicOptionParser()
parser.add_argument("jobStore", type=str,
help="The location of the job store to delete. " + jobStoreLocatorHelp)
parser.add_argument("--version", action='version', version=version)
config = Config()
config.setOptions(parseBasicOptions(parser))
logger.info("Attempting to delete the job store")
jobStore = Toil.getJobStore(config.jobStore)
jobStore.destroy()
logger.info("Successfully deleted the job store")
def main():
""" Reports stats on the workflow, use with --stats option to toil.
"""
parser = getBasicOptionParser()
initializeOptions(parser)
options = parseBasicOptions(parser)
checkOptions(options, parser)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
stats = getStats(jobStore)
collatedStatsTag = processData(jobStore.config, stats)
reportData(collatedStatsTag, options)
def main():
""" Reports stats on the workflow, use with --stats option to toil.
"""
parser = getBasicOptionParser()
initializeOptions(parser)
options = parseBasicOptions(parser)
checkOptions(options, parser)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
stats = getStats(jobStore)
collatedStatsTag = processData(jobStore.config, stats)
reportData(collatedStatsTag, options)
def main():
parser = getBasicOptionParser()
parser.add_argument(
"jobStore",
type=str,
help="The location of the job store used by the workflow." +
jobStoreLocatorHelp)
parser.add_argument("--localFilePath",
nargs=1,
help="Location to which to copy job store files.")
parser.add_argument("--fetch",
nargs="+",
help="List of job-store files to be copied locally."
"Use either explicit names (i.e. 'data.txt'), or "
"specify glob patterns (i.e. '*.txt')")
parser.add_argument(
"--listFilesInJobStore",
help="Prints a list of the current files in the jobStore.")
parser.add_argument(
"--fetchEntireJobStore",
help="Copy all job store files into a local directory.")
parser.add_argument(
"--useSymlinks",
help="Creates symlink 'shortcuts' of files in the localFilePath"
" instead of hardlinking or copying, where possible. If this is"
" not possible, it will copy the files (shutil.copyfile()).")
parser.add_argument("--version", action='version', version=version)
# Load the jobStore
options = parseBasicOptions(parser)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
logger.debug("Connected to job store: %s", config.jobStore)
if options.fetch:
# Copy only the listed files locally
logger.debug("Fetching local files: %s", options.fetch)
fetchJobStoreFiles(jobStore=jobStore, options=options)
elif options.fetchEntireJobStore:
# Copy all jobStore files locally
logger.debug("Fetching all local files.")
options.fetch = "*"
fetchJobStoreFiles(jobStore=jobStore, options=options)
if options.listFilesInJobStore:
# Log filenames and create a file containing these names in cwd
printContentsOfJobStore(jobStorePath=options.jobStore)
def testMultipleJobsPerWorkerStats(self):
"""
Tests case where multiple jobs are run on 1 worker to insure that all jobs report back their data
"""
options = Job.Runner.getDefaultOptions(self._getTestJobStorePath())
options.clean = 'never'
options.stats = True
Job.Runner.startToil(RunTwoJobsPerWorker(), options)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
stats = getStats(jobStore)
collatedStats = processData(jobStore.config, stats)
self.assertTrue(len(collatedStats.job_types) == 2,
"Some jobs are not represented in the stats")
def testMultipleJobsPerWorkerStats(self):
"""
Tests case where multiple jobs are run on 1 worker to insure that all jobs report back their data
"""
options = Job.Runner.getDefaultOptions(self._getTestJobStorePath())
options.clean = 'never'
options.stats = True
Job.Runner.startToil(RunTwoJobsPerWorker(), options)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
stats = getStats(jobStore)
collatedStats = processData(jobStore.config, stats)
self.assertTrue(len(collatedStats.job_types) == 2,
"Some jobs are not represented in the stats")
def main():
"""
This is a Toil pipeline to transfer TCGA data into an S3 Bucket
Data is pulled down with Genetorrent and transferred to S3 via S3AM.
"""
# Define Parser object and add to toil
def existing_file(fname):
"""
Argparse type for an existing file
"""
if not os.path.isfile(fname):
raise ValueError("Invalid file: " + str(fname))
return fname
parser = argparse.ArgumentParser(
description=main.__doc__,
formatter_class=argparse.RawTextHelpFormatter)
parser.add_argument(
'--sudo',
dest='sudo',
default=None,
action='store_true',
help=
'Docker usually needs sudo to execute locally, but not when running Mesos or when '
'the user is a member of a Docker group.')
Job.Runner.addToilOptions(parser)
parser.add_argument('datafiles',
nargs='+',
help='FASTA input',
type=existing_file)
args = parser.parse_args()
assert args.jobStore is not None
config = Config()
config.setOptions(args)
# Store inputs from argparse
inputs = {'sudo': args.sudo}
datafiles = [os.path.abspath(d) for d in args.datafiles]
# Start Pipeline
options = Job.Runner.getDefaultOptions("./toilWorkflow")
Job.Runner.startToil(Job.wrapJobFn(start_batch, datafiles, inputs),
options)
def main():
parser = getBasicOptionParser()
parser.add_argument("jobStore", type=str,
help="The location of the job store used by the workflow whose jobs should "
"be killed." + jobStoreLocatorHelp)
parser.add_argument("--version", action='version', version=version)
options = parseBasicOptions(parser)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
logger.info("Starting routine to kill running jobs in the toil workflow: %s", config.jobStore)
####This behaviour is now broken
batchSystem = Toil.createBatchSystem(jobStore.config) #This should automatically kill the existing jobs.. so we're good.
for jobID in batchSystem.getIssuedBatchJobIDs(): #Just in case we do it again.
batchSystem.killBatchJobs(jobID)
logger.info("All jobs SHOULD have been killed")
def main():
parser = getBasicOptionParser()
parser.add_argument("jobStore",
type=str,
help="The location of the job store to delete. " +
jobStoreLocatorHelp)
parser.add_argument("--version", action='version', version=version)
config = Config()
config.setOptions(parseBasicOptions(parser))
try:
jobStore = Toil.getJobStore(config.jobStore)
jobStore.resume()
jobStore.destroy()
logger.info("Successfully deleted the job store: %s" % config.jobStore)
except NoSuchJobStoreException:
logger.info("Failed to delete the job store: %s is non-existent" %
config.jobStore)
except:
logger.info("Failed to delete the job store: %s" % config.jobStore)
raise
def main() -> None:
"""Reports stats on the workflow, use with --stats option to toil."""
parser = parser_with_common_options()
add_stats_options(parser)
options = parser.parse_args()
for c in options.categories.split(","):
if c.strip() not in category_choices:
raise ValueError(f'{c} not in {category_choices}!')
options.categories = [
x.strip().lower() for x in options.categories.split(",")
]
set_logging_from_options(options)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
stats = getStats(jobStore)
collatedStatsTag = processData(jobStore.config, stats)
reportData(collatedStatsTag, options)
def main():
parser = getBasicOptionParser()
parser.add_argument(
"jobStore",
type=str,
help="The location of the job store used by the workflow." +
jobStoreLocatorHelp)
parser.add_argument("jobID",
nargs=1,
help="The job store id of a job "
"within the provided jobstore to run by itself.")
parser.add_argument(
"--printJobInfo",
nargs=1,
help="Return information about this job to the user"
" including preceding jobs, inputs, outputs, and runtime"
" from the last known run.")
parser.add_argument("--version", action='version', version=version)
# Parse options
options = parseBasicOptions(parser)
config = Config()
config.setOptions(options)
# Load the job store
jobStore = Toil.resumeJobStore(config.jobStore)
if options.printJobInfo:
printContentsOfJobStore(jobStorePath=options.jobStore,
nameOfJob=options.printJobInfo)
# TODO: Option to print list of successor jobs
# TODO: Option to run job within python debugger, allowing step through of arguments
# idea would be to have option to import pdb and set breakpoint at the start of the user's code
# Run the job locally
jobID = options.jobID[0]
logger.debug("Going to run the following job locally: %s", jobID)
workerScript(jobStore, config, jobID, jobID, redirectOutputToLogFile=False)
logger.debug("Ran the following job locally: %s", jobID)
def main() -> None:
parser = parser_with_common_options()
options = parser.parse_args()
set_logging_from_options(options)
config = Config()
config.setOptions(options)
job_store_type, _ = Toil.parseLocator(config.jobStore)
if job_store_type != 'file':
# Remote (aws/google) jobstore; use the old (broken?) method
job_store = Toil.resumeJobStore(config.jobStore)
logger.info("Starting routine to kill running jobs in the toil workflow: %s", config.jobStore)
# TODO: This behaviour is now broken: https://github.com/DataBiosphere/toil/commit/a3d65fc8925712221e4cda116d1825d4a1e963a1
# There's no guarantee that the batch system in use can enumerate
# running jobs belonging to the job store we've attached to. And
# moreover we don't even bother trying to kill the leader at its
# recorded PID, even if it is a local process.
batch_system = Toil.createBatchSystem(job_store.config) # Should automatically kill existing jobs, so we're good.
for job_id in batch_system.getIssuedBatchJobIDs(): # Just in case we do it again.
batch_system.killBatchJobs([job_id])
logger.info("All jobs SHOULD have been killed")
else:
# otherwise, kill the pid recorded in the jobstore.
# TODO: We assume thnis is a local PID.
job_store = Toil.resumeJobStore(config.jobStore)
assert isinstance(job_store, FileJobStore), "Need a FileJobStore which has a sharedFilesDir"
pid_log = os.path.join(job_store.sharedFilesDir, 'pid.log')
with open(pid_log) as f:
pid_to_kill = f.read().strip()
try:
os.kill(int(pid_to_kill), signal.SIGTERM)
logger.info("Toil process %s successfully terminated." % str(pid_to_kill))
except OSError:
logger.error("Toil process %s could not be terminated." % str(pid_to_kill))
raise
def main():
"""Reports the state of the toil.
"""
##########################################
#Construct the arguments.
##########################################
parser = getBasicOptionParser()
parser.add_argument("jobStore", type=str,
help="The location of a job store that holds the information about the "
"workflow whose status is to be reported on." + jobStoreLocatorHelp)
parser.add_argument("--verbose", dest="verbose", action="store_true",
help="Print loads of information, particularly all the log files of \
jobs that failed. default=%(default)s",
default=False)
parser.add_argument("--failIfNotComplete", dest="failIfNotComplete", action="store_true",
help="Return exit value of 1 if toil jobs not all completed. default=%(default)s",
default=False)
parser.add_argument("--version", action='version', version=version)
options = parseBasicOptions(parser)
logger.info("Parsed arguments")
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
##########################################
#Do some checks.
##########################################
logger.info("Checking if we have files for Toil")
assert options.jobStore is not None
config = Config()
config.setOptions(options)
##########################################
#Survey the status of the job and report.
##########################################
jobStore = Toil.resumeJobStore(config.jobStore)
try:
rootJob = jobStore.loadRootJob()
except JobException:
print('The root job of the job store is absent, the workflow completed successfully.',
file=sys.stderr)
sys.exit(0)
toilState = ToilState(jobStore, rootJob )
# The first element of the toilState.updatedJobs tuple is the jobWrapper we want to inspect
totalJobs = set(toilState.successorCounts.keys()) | \
{jobTuple[0] for jobTuple in toilState.updatedJobs}
failedJobs = [ job for job in totalJobs if job.remainingRetryCount == 0 ]
print('There are %i active jobs, %i parent jobs with children, and %i totally failed jobs '
'currently in %s.' % (len(toilState.updatedJobs), len(toilState.successorCounts),
len(failedJobs), config.jobStore), file=sys.stderr)
if options.verbose: #Verbose currently means outputting the files that have failed.
for job in failedJobs:
if job.logJobStoreFileID is not None:
with job.getLogFileHandle(jobStore) as logFileHandle:
logStream(logFileHandle, job.jobStoreID, logger.warn)
else:
print('Log file for job %s is absent.' % job.jobStoreID, file=sys.stderr)
if len(failedJobs) == 0:
print('There are no failed jobs to report.', file=sys.stderr)
if (len(toilState.updatedJobs) + len(toilState.successorCounts)) != 0 and \
options.failIfNotComplete:
sys.exit(1)
def main():
"""Reports the state of the toil.
"""
##########################################
#Construct the arguments.
##########################################
parser = getBasicOptionParser()
parser.add_argument("jobStore", type=str,
help="The location of a job store that holds the information about the "
"workflow whose status is to be reported on." + jobStoreLocatorHelp)
parser.add_argument("--verbose", dest="verbose", action="store_true",
help="Print loads of information, particularly all the log files of \
jobs that failed. default=%(default)s",
default=False)
parser.add_argument("--failIfNotComplete", dest="failIfNotComplete", action="store_true",
help="Return exit value of 1 if toil jobs not all completed. default=%(default)s",
default=False)
parser.add_argument("--version", action='version', version=version)
options = parseBasicOptions(parser)
logger.info("Parsed arguments")
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
##########################################
#Do some checks.
##########################################
logger.info("Checking if we have files for Toil")
assert options.jobStore is not None
config = Config()
config.setOptions(options)
##########################################
#Survey the status of the job and report.
##########################################
jobStore = Toil.resumeJobStore(config.jobStore)
try:
rootJob = jobStore.loadRootJob()
except JobException:
print('The root job of the job store is absent, the workflow completed successfully.',
file=sys.stderr)
sys.exit(0)
toilState = ToilState(jobStore, rootJob )
# The first element of the toilState.updatedJobs tuple is the jobGraph we want to inspect
totalJobs = set(toilState.successorCounts.keys()) | \
{jobTuple[0] for jobTuple in toilState.updatedJobs}
failedJobs = [ job for job in totalJobs if job.remainingRetryCount == 0 ]
print('There are %i active jobs, %i parent jobs with children, and %i totally failed jobs '
'currently in %s.' % (len(toilState.updatedJobs), len(toilState.successorCounts),
len(failedJobs), config.jobStore), file=sys.stderr)
if options.verbose: #Verbose currently means outputting the files that have failed.
for job in failedJobs:
if job.logJobStoreFileID is not None:
with job.getLogFileHandle(jobStore) as logFileHandle:
logStream(logFileHandle, job.jobStoreID, logger.warn)
else:
print('Log file for job %s is absent.' % job.jobStoreID, file=sys.stderr)
if len(failedJobs) == 0:
print('There are no failed jobs to report.', file=sys.stderr)
if (len(toilState.updatedJobs) + len(toilState.successorCounts)) != 0 and \
options.failIfNotComplete:
sys.exit(1)
def main():
"""Reports the state of the toil.
"""
##########################################
#Construct the arguments.
##########################################
parser = getBasicOptionParser()
parser.add_argument(
"jobStore",
type=str,
help="The location of a job store that holds the information about the "
"workflow whose status is to be reported on." + jobStoreLocatorHelp)
parser.add_argument(
"--verbose",
dest="verbose",
action="store_true",
help="Print loads of information, particularly all the log files of \
jobs that failed. default=%(default)s",
default=False)
parser.add_argument(
"--failIfNotComplete",
dest="failIfNotComplete",
action="store_true",
help=
"Return exit value of 1 if toil jobs not all completed. default=%(default)s",
default=False)
parser.add_argument("--version", action='version', version=version)
options = parseBasicOptions(parser)
logger.info("Parsed arguments")
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
##########################################
#Do some checks.
##########################################
logger.info("Checking if we have files for Toil")
assert options.jobStore is not None
config = Config()
config.setOptions(options)
##########################################
#Survey the status of the job and report.
##########################################
jobStore = Toil.resumeJobStore(config.jobStore)
try:
rootJob = jobStore.loadRootJob()
except JobException:
print(
'The root job of the job store is absent, the workflow completed successfully.',
file=sys.stderr)
sys.exit(0)
def traverseGraph(jobGraph):
foundJobStoreIDs = set()
totalJobs = []
def inner(jobGraph):
if jobGraph.jobStoreID in foundJobStoreIDs:
return
foundJobStoreIDs.add(jobGraph.jobStoreID)
totalJobs.append(jobGraph)
# Traverse jobs in stack
for jobs in jobGraph.stack:
for successorJobStoreID in map(lambda x: x.jobStoreID, jobs):
if (successorJobStoreID not in foundJobStoreIDs
and jobStore.exists(successorJobStoreID)):
inner(jobStore.load(successorJobStoreID))
# Traverse service jobs
for jobs in jobGraph.services:
for serviceJobStoreID in map(lambda x: x.jobStoreID, jobs):
if jobStore.exists(serviceJobStoreID):
assert serviceJobStoreID not in foundJobStoreIDs
foundJobStoreIDs.add(serviceJobStoreID)
totalJobs.append(jobStore.load(serviceJobStoreID))
inner(jobGraph)
return totalJobs
logger.info('Traversing the job graph. This may take a couple minutes.')
totalJobs = traverseGraph(rootJob)
failedJobs = []
hasChildren = []
hasServices = []
services = []
currentlyRunnning = []
for job in totalJobs:
if job.logJobStoreFileID is not None:
failedJobs.append(job)
if job.stack:
hasChildren.append(job)
elif job.remainingRetryCount != 0 and job.logJobStoreFileID != 0 and job.command:
# The job has no children, hasn't failed, and has a command to run. This indicates that the job is
# likely currently running, or at least could be run.
currentlyRunnning.append(job)
if job.services:
hasServices.append(job)
if job.startJobStoreID or job.terminateJobStoreID or job.errorJobStoreID:
# these attributes are only set in service jobs
services.append(job)
logger.info(
'There are %i unfinished jobs, %i parent jobs with children, %i jobs with services, %i services, '
'and %i totally failed jobs currently in %s.' %
(len(totalJobs), len(hasChildren), len(hasServices), len(services),
len(failedJobs), config.jobStore))
if currentlyRunnning:
logger.info('These %i jobs are currently active: %s',
len(currentlyRunnning),
' \n'.join(map(str, currentlyRunnning)))
if options.verbose: #Verbose currently means outputting the files that have failed.
if failedJobs:
msg = "Outputting logs for the %i failed jobs" % (len(failedJobs))
msg += ": %s" % ", ".join(
(str(failedJob) for failedJob in failedJobs))
for jobNode in failedJobs:
job = jobStore.load(jobNode.jobStoreID)
msg += "\n=========> Failed job %s \n" % jobNode
with job.getLogFileHandle(jobStore) as fH:
msg += fH.read()
msg += "<=========\n"
print(msg)
else:
print('There are no failed jobs to report.', file=sys.stderr)
if totalJobs and options.failIfNotComplete:
exit(
1
) # when the workflow is complete, all jobs will have been removed from job store
def main():
"""Reports the state of a Toil workflow."""
parser = getBasicOptionParser()
parser.add_argument("jobStore", type=str,
help="The location of a job store that holds the information about the "
"workflow whose status is to be reported on." + jobStoreLocatorHelp)
parser.add_argument("--failIfNotComplete", action="store_true",
help="Return exit value of 1 if toil jobs not all completed. default=%(default)s",
default=False)
parser.add_argument("--noAggStats", dest="stats", action="store_false",
help="Do not print overall, aggregate status of workflow.",
default=True)
parser.add_argument("--printDot", action="store_true",
help="Print dot formatted description of the graph. If using --jobs will "
"restrict to subgraph including only those jobs. default=%(default)s",
default=False)
parser.add_argument("--jobs", nargs='+',
help="Restrict reporting to the following jobs (allows subsetting of the report).",
default=None)
parser.add_argument("--printPerJobStats", action="store_true",
help="Print info about each job. default=%(default)s",
default=False)
parser.add_argument("--printLogs", action="store_true",
help="Print the log files of jobs (if they exist). default=%(default)s",
default=False)
parser.add_argument("--printChildren", action="store_true",
help="Print children of each job. default=%(default)s",
default=False)
parser.add_argument("--version", action='version', version=version)
options = parseBasicOptions(parser)
logger.info("Parsed arguments")
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
config = Config()
config.setOptions(options)
jobStore = Toil.resumeJobStore(config.jobStore)
##########################################
# Gather the jobs to report
##########################################
# Gather all jobs in the workflow in jobsToReport
if options.jobs == None:
rootJob = fetchRootJob(jobStore)
logger.info('Traversing the job graph gathering jobs. This may take a couple of minutes.')
jobsToReport = traverseJobGraph(rootJob, jobStore)
# Only gather jobs specified in options.jobs
else:
jobsToReport = fetchUserJobs(jobStore, jobs=options.jobs)
##########################################
# Report on the jobs
##########################################
jobStats = report_on_jobs(jobsToReport, jobStore, options)
hasChildren = jobStats['hasChildren']
readyToRun = jobStats['readyToRun']
zombies = jobStats['zombies']
hasServices = jobStats['hasServices']
services = jobStats['services']
hasLogFile = jobStats['hasLogFile']
properties = jobStats['properties']
childNumber = jobStats['childNumber']
if options.printPerJobStats:
printAggregateJobStats(jobsToReport, properties, childNumber)
if options.printLogs:
printJobLog(jobsToReport, jobStore)
if options.printChildren:
printJobChildren(jobsToReport)
if options.printDot:
print_dot_chart(jobsToReport, jobStore_name=config.jobStore)
if options.stats:
print('Of the %i jobs considered, '
'there are %i jobs with children, '
'%i jobs ready to run, '
'%i zombie jobs, '
'%i jobs with services, '
'%i services, '
'and %i jobs with log files currently in %s.' %
(len(jobsToReport), len(hasChildren), len(readyToRun), len(zombies),
len(hasServices), len(services), len(hasLogFile), config.jobStore))
if len(jobsToReport) > 0 and options.failIfNotComplete:
# Upon workflow completion, all jobs will have been removed from job store
exit(1)
def main():
"""Reports the state of the toil.
"""
##########################################
#Construct the arguments.
##########################################
parser = getBasicOptionParser()
parser.add_argument("jobStore", type=str,
help="The location of a job store that holds the information about the "
"workflow whose status is to be reported on." + jobStoreLocatorHelp)
parser.add_argument("--verbose", dest="verbose", action="store_true",
help="Print loads of information, particularly all the log files of \
jobs that failed. default=%(default)s",
default=False)
parser.add_argument("--failIfNotComplete", dest="failIfNotComplete", action="store_true",
help="Return exit value of 1 if toil jobs not all completed. default=%(default)s",
default=False)
parser.add_argument("--version", action='version', version=version)
options = parseBasicOptions(parser)
logger.info("Parsed arguments")
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
##########################################
#Do some checks.
##########################################
logger.info("Checking if we have files for Toil")
assert options.jobStore is not None
config = Config()
config.setOptions(options)
##########################################
#Survey the status of the job and report.
##########################################
jobStore = Toil.resumeJobStore(config.jobStore)
try:
rootJob = jobStore.loadRootJob()
except JobException:
print('The root job of the job store is absent, the workflow completed successfully.',
file=sys.stderr)
sys.exit(0)
def traverseGraph(jobGraph):
foundJobStoreIDs = set()
totalJobs = []
def inner(jobGraph):
if jobGraph.jobStoreID in foundJobStoreIDs:
return
foundJobStoreIDs.add(jobGraph.jobStoreID)
totalJobs.append(jobGraph)
# Traverse jobs in stack
for jobs in jobGraph.stack:
for successorJobStoreID in [x.jobStoreID for x in jobs]:
if (successorJobStoreID not in foundJobStoreIDs and jobStore.exists(successorJobStoreID)):
inner(jobStore.load(successorJobStoreID))
# Traverse service jobs
for jobs in jobGraph.services:
for serviceJobStoreID in [x.jobStoreID for x in jobs]:
if jobStore.exists(serviceJobStoreID):
assert serviceJobStoreID not in foundJobStoreIDs
foundJobStoreIDs.add(serviceJobStoreID)
totalJobs.append(jobStore.load(serviceJobStoreID))
inner(jobGraph)
return totalJobs
logger.info('Traversing the job graph. This may take a couple minutes.')
totalJobs = traverseGraph(rootJob)
failedJobs = []
hasChildren = []
hasServices = []
services = []
currentlyRunnning = []
for job in totalJobs:
if job.logJobStoreFileID is not None:
failedJobs.append(job)
if job.stack:
hasChildren.append(job)
elif job.remainingRetryCount != 0 and job.logJobStoreFileID != 0 and job.command:
# The job has no children, hasn't failed, and has a command to run. This indicates that the job is
# likely currently running, or at least could be run.
currentlyRunnning.append(job)
if job.services:
hasServices.append(job)
if job.startJobStoreID or job.terminateJobStoreID or job.errorJobStoreID:
# these attributes are only set in service jobs
services.append(job)
logger.info('There are %i unfinished jobs, %i parent jobs with children, %i jobs with services, %i services, '
'and %i totally failed jobs currently in %s.' %
(len(totalJobs), len(hasChildren), len(hasServices), len(services), len(failedJobs), config.jobStore))
if currentlyRunnning:
logger.info('These %i jobs are currently active: %s',
len(currentlyRunnning), ' \n'.join(map(str, currentlyRunnning)))
if options.verbose: #Verbose currently means outputting the files that have failed.
if failedJobs:
msg = "Outputting logs for the %i failed jobs" % (len(failedJobs))
msg += ": %s" % ", ".join((str(failedJob) for failedJob in failedJobs))
for jobNode in failedJobs:
job = jobStore.load(jobNode.jobStoreID)
msg += "\n=========> Failed job %s \n" % jobNode
with job.getLogFileHandle(jobStore) as fH:
msg += fH.read()
msg += "<=========\n"
print(msg)
else:
print('There are no failed jobs to report.', file=sys.stderr)
if totalJobs and options.failIfNotComplete:
exit(1) # when the workflow is complete, all jobs will have been removed from job store
def main() -> None:
"""Reports the state of a Toil workflow."""
parser = parser_with_common_options()
parser.add_argument(
"--failIfNotComplete",
action="store_true",
help=
"Return exit value of 1 if toil jobs not all completed. default=%(default)s",
default=False)
parser.add_argument(
"--noAggStats",
dest="stats",
action="store_false",
help="Do not print overall, aggregate status of workflow.",
default=True)
parser.add_argument(
"--printDot",
action="store_true",
help=
"Print dot formatted description of the graph. If using --jobs will "
"restrict to subgraph including only those jobs. default=%(default)s",
default=False)
parser.add_argument(
"--jobs",
nargs='+',
help=
"Restrict reporting to the following jobs (allows subsetting of the report).",
default=None)
parser.add_argument("--printPerJobStats",
action="store_true",
help="Print info about each job. default=%(default)s",
default=False)
parser.add_argument(
"--printLogs",
action="store_true",
help="Print the log files of jobs (if they exist). default=%(default)s",
default=False)
parser.add_argument("--printChildren",
action="store_true",
help="Print children of each job. default=%(default)s",
default=False)
options = parser.parse_args()
set_logging_from_options(options)
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
config = Config()
config.setOptions(options)
try:
status = ToilStatus(config.jobStore, options.jobs)
except NoSuchJobStoreException:
print('No job store found.')
return
except JobException: # Workflow likely complete, user informed in ToilStatus()
return
jobStats = status.report_on_jobs()
# Info to be reported.
hasChildren = jobStats['hasChildren']
readyToRun = jobStats['readyToRun']
zombies = jobStats['zombies']
hasServices = jobStats['hasServices']
services = jobStats['services']
hasLogFile = jobStats['hasLogFile']
properties = jobStats['properties']
childNumber = jobStats['childNumber']
if options.printPerJobStats:
status.printAggregateJobStats(properties, childNumber)
if options.printLogs:
status.printJobLog()
if options.printChildren:
status.printJobChildren()
if options.printDot:
status.print_dot_chart()
if options.stats:
print('Of the %i jobs considered, '
'there are %i jobs with children, '
'%i jobs ready to run, '
'%i zombie jobs, '
'%i jobs with services, '
'%i services, '
'and %i jobs with log files currently in %s.' %
(len(status.jobsToReport), len(hasChildren), len(readyToRun),
len(zombies), len(hasServices), len(services), len(hasLogFile),
status.jobStore))
if len(status.jobsToReport) > 0 and options.failIfNotComplete:
# Upon workflow completion, all jobs will have been removed from job store
exit(1)