def GetConf(conf_file):
try:
f = open("%s" % (conf_file), "r")
str_op_conf = f.read()
op_conf = pickle.loads(str_op_conf)
f.close()
return op_conf
except Exception, e:
log("=========unknown error at GetConf--%s" % str(e), "error")
return None
def GetConf(conf_file):
try:
f = open("%s" % (conf_file), 'r')
str_op_conf = f.read()
op_conf = pickle.loads(str_op_conf)
f.close()
return op_conf
except Exception, e:
log('=========unknown error at GetConf--%s' % str(e), 'error')
return None
def FailedRecover(conffile, runtime, run_log):
try:
str_op_conf = pickle.dumps(run_log)
failed_file = "failedrecover_%s_%s" % (conffile, runtime)
folder = conf.basedir + "/failedrecover/"
os.system("mkdir -p %s" % folder)
f = open("%s%s" % (folder, failed_file), "w")
f.write(str_op_conf)
f.close()
except Exception, e:
log("unknown error at FailedRecover--%s" % str(e), "error")
return False
def FailedRecover(conffile, runtime, run_log):
try:
str_op_conf = pickle.dumps(run_log)
failed_file = "failedrecover_%s_%s" % (conffile, runtime)
folder = conf.basedir + '/failedrecover/'
os.system("mkdir -p %s" % folder)
f = open("%s%s" % (folder, failed_file), 'w')
f.write(str_op_conf)
f.close()
except Exception, e:
log('unknown error at FailedRecover--%s' % str(e), 'error')
return False
def getMaxDownloadMethod(self, currentTime, currentDay):
maxDownloadSpeed = 0
if(self.sabSpeedOverride != None and self.sabSpeedOverride.isSpeedOverridden()):
maxDownloadSpeed = self.sabSpeedOverride.getOverriddenDownloadSpeed()
else:
if (currentTime > NIGHT_TIME_START and currentTime < NIGHT_TIME_STOP):
log(currentTime, " - It's night time.")
maxDownloadSpeed = HIGH_SPEED_LIMIT
elif (currentTime > WORK_TIME_START and currentTime < WORK_TIME_STOP):
log(currentTime, " - It's work time.")
maxDownloadSpeed = HIGH_SPEED_LIMIT
else:
log(currentTime, " - We're awake and at home.")
maxDownloadSpeed = LOW_SPEED_LIMIT
# 5 and 6 = Saturday and Sunday
if ((currentDay == 5 or currentDay == 6) and not (currentTime > NIGHT_TIME_START and currentTime < NIGHT_TIME_STOP)):
log("It's the weekend (", currentDay, ") and during the day.")
maxDownloadSpeed = LOW_SPEED_LIMIT
else:
log("It's a weekday (", currentDay, ").")
return maxDownloadSpeed
def printStartupConfiguration():
log("--------------------------------")
log("Initalising SAB Speed Monitor...")
log("--------------------------------\n")
log("Jas SAB Server: ", JAS_SAB)
log("Rich SAB Server: ", RIC_SAB)
log("High speed limit: ", HIGH_SPEED_LIMIT, "kb/s - Times: ", NIGHT_TIME_START, "->", NIGHT_TIME_STOP, " and ", WORK_TIME_START, "->", WORK_TIME_STOP)
log("Low speed limit: ", LOW_SPEED_LIMIT, "kb/s at all other times\n\n")
def EtlSched(etlclass, conffile, timerange, debugmode, job_name, process):
conf_model_path = conf.basedir + "/etl_conf"
etl_model_path = conf.basedir + "/etl"
pdbc_model_path = conf.basedir + "/pdbc"
if sys.path.count(conf_model_path) == 0:
sys.path.append(conf_model_path)
if sys.path.count(etl_model_path) == 0:
sys.path.append(etl_model_path)
if sys.path.count(pdbc_model_path) == 0:
sys.path.append(pdbc_model_path)
db_object_dict = {"loading": {}, "Extraction": {}}
for db_type in conf.db_info.keys():
tmp_db = imp.load_source("tmp_db", pdbc_model_path + "/" + conf.db_info[db_type]["model"])
if tmp_db.__dict__.has_key(conf.db_info[db_type]["classname"]):
tmp_db_class = tmp_db.__dict__[conf.db_info[db_type]["classname"]]
if conf.db_info[db_type]["support"] == "Loading":
db_object_dict["loading"][db_type] = tmp_db_class(log)
elif conf.db_info[db_type]["support"] == "Extraction":
db_object_dict["Extraction"][db_type] = tmp_db_class(log)
else:
db_object_dict["Extraction"][db_type] = tmp_db_class(log)
db_object_dict["loading"][db_type] = tmp_db_class(log)
etl_conf = imp.load_source("etl_conf", conf_model_path + "/" + conffile + ".py")
timelist = []
if etl_conf.etl_op.get("run_mode", "day") == "day":
timelist = ProTimeRange(timerange)
elif etl_conf.etl_op.get("run_mode", "day") == "hour":
timelist = ProTimeRangeHour(timerange, etl_conf.etl_op.get("delay_hours", 2))
else:
log("[%s] unknown run_mode %s" % (conffile, etl_conf.etl_op.get("run_mode", "day")), "error")
return False
etl_res = True
for runtime in timelist:
etl_op = copy.deepcopy(etl_conf.etl_op)
global_args = {}
global_args["conffile"] = conffile
global_args["tmpdir"] = conf.tmp_dir + "/" + conffile
global_args["tmp_dir"] = global_args["tmpdir"]
global_args["date"] = runtime[:8]
global_args["hour"] = runtime[8:10] or "23"
run_log = {}
RunLogInit(run_log, etl_op, runtime, conffile)
for job in etl_op["jobs"]:
if job_name:
if job["job_name"] != job_name:
continue
for i in range(len(job["analysis"])):
if job["analysis"][i]["db_type"] in ["hive", "hivecli"]:
job["analysis"][i]["pre_sql"].insert(0, "set mapred.job.queue.name=%s" % conf.mapred_queue)
if job["analysis"][i]["db_type"] in ["mapred"]:
job["analysis"][i]["sql"] = (
job["analysis"][i]["sql"] + " -jobconf mapred.job.queue.name=%s " % conf.mapred_queue
)
action_list = []
action_list.extend(job["analysis"])
action_list.extend(job["transform"])
action_list.extend(job["loading"])
for action in action_list:
action["job_name"] = job["job_name"]
action.update(global_args)
etlclass = action.get("etl_class_name", "")
if process:
if action["step_name"] == process:
etlmodel = imp.load_source("cometl", etl_model_path + "/" + etlclass + ".py")
cometl = etlmodel.__dict__[etlclass](action, db_object_dict, log, conffile, debugmode)
if cometl.run():
log(
"[%s-%s-%s-%s] success" % (conffile, job["job_name"], action["step_name"], runtime),
"info",
)
return (True, "")
else:
log(
"[%s-%s-%s-%s] failed" % (conffile, job["job_name"], action["step_name"], runtime),
"error",
)
# op_conf={"etlclass":etlclass,"conffile":conffile,"timerange":runtime,"process":action["step_name"],"debugmode":debugmode}
# FailedRecover(op_conf)
# dqsvalue = {'type':'job','title':'ETL','program':'%s_%s_%s' % (conffile,action["step_name"],runtime),'state':0,'msg':'%s_%s_%s_etl_failed' % (conffile,action["step_name"],runtime)}
# DqsAlarm.DQSAlarm('job',dqsvalue)
return (False, "%s_%s_%s_%s" % (conffile, job["job_name"], action["step_name"], runtime))
else:
if run_log[job["job_name"]][action["step_name"]] == "ok":
continue
etlmodel = imp.load_source("cometl", etl_model_path + "/" + etlclass + ".py")
cometl = etlmodel.__dict__[etlclass](action, db_object_dict, log, conffile, debugmode)
if cometl.run():
log("[%s-%s-%s-%s] success" % (conffile, job["job_name"], action["step_name"], runtime), "info")
run_log[job["job_name"]][action["step_name"]] = "ok"
else:
run_log[job["job_name"]][action["step_name"]] = "failed"
log("[%s-%s-%s-%s] failed" % (conffile, job["job_name"], action["step_name"], runtime), "error")
# op_conf={"etlclass":etlclass,"conffile":conffile,"timerange":runtime,"process":action["step_name"],"debugmode":debugmode}
if job_name:
return (False, "%s_%s_%s_%s" % (conffile, job["job_name"], action["step_name"], runtime))
else:
FailedRecover(conffile, runtime, run_log)
# dqsvalue = {'type':'job','title':'ETL','program':'%s_%s_%s_%s' % (conffile,job['job_name'],action["step_name"],runtime),'state':0,'msg':'%s_%s_%s_etl_failed' % (conffile,action["step_name"],runtime)}
# DqsAlarm.DQSAlarm('job',dqsvalue)
return (False, "%s_%s_%s_%s" % (conffile, job["job_name"], action["step_name"], runtime))
return (True, "")
def startControllingSabSpeeds(self):
while True :
currentTime = datetime.datetime.now().time()
currentDay = datetime.datetime.now().weekday()
try:
jasSABSpeed = getCurrentDownloadSpeed(JAS_SAB, JAS_SAB_API_KEY)
except IOError:
log("**Exception**", currentTime, "I/O error trying to access Jas's SAB.")
time.sleep(TIME_BETWEEN_SAB_SPEED_CHECKS_IF_EXCEPTION)
continue
try:
richSABSpeed = getCurrentDownloadSpeed(RIC_SAB, RIC_SAB_API_KEY)
except IOError:
log("**Exception**", currentTime, "I/O error trying to access Rich's SAB.")
time.sleep(TIME_BETWEEN_SAB_SPEED_CHECKS_IF_EXCEPTION)
continue
log("Jas SAB current speed:", jasSABSpeed, "kb/s\t\tRich SAB current speed:", richSABSpeed, "kb/s")
maxDownloadSpeed = self.getMaxDownloadMethod(currentTime, currentDay)
log("MaxSpeed: ", maxDownloadSpeed)
if (self.isSabActive(jasSABSpeed) and self.isSabActive(richSABSpeed)):
log("Both servers idle. Standby.")
elif (self.isSabActive(richSABSpeed)):
log("Set Rich's server to", maxDownloadSpeed, "kb/s")
setNewDownloadSpeed(RIC_SAB, RIC_SAB_API_KEY, maxDownloadSpeed)
elif (self.isSabActive(jasSABSpeed)):
log("Set Jas's server to", maxDownloadSpeed, "kb/s")
setNewDownloadSpeed(JAS_SAB, JAS_SAB_API_KEY, maxDownloadSpeed)
else:
log("Aggregate", maxDownloadSpeed, "kb/s evenly across both servers")
aggregatedSpeed = maxDownloadSpeed / 2
setNewDownloadSpeed(RIC_SAB, RIC_SAB_API_KEY, aggregatedSpeed)
setNewDownloadSpeed(JAS_SAB, JAS_SAB_API_KEY, aggregatedSpeed)
print("\n")
time.sleep(TIME_BETWEEN_SAB_SPEED_CHECKS)
def EtlSched(etlclass, conffile, timerange, debugmode, job_name, process):
conf_model_path = conf.basedir + '/etl_conf'
etl_model_path = conf.basedir + '/etl'
pdbc_model_path = conf.basedir + '/pdbc'
if sys.path.count(conf_model_path) == 0:
sys.path.append(conf_model_path)
if sys.path.count(etl_model_path) == 0:
sys.path.append(etl_model_path)
if sys.path.count(pdbc_model_path) == 0:
sys.path.append(pdbc_model_path)
db_object_dict = {'loading': {}, 'Extraction': {}}
for db_type in conf.db_info.keys():
tmp_db = imp.load_source(
'tmp_db', pdbc_model_path + '/' + conf.db_info[db_type]['model'])
if tmp_db.__dict__.has_key(conf.db_info[db_type]['classname']):
tmp_db_class = tmp_db.__dict__[conf.db_info[db_type]['classname']]
if conf.db_info[db_type]['support'] == 'Loading':
db_object_dict['loading'][db_type] = tmp_db_class(log)
elif conf.db_info[db_type]['support'] == 'Extraction':
db_object_dict['Extraction'][db_type] = tmp_db_class(log)
else:
db_object_dict['Extraction'][db_type] = tmp_db_class(log)
db_object_dict['loading'][db_type] = tmp_db_class(log)
etl_conf = imp.load_source('etl_conf',
conf_model_path + '/' + conffile + '.py')
timelist = []
if etl_conf.etl_op.get('run_mode', 'day') == 'day':
timelist = ProTimeRange(timerange)
elif etl_conf.etl_op.get('run_mode', 'day') == 'hour':
timelist = ProTimeRangeHour(timerange,
etl_conf.etl_op.get('delay_hours', 2))
else:
log(
'[%s] unknown run_mode %s' %
(conffile, etl_conf.etl_op.get('run_mode', 'day')), 'error')
return False
etl_res = True
for runtime in timelist:
etl_op = copy.deepcopy(etl_conf.etl_op)
global_args = {}
global_args['conffile'] = conffile
global_args['tmpdir'] = conf.tmp_dir + "/" + conffile
global_args['tmp_dir'] = global_args['tmpdir']
global_args['date'] = runtime[:8]
global_args['hour'] = runtime[8:10] or '23'
run_log = {}
RunLogInit(run_log, etl_op, runtime, conffile)
for job in etl_op['jobs']:
if job_name:
if job['job_name'] != job_name:
continue
for i in range(len(job['analysis'])):
if job['analysis'][i]['db_type'] in ['hive', 'hivecli']:
job['analysis'][i]['pre_sql'].insert(
0, "set mapred.job.queue.name=%s" % conf.mapred_queue)
if job['analysis'][i]['db_type'] in ['mapred']:
job['analysis'][i]['sql'] = job['analysis'][i][
'sql'] + " -jobconf mapred.job.queue.name=%s " % conf.mapred_queue
action_list = []
action_list.extend(job['analysis'])
action_list.extend(job['transform'])
action_list.extend(job['loading'])
for action in action_list:
action['job_name'] = job['job_name']
action.update(global_args)
etlclass = action.get('etl_class_name', '')
if process:
if action["step_name"] == process:
etlmodel = imp.load_source(
'cometl', etl_model_path + '/' + etlclass + '.py')
cometl = etlmodel.__dict__[etlclass](action,
db_object_dict,
log, conffile,
debugmode)
if cometl.run():
log(
'[%s-%s-%s-%s] success' %
(conffile, job['job_name'],
action["step_name"], runtime), 'info')
return (True, '')
else:
log(
'[%s-%s-%s-%s] failed' %
(conffile, job['job_name'],
action["step_name"], runtime), 'error')
#op_conf={"etlclass":etlclass,"conffile":conffile,"timerange":runtime,"process":action["step_name"],"debugmode":debugmode}
#FailedRecover(op_conf)
#dqsvalue = {'type':'job','title':'ETL','program':'%s_%s_%s' % (conffile,action["step_name"],runtime),'state':0,'msg':'%s_%s_%s_etl_failed' % (conffile,action["step_name"],runtime)}
#DqsAlarm.DQSAlarm('job',dqsvalue)
return (False, '%s_%s_%s_%s' %
(conffile, job['job_name'],
action["step_name"], runtime))
else:
if run_log[job['job_name']][action["step_name"]] == 'ok':
continue
etlmodel = imp.load_source(
'cometl', etl_model_path + '/' + etlclass + '.py')
cometl = etlmodel.__dict__[etlclass](action,
db_object_dict, log,
conffile, debugmode)
if cometl.run():
log(
'[%s-%s-%s-%s] success' %
(conffile, job['job_name'], action["step_name"],
runtime), 'info')
run_log[job['job_name']][action["step_name"]] = 'ok'
else:
run_log[job['job_name']][
action["step_name"]] = 'failed'
log(
'[%s-%s-%s-%s] failed' %
(conffile, job['job_name'], action["step_name"],
runtime), 'error')
#op_conf={"etlclass":etlclass,"conffile":conffile,"timerange":runtime,"process":action["step_name"],"debugmode":debugmode}
if job_name:
return (False, '%s_%s_%s_%s' %
(conffile, job['job_name'],
action["step_name"], runtime))
else:
FailedRecover(conffile, runtime, run_log)
#dqsvalue = {'type':'job','title':'ETL','program':'%s_%s_%s_%s' % (conffile,job['job_name'],action["step_name"],runtime),'state':0,'msg':'%s_%s_%s_etl_failed' % (conffile,action["step_name"],runtime)}
#DqsAlarm.DQSAlarm('job',dqsvalue)
return (False, '%s_%s_%s_%s' %
(conffile, job['job_name'],
action["step_name"], runtime))
return (True, '')
