def insertFileIntoHdfs(fileName):
pathFileName = '/user/' + HADOOPQA_USER + '/' + fileName
if (not(HDFS.fileExists(pathFileName))):
sourceFile = DATA_PATH + '/' + fileName
destFile = '/user/' + HADOOPQA_USER + '/' + fileName
putCmd = "dfs -put " + sourceFile + ' ' + destFile
out = Hadoop.run(putCmd)
return out
def verifyLongRunningJob(cls):
'''
Verify long running background job after it finishes
:return:
'''
ruAssert(
"Tez",
YARN.getAppFinalStateFromID(
cls._background_job_appId) == 'SUCCEEDED')
for output in cls._hdfsOutputList:
ruAssert("Tez", HDFS.fileExists(output + '/part*'))
logger.info("**** Verified long running job for Tez ****")
def verifyLongRunningJob(cls):
'''
Validate long running background job after end of all component upgrade
'''
from beaver.component.rollingupgrade.ruUpgrade import UpgradePerNode
UpgradePerNode.reportProgress(
"### Verifying Long running job for Pig ####")
# Check if the Long Running process is not None
if cls._process is not None:
# If process poll() returns None, it means the process has not finished yet.
if cls._process.poll() is None:
UpgradePerNode.reportProgress(
"### Long running job for Pig has not Finished yet. Waiting for it to complete ####"
)
# Wait for _long_running_job_timeout_secs for Job to complete
starttime = time.time()
while (starttime - time.time() <
cls._long_running_job_timeout_secs
) and cls._process.poll() is None:
time.sleep(5)
exit_code = cls._process.poll()
if exit_code is None:
logger.info("Killing Pig Long running job process '%d'" %
cls._process.pid)
Machine.killProcess(cls._process.pid)
UpgradePerNode.reportProgress(
"### [FAILED][Pig][BGJob] Long running job for Pig Failed to finish ####"
)
elif exit_code != 0:
UpgradePerNode.reportProgress(
"### [FAILED][Pig][BGJob] Long running job for Pig Failed and Exited with '%d' ####"
% exit_code)
else:
UpgradePerNode.reportProgress(
"### Long running job for Pig Finished ####")
#Check for _SUCCESS file in HDFS Path
if HDFS.fileExists(cls._hdfs_success_filepath, cls._job_user):
UpgradePerNode.reportProgress(
"### [PASSED][Pig][BGJob] Found _SUCCESS file in HDFS for Pig Long running job ####"
)
else:
UpgradePerNode.reportProgress(
"### [FAILED][Pig][BGJob] Not Found _SUCCESS file in HDFS for Pig Long running job. ####"
)
else:
UpgradePerNode.reportProgress(
"### [FAILED][Pig][BGJob] Long Running Pig Job Failed. No Process found ####"
)
def insertCSVDataViaCSVBuildLoad(cls,
csvFile,
tableName,
putIntoHDFS=True,
deleteAfterExec=True,
runInBackground=False,
user=None,
config=None,
optionAndParameter="",
env=None,
delimiter=",",
arrayDelimiter=None,
schema=None):
"""
By default, the files will be allocated under /tmp/ folder in HDFS.
"""
global ZK_ZPARENT
if Slider.isSlider():
ZK_ZPARENT = util.getPropertyValueFromConfigXMLFile(
os.path.join(Config.get('hbase', 'HBASE_CONF_DIR'),
'hbase-site.xml'), "zookeeper.znode.parent")
if Machine.isLinux():
clientjar = Machine.find(user=Machine.getAdminUser(),
host="localhost",
filepath=PHOENIX_HOME,
searchstr="phoenix-*[0-9]-client.jar",
passwd=Machine.getAdminPasswd())
else:
clientjar = Machine.find(user=Machine.getAdminUser(),
host="localhost",
filepath=PHOENIX_HOME,
searchstr="phoenix-*-client.jar",
passwd=Machine.getAdminPasswd())
if Machine.isWindows():
clientjar = (clientjar[0].strip("\\localhost")).replace("$", ":")
fileName = csvFile.split('\\')[-1]
else:
clientjar = clientjar[0]
fileName = csvFile.split('/')[-1]
# If we need to, we insert it into HDFS, since the library will take it from there.
executingUser = (HADOOPQA_USER) if user is None else user
if putIntoHDFS:
if not HDFS.fileExists('/tmp/'):
HDFS.mkdir('/tmp/')
HDFS.copyFromLocal(csvFile, '/tmp/', executingUser, config,
optionAndParameter)
hbaseConfDir = HBASE_CONF_DIR
if Slider.isSlider():
hbaseConfDir = Config.get('hbase', 'HBASE_CONF_DIR')
classpath = hbaseConfDir
finalCommand = "%s jar %s org.apache.phoenix.mapreduce.CsvBulkLoadTool --table %s --input %s" \
% (HADOOP_CMD, clientjar, tableName, '/tmp/%s' % fileName)
if schema is not None:
finalCommand = '%s -schema %s' % (finalCommand, schema)
if Machine.isWindows():
os.environ['HADOOP_USER_CLASSPATH_FIRST'] = 'true'
os.environ['HADOOP_CLASSPATH'] = classpath
if delimiter != "," or arrayDelimiter != None:
finalCommand = "%s -d `\\\"`%s`\\\" -a `\\\"`%s`\\\"" \
% (finalCommand, delimiter, arrayDelimiter.strip("'"))
finalCommand = "%s --zookeeper %s" % (finalCommand, ZK_HOST)
if runInBackground:
exit_code = 0
stdout = ''
Machine.runinbackground(
finalCommand,
env=dict(env.items() + ENVIRONMENT.items()
if env is not None else ENVIRONMENT))
else:
exit_code, stdout = Machine.run(
finalCommand,
env=dict(env.items() + ENVIRONMENT.items()
if env is not None else ENVIRONMENT))
else:
# delimiter options
if delimiter != "," or arrayDelimiter != None:
finalCommand = "%s --delimiter %s --array-delimiter %s" % (
finalCommand, delimiter, arrayDelimiter)
# ZKHosts options
finalCommand = "%s --zookeeper %s" % (finalCommand,
cls.getZKConnectString())
if runInBackground:
exit_code = 0
stdout = ''
Machine.runinbackground(
"HADOOP_CLASSPATH=%s %s" % (classpath, finalCommand),
env=dict(env.items() + ENVIRONMENT.items()
if env is not None else ENVIRONMENT))
else:
exit_code, stdout = Machine.run(
"HADOOP_CLASSPATH=%s %s" % (classpath, finalCommand),
env=dict(env.items() + ENVIRONMENT.items()
if env is not None else ENVIRONMENT))
# If selected, after insertion into HBase we will delete the csvFile from HDFS
if deleteAfterExec and not runInBackground:
# Does not work for "run in background" option
HDFS.deleteFile('/tmp/%s' % fileName, executingUser)
# return 0,""
return exit_code, stdout
