def GetExecutable():
configuration = GetConfiguration()
(machine, machineEntry, sourceBaseDir) = simlib.GetLocalEnvironment()
configPath = simlib.BuildPath(simenv.CONFIGS_PATH, configuration)
if not (simlib.FileExists(configPath)):
fatal(
"configuration '%s', which has path '%s' does not exist or is not readable"
% (configuration, configPath))
exe = simlib.BuildPath(simenv.CACTUS_PATH, 'exe',
'cactus_%s' % configuration)
if not (simlib.FileExists(exe)):
fatal(
"Executable %s for configuration %s does not exist or is not readable"
% (exe, configuration))
submitScript = simlib.BuildPath(configPath, "SubmitScript")
if not (simlib.FileExists(submitScript)):
warning("empty submit script for configuration %s" % configuration)
submitScript = None
runScript = simlib.BuildPath(configPath, "RunScript")
if not (simlib.FileExists(runScript)):
fatal("Error: empty/missing run script for configuration %s" %
configuration)
return (exe, submitScript, runScript)
def CreateRestartSkeleton(simulationName):
(machine, machineEntry, sourceBaseDir) = simlib.GetLocalEnvironment()
basedir = simlib.GetBaseDir(machineEntry)
try:
simlib.MakeDirs(basedir)
except OSError, e:
fatal(
"could not access simulation base directory %s for reading and writing: %s"
% (basedir, e))
def CreateSimulationId(simulationName):
(machine, machineEntry, sourceBaseDir) = simlib.GetLocalEnvironment()
hostname = machineEntry.hostname
# make sure fd is closed.
fd = simenv.popen('whoami')
user = fd.read().strip()
fd.close()
tt = time.localtime()
timestamp = "%4d.%02d.%02d-%02d.%02d.%02d" % (
tt.tm_year, tt.tm_mon, tt.tm_mday, tt.tm_hour, tt.tm_min, tt.tm_sec)
pid = os.getpid()
simulation_id = "simulation-%s-%s-%s-%s-%s-%s" % (
simulationName, machine, hostname, user, timestamp, pid)
return simulation_id
def GetExecHost(restart):
DefineDatabase = simsubs.DefineDatabase()
(machine, machineEntry, sourceBaseDir) = simlib.GetLocalEnvironment()
job_id = restart.GetJobId()
job_status = 'U'
if job_id != -1:
job_status = GetJobStatus(job_id)
if job_status != 'R':
warning("Job is not running, cannot retreive exechost")
return None
simlib.VerifyKeys(machineEntry, ['exechost', 'exechostpattern'])
DefineDatabase.Set('JOB_ID', job_id)
DefineDatabase.Set('USER', machineEntry.user)
DefineDatabase.Set('SIMULATION_NAME', restart.SimulationName)
exechost = DefineDatabase.SubAll(machineEntry.GetKey('exechost'))
exechostpattern = DefineDatabase.SubAll(
machineEntry.GetKey('exechostpattern'))
output, rc = simlib.ExecuteCommand(exechost, True)
rx = re.compile(exechostpattern, re.MULTILINE)
matches = rx.search(output)
if rc or matches == None:
warning("Unable to retrieve exechost using pattern %s" %
exechostpattern)
return None
return matches.group(1)
def GetJobStatus(job_id):
# TODO: there needs to be documentation stating what the possible
# return values are
DefineDatabase = simsubs.DefineDatabase()
(machine, machineEntry, sourceBaseDir) = simlib.GetLocalEnvironment()
simlib.VerifyKeys(machineEntry, [
'getstatus', 'queuedpattern', 'runningpattern', 'statuspattern', 'user'
])
status_command = machineEntry.GetKey('getstatus')
status_pattern = machineEntry.GetKey('statuspattern')
queued_pattern = machineEntry.GetKey('queuedpattern')
running_pattern = machineEntry.GetKey('runningpattern')
holding_pattern = machineEntry.GetKey('holdingpattern')
user = machineEntry.GetKey('user')
DefineDatabase.Set('USER', user)
DefineDatabase.Set('JOB_ID', job_id)
status_command = DefineDatabase.SubAll(status_command)
status_pattern = DefineDatabase.SubAll(status_pattern)
queued_pattern = DefineDatabase.SubAll(queued_pattern)
running_pattern = DefineDatabase.SubAll(running_pattern)
holding_pattern = DefineDatabase.SubAll(holding_pattern)
#capture output.
output, ret = simlib.ExecuteCommand(status_command, output=True)
# U == unknown?
status = 'U'
if ret != None:
lines = output.split("\n")
#first, lets see if the job is still in the queue, regardless of whether or not we can determine the job status
# TODO: counting lines doesn't help detecting errors
InQueue = lines.count(job_id) > 0
matched = []
for line in lines:
matches = re.search(status_pattern, line)
if matches != None:
# queued_pattern
if re.search(queued_pattern, line):
status = 'Q'
matched.append(queued_pattern)
# running_pattern
if re.search(running_pattern, line):
status = 'R'
matched.append(running_pattern)
# holding_pattern
if holding_pattern != None:
if re.search(holding_pattern, line):
status = 'H'
matched.append(holding_pattern)
# TODO: matches is set in the last loop iteration only; this if
# statement is bogus
if matches > 1:
# TODO: output a better error message; the list of patterns
# alone is not useful because patterns are difficult to read;
# better would be also what state the patterns are for
fatal("multiple status patterns matched: %s" % matched)
if InQueue and status == 'U':
status = 'E'
if status == 'U':
warning("job status is U")
dprint("queue status return code is: %d" % ret)
dprint("queue status output is:")
for line in lines:
dprint(" lines=[%s]" % line)
dprint("patterns are:")
dprint(" status_pattern=[%s]" % status_pattern)
dprint(" queued_pattern=[%s]" % queued_pattern)
dprint(" running_pattern=[%s]" % running_pattern)
dprint(" holding_pattern=[%s]" % holding_pattern)
return status