def register_gratia(self, name):
Gratia.RegisterReporter(name)
try:
slurm_version = self.get_slurm_version()
except Exception as e:
DebugPrint(0, "Unable to get SLURM version: %s" % str(e))
raise
Gratia.RegisterService("SLURM", slurm_version)
Gratia.setProbeBatchManager("slurm")
def register_gratia(self):
"""Register in Gratia the Reporter (gratia probe), ReporterLibrary (Gratia library version)
and the Service (input)
:return:
"""
Gratia.RegisterReporter(self.probe_name)
try:
input_version = self.get_version()
except SystemExit:
raise
except KeyboardInterrupt:
raise
except Exception as e:
DebugPrint(0, "Unable to get input version: %s" % str(e))
raise
# TODO: check the meaning of RegisterReporter vs RegisterService
Gratia.RegisterService(self._probeinput.get_name(), input_version)
if p.returncode != 0:
raise Exception("Unable to invoke %s" % cmd)
name, version = output.split()
return version
def register_gratia(self, name):
Gratia.RegisterReporter(name)
try:
slurm_version = self.get_slurm_version()
except Exception, e:
DebugPrint(0, "Unable to get SLURM version: %s" % str(e))
raise
Gratia.RegisterService("SLURM", slurm_version)
Gratia.setProbeBatchManager("slurm")
class SlurmCheckpoint(object):
"""Read and write a checkpoint file
If class is instantiated without a filename, class works as expected but
data is not stored to disk
"""
_val = None
_fp = None
def __init__(self, target=None):
"""
Create a checkpoint file
def main():
# We need the logger variable in the exception handler.
# So we create it here.
logger = logging.getLogger('DCacheAggregator')
# Ignore hangup signals. We shouldn't die just because our parent
# shell logs out.
signal.signal(signal.SIGHUP, signal.SIG_IGN)
# Try to catch common signals and send email before we die
signal.signal(signal.SIGINT, warn_of_signal)
signal.signal(signal.SIGQUIT, warn_of_signal)
signal.signal(signal.SIGTERM, warn_of_signal)
try:
# Tell Gratia what versions we are using.
# CHRIS: is there a way to automate the version extraction
# using the pkg_resource package?
Gratia.RegisterReporterLibrary("psycopg2", "2.0.6")
#Gratia.RegisterReporterLibrary( "SQLAlchemy", "0.4.1" )
rev = Gratia.ExtractCvsRevision("$Revision: 1.13 $")
tag = Gratia.ExtractCvsRevision("$Name: $")
Gratia.RegisterReporter("dCacheBillingAggregator.py",
str(rev) + " (tag " + str(tag) + ")")
# BRIAN: attempt to pull the dCache version from RPM.
version = "UNKNOWN"
try:
version = os.popen("rpm -q --qf '%{VERSION}-%{RELEASE}' " \
"dcache-server").read()
except:
pass
Gratia.RegisterService("dCache", version)
# Initialize gratia before attempting to read its config file.
Gratia.Initialize()
# Extract the configuration information into local variables.
myconf = dCacheProbeConfig()
# Get the name of the directory where we are to store the log files.
logDir = myconf.get_LogFolder()
# Make sure that the logging directory is present
if not os.path.isdir(logDir):
os.mkdir(logDir, 0755)
logFileName = os.path.join(logDir, "dcacheTransfer.log")
# Set up an alarm to send an email if the program terminates.
termSubject = "dCache-transfer probe is going down"
termMessage = "The dCache transfer probe for Gratia has " + \
"terminated.\nPlease check the logfile\n\n " + \
logFileName + \
"\n\nfor the cause.\n"
terminationAlarm = Alarm(myconf.get_EmailServerHost(),
myconf.get_EmailFromAddress(),
myconf.get_EmailToList(), termSubject,
termMessage, 0, 0, False)
# Set up the logger with a suitable format
hdlr = RotatingFileHandler(logFileName, 'a', 512000, 10)
formatter = logging.Formatter('%(asctime)s %(levelname)s %(message)s')
hdlr.setFormatter(formatter)
logger.addHandler(hdlr)
logger.setLevel(myconf.get_AggrLogLevel())
logger.info("starting " + ProgramName)
stopFileName = myconf.get_StopFileName()
updateFreq = float(myconf.get_UpdateFrequency())
logger.warn("update freq = %.2f" % updateFreq)
# Create the aggregator instance that we will use.
dataDir = myconf.get_DataFolder()
aggregator = DCacheAggregator(myconf, dataDir)
# If profiling was requested, turn it on.
profiling = sys.argv.count('-profile') > 0
if profiling:
profiler = hotshot.Profile("profile.dat")
logger.info("Enabling Profiling")
# Now aggregate new records, then sleep, until somebody creates
# the stop file...
while 1:
# Make sure we (still) have a connection to Gratia.
if (not TestContainer.isTest()
): # no need in that during self test
Gratia.Maintenance()
if profiling:
profiler.run("aggregator.sendBillingInfoRecordsToGratia()")
else:
try:
aggregator.sendBillingInfoRecordsToGratia()
except TestContainer.SimInterrupt:
logger.info("BillingRecSimulator.SimInterrupt caught, " \
"restarting")
aggregator = DCacheAggregator(myconf, dataDir)
continue
# Are we are shutting down?
if os.path.exists(stopFileName):
break
if TestContainer.isTest():
break
logger.warn("sleeping for = %.2f seconds" % updateFreq)
sleep_check(updateFreq, stopFileName)
# If we are profiling, print the results...
if profiling:
profiler.close()
stats = hotshot.stats.load("profile.dat")
stats.sort_stats('time', 'calls')
stats.print_stats()
logger.warn(ProgramName + " stop file detected.")
except (KeyboardInterrupt, SystemExit):
raise
except:
# format the traceback into a string
tblist = traceback.format_exception(sys.exc_type, sys.exc_value,
sys.exc_traceback)
msg = ProgramName + " caught an exception:\n" + "".join(tblist)
print msg
logger.error(msg)
TestContainer.dumpStatistics(logger)
# shut down the logger to make sure nothing is lost.
logger.critical(ProgramName + " shutting down.")
logging.shutdown()
# try to send an email warning of the shutdown.
if terminationAlarm != None:
terminationAlarm.event()
sys.exit(1)