def main():
parser = getBasicOptionParser("usage: %prog [options]", "%prog 0.1")
parser.add_option("--job", dest="jobFile", help="Job file containing command to run", default="None")
parser.add_option(
"--treePointer", dest="treePointerFile", help="File containing pointer to the tree data", default="None"
)
options, args = parseBasicOptions(parser)
logger.info("Parsed the input arguments")
job = ET.parse(options.jobFile).getroot()
setLogLevel(job.attrib["log_level"])
logger.info("Parsed the job XML")
treePointer = ET.parse(options.treePointerFile).getroot()
logger.info("Parsed the tree pointer XML")
tree = ET.parse(treePointer.attrib["file"]).getroot()
logger.info("Parsed the tree XML")
for child in tree.find("children").findall("child"):
# Make the chuld tree pointer
childTreePointerFile = makeTreePointer(child.attrib["file"], getTempFile(rootDir=job.attrib["global_temp_dir"]))
# Make the child command
unbornChild = ET.SubElement(job.find("children"), "child")
command = "jobTreeTest_CommandFirst.py --treePointer %s --job JOB_FILE" % (childTreePointerFile,)
unbornChild.attrib["command"] = command
if random.random() > 0.2:
unbornChild.attrib["time"] = str(random.random() * 10)
# Make the child tree pointer
ET.SubElement(treePointer.find("children"), "child", {"file": childTreePointerFile})
job.attrib["command"] = "jobTreeTest_CommandSecond.py --treePointer %s --job JOB_FILE" % (options.treePointerFile,)
logger.info("Made new command")
fileHandle = open(options.jobFile, "w")
ET.ElementTree(job).write(fileHandle)
fileHandle.close()
logger.info("Updated the job file")
print >> sys.stderr, "Checking that we can report to std err" # These lines should end up in the logs
print "Checking that we can report to std out"
if random.random() > 0.9:
logger.info("Going to fail the job")
sys.exit(1)
logger.info("Going to pass the job done okay")
sys.exit(0)
def main():
parser = getBasicOptionParser("usage: %prog [options]", "%prog 0.1")
parser.add_option("--jobTree", dest="jobTree",
help="Directory containing the job tree to kill")
options, args = parseBasicOptions(parser)
logger.info("Parsed arguments")
assert len(args) == 0 #This program takes no arguments
assert options.jobTree != None #The jobtree should not be null
assert os.path.isdir(options.jobTree) #The job tree must exist if we are going to kill it.
logger.info("Starting routine to kill running jobs in the jobTree: %s" % options.jobTree)
config = ET.parse(os.path.join(options.jobTree, "config.xml")).getroot()
batchSystem = loadTheBatchSystem(config) #This should automatically kill the existing jobs.. so we're good.
for job in batchSystem.getIssuedJobIDs(): #Just in case we do it again.
batchSystem.killJobs(job)
logger.info("All jobs SHOULD have been killed")
def main():
sys.path += [ sys.argv[1] ]
sys.argv.remove(sys.argv[1])
#Now we can import all the stuff..
from workflow.jobTree.lib.bioio import getBasicOptionParser
from workflow.jobTree.lib.bioio import parseBasicOptions
from workflow.jobTree.lib.bioio import logger
from workflow.jobTree.lib.bioio import addLoggingFileHandler
from workflow.jobTree.lib.bioio import setLogLevel
from workflow.jobTree.lib.bioio import getTotalCpuTime
from workflow.jobTree.lib.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()
def main():
"""Reports the state of the job tree.
"""
##########################################
#Construct the arguments.
##########################################
parser = getBasicOptionParser("usage: %prog [options] \nThe colours returned indicate the state of the job.\n\
\twhite: job has not been started yet\n\
\tgrey: job is issued to batch system\n\
\tred: job failed\n\
\tblue: job has children currently being processed\n\
\tblack: job has finished and will be processed (transient state)\n\
\tdead: job is totally finished and is awaiting deletion (transient state)", "%prog 0.1")
parser.add_option("--jobTree", dest="jobTree",
help="Directory containing the job tree")
parser.add_option("--verbose", dest="verbose", action="store_true",
help="Print loads of information, particularly all the log files of errors",
default=False)
parser.add_option("--failIfNotComplete", dest="failIfNotComplete", action="store_true",
help="Return exit value of 1 if job tree jobs not all completed",
default=False)
options, args = parseBasicOptions(parser)
logger.info("Parsed arguments")
assert len(args) == 0
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
##########################################
#Do some checks.
##########################################
logger.info("Checking if we have files for job tree")
assert options.jobTree != None
assert os.path.isdir(options.jobTree) #The given job dir tree must exist.
assert os.path.isfile(os.path.join(options.jobTree, "config.xml")) #A valid job tree must contain the config gile
assert os.path.isdir(os.path.join(options.jobTree, "jobs")) #A job tree must have a directory of jobs.
assert os.path.isdir(os.path.join(options.jobTree, "tempDirDir")) #A job tree must have a directory of temporary directories (for jobs to make temp files in).
assert os.path.isdir(os.path.join(options.jobTree, "logFileDir")) #A job tree must have a directory of log files.
assert os.path.isdir(os.path.join(options.jobTree, "slaveLogFileDir")) #A job tree must have a directory of slave log files.
##########################################
#Read the total job number
##########################################
config = ET.parse(os.path.join(options.jobTree, "config.xml")).getroot()
##########################################
#Survey the status of the job and report.
##########################################
colours = {}
jobFiles = TempFileTree(config.attrib["job_file_dir"]).listFiles()
if len(jobFiles) > 0:
logger.info("Collating the colours of the job tree")
for absFileName in jobFiles:
job = parseJobFile(absFileName)
if job != None:
if not colours.has_key(job.attrib["colour"]):
colours[job.attrib["colour"]] = 0
colours[job.attrib["colour"]] += 1
else:
logger.info("There are no jobs to collate")
print "There are %i jobs currently in job tree: %s" % \
(len(jobFiles), options.jobTree)
for colour in colours.keys():
print "\tColour: %s, number of jobs: %s" % (colour, colours[colour])
if options.verbose: #Verbose currently means outputting the files that have failed.
for absFileName in jobFiles:
job = parseJobFile(absFileName)
if job != None:
if job.attrib["colour"] == "red":
if os.path.isfile(job.attrib["log_file"]):
def fn(string):
print string
logFile(job.attrib["log_file"], fn)
else:
logger.info("Log file for job %s is not present" % job.attrib["file"])
if len(jobFiles) != 0 and options.failIfNotComplete:
sys.exit(1)
#
#THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
#IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
#FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
#AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
#LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
#OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
#THE SOFTWARE.
from workflow.jobTree.lib.bioio import logger
from workflow.jobTree.lib.bioio import getBasicOptionParser
from workflow.jobTree.lib.bioio import parseBasicOptions
from workflow.jobTree.scriptTree.stack import loadPickleFile
parser = getBasicOptionParser("usage: %prog [options]", "%prog 0.1")
parser.add_option("--job", dest="jobFile",
help="Job file containing command to run")
parser.add_option("--target", dest="target",
help="File containing a pickled, wrapped instance of target classes")
options, args = parseBasicOptions(parser)
assert options.target != None
logger.info("Parsed the input arguments")
#Naughty stuff to do the import of the target we need
for className in args:
def main():
parser = getBasicOptionParser("usage: %prog [options]", "%prog 0.1")
parser.add_option("--job", dest="jobFile",
help="Job file containing command to run",
default="None")
parser.add_option("--treePointer", dest="treePointer",
help="File containing pointer to the tree data",
default="None")
options, args = parseBasicOptions(parser)
logger.info("Parsed the input arguments")
print >>sys.stderr, "Checking that we can report to std err" #These lines should end up in the logs
print "Checking that we can report to std out"
job = ET.parse(options.jobFile).getroot()
setLogLevel(job.attrib["log_level"])
logger.info("Parsed the job XML")
treePointer = ET.parse(options.treePointer).getroot()
logger.info("Parsed the tree pointer XML")
tree = ET.parse(treePointer.attrib["file"]).getroot()
logger.info("Parsed the tree XML")
i = 0
children = tree.find("children").findall("child")
if len(children) > 0:
for child in children:
#Parse the child XML tree
childTree = ET.parse(child.attrib["file"]).getroot()
i += int(childTree.attrib["count"])
else:
i = 1
tree.attrib["count"] = str(i)
logger.info("Calculated the leaf count: %i" % i)
fileHandle = open(treePointer.attrib["file"], 'w')
ET.ElementTree(tree).write(fileHandle)
fileHandle.close()
logger.info("Updated the tree file: %s" % treePointer.attrib["file"])
for childPointer in treePointer.find("children").findall("child"):
if os.path.isfile(childPointer.attrib["file"]):
os.remove(childPointer.attrib["file"])
logger.info("Removed the child pointer files")
logger.info("No need to update the job file, as we didn't make anything new!")
if random.random() > 0.9:
logger.info("Going to fail the job")
sys.exit(1)
logger.info("Going to pass the job done okay")
sys.exit(0)
def main():
"""Reports stats on the job-tree, use in conjunction with --stats options to jobTree.
"""
##########################################
#Construct the arguments.
##########################################
parser = getBasicOptionParser("usage: %prog", "%prog 0.1")
parser.add_option("--jobTree", dest="jobTree",
help="Directory containing the job tree")
parser.add_option("--outputFile", dest="outputFile", default=None,
help="File in which to write results")
options, args = parseBasicOptions(parser)
logger.info("Parsed arguments")
assert len(args) == 0
if len(sys.argv) == 1:
parser.print_help()
sys.exit(0)
##########################################
#Do some checks.
##########################################
logger.info("Checking if we have files for job tree")
if options.jobTree == None:
raise RuntimeError("You did not specify the job-tree")
if not os.path.isdir(options.jobTree):
raise RuntimeError("The given job dir tree does not exist: %s" % options.jobTree)
if not os.path.isfile(os.path.join(options.jobTree, "config.xml")):
raise RuntimeError("A valid job tree must contain the config file")
if not os.path.isfile(os.path.join(options.jobTree, "stats.xml")):
raise RuntimeError("The job-tree was run without the --stats flag, so no stats were created")
##########################################
#Read the stats and config
##########################################
config = ET.parse(os.path.join(options.jobTree, "config.xml")).getroot()
stats = ET.parse(os.path.join(options.jobTree, "stats.xml")).getroot()
##########################################
#Collate the stats and report
##########################################
def fn(element, items, itemName):
itemTimes = [ float(item.attrib["time"]) for item in items ]
itemTimes.sort()
itemClocks = [ float(item.attrib["clock"]) for item in items ]
itemClocks.sort()
itemWaits = [ float(item.attrib["time"]) - float(item.attrib["clock"]) for item in items ]
itemWaits.sort()
if len(itemTimes) == 0:
itemTimes.append(0)
itemClocks.append(0)
itemWaits.append(0)
return ET.SubElement(element, itemName, { "total_number":str(len(items)),
"total_time":str(sum(itemTimes)),
"median_time":str(itemTimes[len(itemTimes)/2]),
"average_time":str(sum(itemTimes)/len(itemTimes)),
"min_time":str(min(itemTimes)),
"max_time":str(max(itemTimes)),
"total_clock":str(sum(itemClocks)),
"median_clock":str(itemClocks[len(itemClocks)/2]),
"average_clock":str(sum(itemClocks)/len(itemClocks)),
"min_clock":str(min(itemClocks)),
"max_clock":str(max(itemClocks)),
"total_wait":str(sum(itemWaits)),
"median_wait":str(itemWaits[len(itemWaits)/2]),
"average_wait":str(sum(itemWaits)/len(itemWaits)),
"min_wait":str(min(itemWaits)),
"max_wait":str(max(itemWaits))
})
def fn2(element, containingItems, containingItemName, getFn):
itemCounts = [ len(getFn(containingItem)) for containingItem in containingItems ]
itemCounts.sort()
if len(itemCounts) == 0:
itemCounts.append(0)
element.attrib["median_number_per_%s" % containingItemName] = str(itemCounts[len(itemCounts)/2])
element.attrib["average_number_per_%s" % containingItemName] = str(float(sum(itemCounts))/len(itemCounts))
element.attrib["min_number_per_%s" % containingItemName] = str(min(itemCounts))
element.attrib["max_number_per_%s" % containingItemName] = str(max(itemCounts))
if stats.find("total_time") == None: #Hack to allow it to work on unfinished jobtrees.
ET.SubElement(stats, "total_time", { "time":"0.0", "clock":"0.0"})
collatedStatsTag = ET.Element("collated_stats", { "total_run_time":stats.find("total_time").attrib["time"],
"total_clock":stats.find("total_time").attrib["clock"],
"batch_system":config.attrib["batch_system"],
"job_time":config.attrib["job_time"],
"default_memory":config.attrib["default_memory"],
"default_cpu":config.attrib["default_cpu"],
"max_jobs":config.attrib["max_jobs"],
"max_threads":config.attrib["max_threads"] })
#Add slave info
slaves = stats.findall("slave")
fn(collatedStatsTag, slaves, "slave")
#Add job info
jobs = []
for slave in slaves:
jobs += slave.findall("job")
def fn3(slave):
return slave.findall("job")
fn2(fn(collatedStatsTag, jobs, "job"), slaves, "slave", fn3)
#Add aggregated target info
targets = []
for job in jobs:
for stack in job.findall("stack"):
targets += stack.findall("target")
def fn4(job):
targets = []
for stack in job.findall("stack"):
targets += stack.findall("target")
return targets
fn2(fn(collatedStatsTag, targets, "target"), jobs, "job", fn4)
#Get info for each target
targetNames = set()
for target in targets:
targetNames.add(target.attrib["class"])
targetTypesTag = ET.SubElement(collatedStatsTag, "target_types")
for targetName in targetNames:
targetTypes = [ target for target in targets if target.attrib["class"] == targetName ]
targetTypeTag = fn(targetTypesTag, targetTypes, targetName)
estimatedRunTimes = [ float(target.attrib["e_time"]) for target in targetTypes ]
targetTypeTag.attrib["estimated_time"] = str(sum(estimatedRunTimes)/len(estimatedRunTimes))
def prettify(elem):
"""Return a pretty-printed XML string for the Element.
"""
rough_string = ET.tostring(elem, 'utf-8')
reparsed = minidom.parseString(rough_string)
return reparsed.toprettyxml(indent=" ")
#Now dump it all out to file
if options.outputFile != None:
fileHandle = open(options.outputFile, 'w')
#ET.ElementTree(collatedStatsTag).write(fileHandle)
fileHandle.write(prettify(collatedStatsTag))
fileHandle.close()
#Now dump onto the screen
print prettify(collatedStatsTag)