class HawqUtil:
def __init__(self, conf):
self.conf = conf
self.hdfsUtil = HdfsUtil(conf)
self.pgUtil = PostgresUtil(conf)
self.segDirs = self.getSegmentDirs()
def printReport(self):
printInfo("Getting database names from HAWQ")
dbNames = self.getDatabaseNames()
printInfo("Getting quotas from these databases %s" % (dbNames))
for db in dbNames:
self.printDatabaseQuota(db)
self.printTopKLargestDatabases(dbNames)
def getDatabaseNames(self):
names = queryPostgres(self.conf.get(Config.HAWQ_PORT), self.conf.get(Config.HAWQ_USERNAME), self.conf.get(Config.HAWQ_METADATA_DB), "SELECT DISTINCT datname FROM pg_database", split=False)
retval = []
for name in names.split():
if not name in self.conf.get(Config.HAWQ_SYSTEM_DB_BLACKLIST):
retval.append(name)
return retval
def getDatabaseOID(self, name):
oids = queryPostgres(self.conf.get(Config.HAWQ_PORT), self.conf.get(Config.HAWQ_USERNAME), self.conf.get(Config.HAWQ_METADATA_DB), "SELECT oid FROM pg_database WHERE datname = '%s'" % (name))
if not oids is None and len(oids) == 1:
return oids[0]
elif oids is None:
printError("Database %s not found" % (name))
else:
printError("Received %i OIDs, expecting 1" %(len(oids)))
return None
def getTableOID(self, database, table):
oids = queryPostgres(self.conf.get(Config.HAWQ_PORT), self.conf.get(Config.HAWQ_USERNAME), database, "SELECT oid FROM pg_class WHERE relname = '%s'" % (table))
if len(oids) == 1:
return oids[0]
else:
printError("Received %i OIDs, expecting 1" %(len(oids)))
return None
def getSegmentDirs(self):
cmd = "hdfs dfs -ls %s | grep gpseg | awk '{print $8}'" % (self.conf.get(Config.HAWQ_HDFS_DIR))
dirs = getCommandOutput(cmd).split()
if len(dirs) == 0:
printError("Failed to get any segment directories from HDFS")
sys.exit(1)
return dirs
def getSchemaTables(self, database):
output = queryPostgres(port, username, database, "SELECT table_schema, table_name FROM information_schema.tables", split=False)
schemaTableMap = dict()
for record in output.split('\n'):
(schema, table) = record.strip().replace('|', '').split()
if not schema in self.schemaBlackList:
try:
schemaTableMap[schema] = schemaTableMap[schema] + [table]
except KeyError:
schemaTableMap[schema] = [table]
return schemaTableMap
def printDatabaseQuota(self, db):
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
printInfo("Getting quota status for database %s" % (db))
hdfsDBDirs = []
for segDir in self.segDirs:
hdfsDBDirs.append("%s/16385/%s" % (segDir, dbOID))
quotas = self.hdfsUtil.getSpaceQuotas(hdfsDBDirs)
self.__printDBQuotaInserts(db, quotas)
row = namedtuple('Row', ['Database', 'Directory', 'Quota', 'Remaining', 'QuotaHR', 'RemainingHR'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(remainingQuota) if remainingQuota != 'inf' else remainingQuota
toPrint.append(row(db, directory, quota, remainingQuota, quotaHR, remainingQuotaHR))
pprinttable(toPrint)
def __printDBQuotaInserts(self, db, quotas):
for (directory, quota, remainingQuota) in quotas:
row = HawqDBQuotaRow()
row.database = db
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def getDatabaseSize(self, db):
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
dbDir = "%s/*/16385/%s" % (self.conf.get(Config.HAWQ_HDFS_DIR), dbOID)
sizes = self.hdfsUtil.getDirSizes([dbDir])
sum = 0
for (dir, size) in sizes:
sum += size
return (db, sum)
else:
return None
def printTopKLargestDatabases(self, dbNames):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s largest HAWQ databases" % (k))
dbSizes = []
for db in dbNames:
tDbSize = self.getDatabaseSize(db)
if not tDbSize is None:
dbSizes.append(tDbSize)
if len(dbSizes) == 0:
printInfo("No HAWQ databases found in HDFS")
return
dbSizes.sort(key=operator.itemgetter(1), reverse=True)
if len(dbSizes) > k:
dbSizes = dbSizes[0:k]
self.__printTopKLargestDatabasesInserts(dbSizes)
# print sizes
row = namedtuple('Row', ['Database', 'Size', 'SizeHR'])
toPrint = []
for (db, size) in dbSizes:
sizeHR = bytes2human(size)
toPrint.append(row(db, str(size), str(sizeHR)))
pprinttable(toPrint)
def __printTopKLargestDatabasesInserts(self, dbSizes):
for (db, size) in dbSizes:
row = HawqDBSizeRow()
row.database = db
row.size = size
self.pgUtil.writeInsert(row)
def setDatabaseQuota(self, db, quota):
if db == self.conf.get(Config.HAWQ_DB_BLACKLIST).split():
printError("Database %s is in the blacklist. Remove to set quota" % (db))
return
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
printInfo("Setting quota for %s to %s bytes" % (db, quota))
hdfsDBDirs = []
for segDir in self.segDirs:
hdfsDBDirs.append("%s/16385/%s" % (segDir, dbOID))
self.hdfsUtil.setSpaceQuotas(hdfsDBDirs, quota)
else:
sys.exit(1)
def clearDatabaseQuota(self, db):
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
printInfo("Clearing quota for database %s" % (db))
hdfsDBDirs = []
for segDir in self.segDirs:
hdfsDBDirs.append("%s/16385/%s" % (segDir, dbOID))
self.hdfsUtil.clearSpaceQuotas(hdfsDBDirs)
else:
sys.exit(1)
class UserUtil:
def __init__(self, conf):
self.conf = conf
self.hdfsUtil = HdfsUtil(conf)
self.pgUtil = PostgresUtil(conf)
def printReport(self):
self.printUserSpaceQuotas()
self.printUserINodeQuotas()
self.printTopKSpaceUsers()
self.printTopKINodeUsers()
def printUserSpaceQuotas(self):
printInfo("Getting space quota status for users")
quotas = self.hdfsUtil.getSpaceQuotas(self.getUserDirectories())
if len(quotas) == 0:
printInfo("No user directories found in HDFS")
return
quotas.sort()
self.__printUserSpaceQuotasInserts(quotas)
row = namedtuple(
'Row',
['Directory', 'Quota', 'Remaining', 'QuotaHR', 'RemainingHR'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
quotaHR = bytes2human(quota) if quota != 'none' else quota
# Sometimes the remaining quota is negative...
if remainingQuota != 'inf':
if long(remainingQuotaHR) < 0:
remainingQuotaHR = "-" + bytes2human(-long(remainingQuota))
else:
remainingQuotaHR = bytes2human(remainingQuota)
else:
remainingQuotaHR = remainingQuota
toPrint.append(
row(directory, quota, remainingQuota, quotaHR,
remainingQuotaHR))
pprinttable(toPrint)
def __printUserSpaceQuotasInserts(self, quotas):
for (directory, quota, remainingQuota) in quotas:
row = UserSpaceQuotaRow()
row.username = directory[6:]
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def printUserINodeQuotas(self):
printInfo("Getting inode quota status for users")
quotas = self.hdfsUtil.getINodeQuotas(self.getUserDirectories())
if len(quotas) == 0:
printInfo("No user directories found in HDFS")
return
quotas.sort()
self.__printUserINodeQuotasInserts(quotas)
row = namedtuple('Row', ['Directory', 'Quota', 'Remaining'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
toPrint.append(row(directory, quota, remainingQuota))
pprinttable(toPrint)
def __printUserINodeQuotasInserts(self, quotas):
for (directory, quota, remainingQuota) in quotas:
row = UserINodeQuotaRow()
row.username = directory[6:]
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def printUserSpaceQuota(self, user):
printInfo("Getting space quota status for user %s" % (user))
quotas = self.hdfsUtil.getSpaceQuotas(["/user/%s" % (user)])
if len(quotas) == 0:
printInfo("Directory for user %s not found in HDFS" % (quotas))
return
row = namedtuple(
'Row',
['Directory', 'Quota', 'Remaining', 'QuotaHR', 'RemainingHR'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(
remainingQuota) if remainingQuota != 'inf' else remainingQuota
toPrint.append(
row(directory, quota, remainingQuota, quotaHR,
remainingQuotaHR))
pprinttable(toPrint)
def printUserINodeQuota(self, user):
printInfo("Getting inode quota status for user %s" % (user))
quotas = self.hdfsUtil.getINodeQuotas(["/user/%s" % (user)])
if len(quotas) == 0:
printInfo("Directory for user %s not found in HDFS" % (quotas))
return
row = namedtuple('Row', ['Directory', 'Quota', 'Remaining'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
toPrint.append(row(directory, quota, remainingQuota))
pprinttable(toPrint)
def printTopKSpaceUsers(self):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s space users" % (k))
sizes = self.hdfsUtil.getDirSizes(['/user'])
if len(sizes) == 0:
printInfo("No user directories found in HDFS")
return
sizes.sort(key=operator.itemgetter(1), reverse=True)
if len(sizes) > k:
sizes = sizes[0:k]
self.__printTopKSpaceInserts(sizes)
row = namedtuple('Row', ['User', 'Size', 'SizeHR'])
toPrint = []
for (dir, size) in sizes:
sizeHR = bytes2human(size)
toPrint.append(row(dir, str(size), str(sizeHR)))
pprinttable(toPrint)
def __printTopKSpaceInserts(self, sizes):
for (dir, size) in sizes:
row = UserSpaceSizeRow()
row.username = dir[6:]
row.dir = dir
row.size = size
self.pgUtil.writeInsert(row)
def printTopKINodeUsers(self):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s inode users" % (k))
counts = self.hdfsUtil.getINodeCounts(self.getUserDirectories())
if len(counts) == 0:
printInfo("No user directories found in HDFS")
return
counts.sort(key=operator.itemgetter(1), reverse=True)
if len(counts) > k:
counts = counts[0:k]
self.__printTopKINodeUsersInserts(counts)
row = namedtuple('Row', ['User', 'INodes'])
toPrint = []
for (dir, count) in counts:
toPrint.append(row(dir, str(count)))
pprinttable(toPrint)
def __printTopKINodeUsersInserts(self, counts):
for (dir, count) in counts:
row = UserINodeSizeRow()
row.username = dir[6:]
row.dir = dir
row.size = count
self.pgUtil.writeInsert(row)
def setUserSpaceQuota(self, user, quota):
if user == self.conf.get(Config.USER_DIR_BLACKLIST).split():
printError("User %s is in the blacklist. Remove to set quota" %
(db))
return
self.hdfsUtil.setSpaceQuotas(["/user/%s" % (user)], quota)
def clearUserSpaceQuota(self, user):
self.hdfsUtil.clearSpaceQuotas(["/user/%s" % (user)])
def setUserINodeQuota(self, user, quota):
if user == self.conf.get(Config.USER_DIR_BLACKLIST).split():
printError("User %s is in the blacklist. Remove to set quota" %
(db))
return
self.hdfsUtil.setINodeQuotas(["/user/%s" % (user)], quota)
def clearUserINodeQuota(self, user):
self.hdfsUtil.clearINodeQuotas(["/user/%s" % (user)])
def getUserDirectories(self):
return self.hdfsUtil.listDirs(['/user'])
class HawqUtil:
def __init__(self, conf):
self.conf = conf
self.hdfsUtil = HdfsUtil(conf)
self.pgUtil = PostgresUtil(conf)
self.segDirs = self.getSegmentDirs()
def printReport(self):
printInfo("Getting database names from HAWQ")
dbNames = self.getDatabaseNames()
printInfo("Getting quotas from these databases %s" % (dbNames))
for db in dbNames:
self.printDatabaseQuota(db)
self.printTopKLargestDatabases(dbNames)
def getDatabaseNames(self):
names = queryPostgres(self.conf.get(Config.HAWQ_PORT),
self.conf.get(Config.HAWQ_USERNAME),
self.conf.get(Config.HAWQ_METADATA_DB),
"SELECT DISTINCT datname FROM pg_database",
split=False)
retval = []
for name in names.split():
if not name in self.conf.get(Config.HAWQ_SYSTEM_DB_BLACKLIST):
retval.append(name)
return retval
def getDatabaseOID(self, name):
oids = queryPostgres(
self.conf.get(Config.HAWQ_PORT),
self.conf.get(Config.HAWQ_USERNAME),
self.conf.get(Config.HAWQ_METADATA_DB),
"SELECT oid FROM pg_database WHERE datname = '%s'" % (name))
if not oids is None and len(oids) == 1:
return oids[0]
elif oids is None:
printError("Database %s not found" % (name))
else:
printError("Received %i OIDs, expecting 1" % (len(oids)))
return None
def getTableOID(self, database, table):
oids = queryPostgres(
self.conf.get(Config.HAWQ_PORT),
self.conf.get(Config.HAWQ_USERNAME), database,
"SELECT oid FROM pg_class WHERE relname = '%s'" % (table))
if len(oids) == 1:
return oids[0]
else:
printError("Received %i OIDs, expecting 1" % (len(oids)))
return None
def getSegmentDirs(self):
cmd = "hdfs dfs -ls %s | grep gpseg | awk '{print $8}'" % (
self.conf.get(Config.HAWQ_HDFS_DIR))
dirs = getCommandOutput(cmd).split()
if len(dirs) == 0:
printError("Failed to get any segment directories from HDFS")
sys.exit(1)
return dirs
def getSchemaTables(self, database):
output = queryPostgres(
port,
username,
database,
"SELECT table_schema, table_name FROM information_schema.tables",
split=False)
schemaTableMap = dict()
for record in output.split('\n'):
(schema, table) = record.strip().replace('|', '').split()
if not schema in self.schemaBlackList:
try:
schemaTableMap[schema] = schemaTableMap[schema] + [table]
except KeyError:
schemaTableMap[schema] = [table]
return schemaTableMap
def printDatabaseQuota(self, db):
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
printInfo("Getting quota status for database %s" % (db))
hdfsDBDirs = []
for segDir in self.segDirs:
hdfsDBDirs.append("%s/16385/%s" % (segDir, dbOID))
quotas = self.hdfsUtil.getSpaceQuotas(hdfsDBDirs)
self.__printDBQuotaInserts(db, quotas)
row = namedtuple('Row', [
'Database', 'Directory', 'Quota', 'Remaining', 'QuotaHR',
'RemainingHR'
])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(
remainingQuota
) if remainingQuota != 'inf' else remainingQuota
toPrint.append(
row(db, directory, quota, remainingQuota, quotaHR,
remainingQuotaHR))
pprinttable(toPrint)
def __printDBQuotaInserts(self, db, quotas):
for (directory, quota, remainingQuota) in quotas:
row = HawqDBQuotaRow()
row.database = db
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def getDatabaseSize(self, db):
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
dbDir = "%s/*/16385/%s" % (self.conf.get(
Config.HAWQ_HDFS_DIR), dbOID)
sizes = self.hdfsUtil.getDirSizes([dbDir])
sum = 0
for (dir, size) in sizes:
sum += size
return (db, sum)
else:
return None
def printTopKLargestDatabases(self, dbNames):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s largest HAWQ databases" % (k))
dbSizes = []
for db in dbNames:
tDbSize = self.getDatabaseSize(db)
if not tDbSize is None:
dbSizes.append(tDbSize)
if len(dbSizes) == 0:
printInfo("No HAWQ databases found in HDFS")
return
dbSizes.sort(key=operator.itemgetter(1), reverse=True)
if len(dbSizes) > k:
dbSizes = dbSizes[0:k]
self.__printTopKLargestDatabasesInserts(dbSizes)
# print sizes
row = namedtuple('Row', ['Database', 'Size', 'SizeHR'])
toPrint = []
for (db, size) in dbSizes:
sizeHR = bytes2human(size)
toPrint.append(row(db, str(size), str(sizeHR)))
pprinttable(toPrint)
def __printTopKLargestDatabasesInserts(self, dbSizes):
for (db, size) in dbSizes:
row = HawqDBSizeRow()
row.database = db
row.size = size
self.pgUtil.writeInsert(row)
def setDatabaseQuota(self, db, quota):
if db == self.conf.get(Config.HAWQ_DB_BLACKLIST).split():
printError("Database %s is in the blacklist. Remove to set quota" %
(db))
return
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
printInfo("Setting quota for %s to %s bytes" % (db, quota))
hdfsDBDirs = []
for segDir in self.segDirs:
hdfsDBDirs.append("%s/16385/%s" % (segDir, dbOID))
self.hdfsUtil.setSpaceQuotas(hdfsDBDirs, quota)
else:
sys.exit(1)
def clearDatabaseQuota(self, db):
dbOID = self.getDatabaseOID(db)
if not dbOID is None:
printInfo("Clearing quota for database %s" % (db))
hdfsDBDirs = []
for segDir in self.segDirs:
hdfsDBDirs.append("%s/16385/%s" % (segDir, dbOID))
self.hdfsUtil.clearSpaceQuotas(hdfsDBDirs)
else:
sys.exit(1)
class UserUtil:
def __init__(self, conf):
self.conf = conf
self.hdfsUtil = HdfsUtil(conf)
self.pgUtil = PostgresUtil(conf)
def printReport(self):
self.printUserSpaceQuotas()
self.printUserINodeQuotas()
self.printTopKSpaceUsers()
self.printTopKINodeUsers()
def printUserSpaceQuotas(self):
printInfo("Getting space quota status for users")
quotas = self.hdfsUtil.getSpaceQuotas(self.getUserDirectories())
if len(quotas) == 0:
printInfo("No user directories found in HDFS")
return
quotas.sort()
self.__printUserSpaceQuotasInserts(quotas)
row = namedtuple('Row', ['Directory', 'Quota', 'Remaining', 'QuotaHR', 'RemainingHR'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
quotaHR = bytes2human(quota) if quota != 'none' else quota
# Sometimes the remaining quota is negative...
if remainingQuota != 'inf':
if long(remainingQuotaHR) < 0:
remainingQuotaHR = "-" + bytes2human(-long(remainingQuota))
else:
remainingQuotaHR = bytes2human(remainingQuota)
else:
remainingQuotaHR = remainingQuota
toPrint.append(row(directory, quota, remainingQuota, quotaHR, remainingQuotaHR))
pprinttable(toPrint)
def __printUserSpaceQuotasInserts(self, quotas):
for (directory, quota, remainingQuota) in quotas:
row = UserSpaceQuotaRow()
row.username = directory[6:]
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def printUserINodeQuotas(self):
printInfo("Getting inode quota status for users")
quotas = self.hdfsUtil.getINodeQuotas(self.getUserDirectories())
if len(quotas) == 0:
printInfo("No user directories found in HDFS")
return
quotas.sort()
self.__printUserINodeQuotasInserts(quotas)
row = namedtuple('Row', ['Directory', 'Quota', 'Remaining'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
toPrint.append(row(directory, quota, remainingQuota))
pprinttable(toPrint)
def __printUserINodeQuotasInserts(self, quotas):
for (directory, quota, remainingQuota) in quotas:
row = UserINodeQuotaRow()
row.username = directory[6:]
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def printUserSpaceQuota(self, user):
printInfo("Getting space quota status for user %s" % (user))
quotas = self.hdfsUtil.getSpaceQuotas(["/user/%s" % (user)])
if len(quotas) == 0:
printInfo("Directory for user %s not found in HDFS" % (quotas))
return
row = namedtuple('Row', ['Directory', 'Quota', 'Remaining', 'QuotaHR', 'RemainingHR'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(remainingQuota) if remainingQuota != 'inf' else remainingQuota
toPrint.append(row(directory, quota, remainingQuota, quotaHR, remainingQuotaHR))
pprinttable(toPrint)
def printUserINodeQuota(self, user):
printInfo("Getting inode quota status for user %s" % (user))
quotas = self.hdfsUtil.getINodeQuotas(["/user/%s" % (user)])
if len(quotas) == 0:
printInfo("Directory for user %s not found in HDFS" % (quotas))
return
row = namedtuple('Row', ['Directory', 'Quota', 'Remaining'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
toPrint.append(row(directory, quota, remainingQuota))
pprinttable(toPrint)
def printTopKSpaceUsers(self):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s space users" % (k))
sizes = self.hdfsUtil.getDirSizes(['/user'])
if len(sizes) == 0:
printInfo("No user directories found in HDFS")
return
sizes.sort(key=operator.itemgetter(1), reverse=True)
if len(sizes) > k:
sizes = sizes[0:k]
self.__printTopKSpaceInserts(sizes)
row = namedtuple('Row', ['User', 'Size', 'SizeHR'])
toPrint = []
for (dir, size) in sizes:
sizeHR = bytes2human(size)
toPrint.append(row(dir, str(size), str(sizeHR)))
pprinttable(toPrint)
def __printTopKSpaceInserts(self, sizes):
for (dir, size) in sizes:
row = UserSpaceSizeRow()
row.username = dir[6:]
row.dir = dir
row.size = size
self.pgUtil.writeInsert(row)
def printTopKINodeUsers(self):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s inode users" % (k))
counts = self.hdfsUtil.getINodeCounts(self.getUserDirectories())
if len(counts) == 0:
printInfo("No user directories found in HDFS")
return
counts.sort(key=operator.itemgetter(1), reverse=True)
if len(counts) > k:
counts = counts[0:k]
self.__printTopKINodeUsersInserts(counts)
row = namedtuple('Row', ['User', 'INodes'])
toPrint = []
for (dir, count) in counts:
toPrint.append(row(dir, str(count)))
pprinttable(toPrint)
def __printTopKINodeUsersInserts(self, counts):
for (dir, count) in counts:
row = UserINodeSizeRow()
row.username = dir[6:]
row.dir = dir
row.size = count
self.pgUtil.writeInsert(row)
def setUserSpaceQuota(self, user, quota):
if user == self.conf.get(Config.USER_DIR_BLACKLIST).split():
printError("User %s is in the blacklist. Remove to set quota" % (db))
return
self.hdfsUtil.setSpaceQuotas(["/user/%s" % (user)], quota)
def clearUserSpaceQuota(self, user):
self.hdfsUtil.clearSpaceQuotas(["/user/%s" % (user)])
def setUserINodeQuota(self, user, quota):
if user == self.conf.get(Config.USER_DIR_BLACKLIST).split():
printError("User %s is in the blacklist. Remove to set quota" % (db))
return
self.hdfsUtil.setINodeQuotas(["/user/%s" % (user)], quota)
def clearUserINodeQuota(self, user):
self.hdfsUtil.clearINodeQuotas(["/user/%s" % (user)])
def getUserDirectories(self):
return self.hdfsUtil.listDirs(['/user'])
class HiveUtil:
def __init__(self, conf):
self.conf = conf
self.hdfsUtil = HdfsUtil(conf)
self.pgUtil = PostgresUtil(conf)
def printReport(self):
printInfo("Fetching contents of Hive warehouse")
hivedbdirs = self.getHiveDatabaseDirectories()
self.printDatabaseQuotas(hivedbdirs)
self.printTopKLargestDatabases(hivedbdirs)
def getHiveDatabaseDirectories(self):
hivedirs = self.hdfsUtil.listDirs(
[self.conf.get(Config.HIVE_WAREHOUSE_DIR)])
retval = []
for dir in hivedirs:
if dir.endswith(".db"):
retval.append(dir)
return retval
def printDatabaseQuota(self, db):
printInfo("Getting quota status for Hive database %s" % (db))
quotas = self.hdfsUtil.getSpaceQuotas(
["%s%s.db" % (self.conf.get(Config.HIVE_WAREHOUSE_DIR), db)])
if len(quotas) == 0:
printInfo("No Hive databases found")
return
row = namedtuple('Row', [
'Database', 'Directory', 'Quota', 'Remaining', 'QuotaHR',
'RemainingHR'
])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
dbName = directory.replace(".db", "").replace(
self.conf.get(Config.HIVE_WAREHOUSE_DIR), "")
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(
remainingQuota) if remainingQuota != 'inf' else remainingQuota
toPrint.append(
row(dbName, directory, quota, remainingQuota, quotaHR,
remainingQuotaHR))
pprinttable(toPrint)
def printDatabaseQuotas(self, hivedbdirs):
printInfo("Getting quota status for Hive databases")
hdfsDirs = []
for dir in hivedbdirs:
db = self.getDbNameFromPath(dir)
hdfsDirs.append("%s/%s.db" %
(self.conf.get(Config.HIVE_WAREHOUSE_DIR), db))
quotas = self.hdfsUtil.getSpaceQuotas(hdfsDirs)
if len(quotas) == 0:
printInfo("No Hive databases found")
return
quotas.sort()
self.__printDBQuotasInserts(quotas)
row = namedtuple('Row', [
'Database', 'Directory', 'Quota', 'Remaining', 'QuotaHR',
'RemainingHR'
])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
dbName = directory.replace(".db", "").replace(
self.conf.get(Config.HIVE_WAREHOUSE_DIR), "")
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(
remainingQuota) if remainingQuota != 'inf' else remainingQuota
toPrint.append(
row(dbName, directory, quota, remainingQuota, quotaHR,
remainingQuotaHR))
pprinttable(toPrint)
def __printDBQuotasInserts(self, quotas):
for (directory, quota, remainingQuota) in quotas:
row = HiveDBQuotaRow()
row.database = directory.replace(".db", "").replace(
self.conf.get(Config.HIVE_WAREHOUSE_DIR), "")
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def getDatabaseSize(self, dbDir):
sizes = self.hdfsUtil.getDirSizes([dbDir])
sum = 0
for (dir, size) in sizes:
sum += size
return (dbDir, sum)
def printTopKLargestDatabases(self, hivedbdirs):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s largest Hive databases" % (k))
dbSizes = []
for dbDir in hivedbdirs:
tDbSize = self.getDatabaseSize(dbDir)
if not tDbSize is None:
dbSizes.append(tDbSize)
if len(dbSizes) == 0:
printInfo("No Hive databases found in HDFS")
return
dbSizes.sort(key=operator.itemgetter(1), reverse=True)
if len(dbSizes) > k:
dbSizes = dbSizes[0:k]
self.__printTopKLargestDatabases(dbSizes)
# print sizes
row = namedtuple('Row', ['Database', 'Size', 'SizeHR'])
toPrint = []
for (db, size) in dbSizes:
sizeHR = bytes2human(size)
toPrint.append(row(db, str(size), str(sizeHR)))
pprinttable(toPrint)
def __printTopKLargestDatabases(self, dbSizes):
for (db, size) in dbSizes:
row = HiveDBSizeRow()
row.database = db
row.size = size
self.pgUtil.writeInsert(row)
def setDatabaseQuota(self, db, quota):
if db == self.conf.get(Config.HIVE_DB_BLACKLIST).split():
printError("Database %s is in the blacklist. Remove to set quota" %
(db))
return
printInfo("Setting quota for %s to %s bytes" % (db, quota))
self.hdfsUtil.setSpaceQuotas([self.getDbPathFromName(db)], quota)
def clearDatabaseQuota(self, db):
printInfo("Clearing quota for database %s" % (db))
self.hdfsUtil.clearSpaceQuotas([self.getDbPathFromName(db)])
def getDbNameFromPath(self, dir):
return dir.replace(self.conf.get(Config.HIVE_WAREHOUSE_DIR),
"").replace(".db", "")
def getDbPathFromName(self, db):
return "%s%s.db" % (self.conf.get(Config.HIVE_WAREHOUSE_DIR), db)
class HiveUtil:
def __init__(self, conf):
self.conf = conf
self.hdfsUtil = HdfsUtil(conf)
self.pgUtil = PostgresUtil(conf)
def printReport(self):
printInfo("Fetching contents of Hive warehouse")
hivedbdirs = self.getHiveDatabaseDirectories()
self.printDatabaseQuotas(hivedbdirs)
self.printTopKLargestDatabases(hivedbdirs)
def getHiveDatabaseDirectories(self):
hivedirs = self.hdfsUtil.listDirs([self.conf.get(Config.HIVE_WAREHOUSE_DIR)])
retval = []
for dir in hivedirs:
if dir.endswith(".db"):
retval.append(dir)
return retval
def printDatabaseQuota(self, db):
printInfo("Getting quota status for Hive database %s" % (db))
quotas = self.hdfsUtil.getSpaceQuotas(["%s%s.db" % (self.conf.get(Config.HIVE_WAREHOUSE_DIR), db)])
if len(quotas) == 0:
printInfo("No Hive databases found")
return;
row = namedtuple('Row', ['Database', 'Directory', 'Quota', 'Remaining', 'QuotaHR', 'RemainingHR'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
dbName = directory.replace(".db", "").replace(self.conf.get(Config.HIVE_WAREHOUSE_DIR), "")
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(remainingQuota) if remainingQuota != 'inf' else remainingQuota
toPrint.append(row(dbName, directory, quota, remainingQuota, quotaHR, remainingQuotaHR))
pprinttable(toPrint)
def printDatabaseQuotas(self, hivedbdirs):
printInfo("Getting quota status for Hive databases")
hdfsDirs = []
for dir in hivedbdirs:
db = self.getDbNameFromPath(dir)
hdfsDirs.append("%s/%s.db" % (self.conf.get(Config.HIVE_WAREHOUSE_DIR), db))
quotas = self.hdfsUtil.getSpaceQuotas(hdfsDirs)
if len(quotas) == 0:
printInfo("No Hive databases found")
return;
quotas.sort()
self.__printDBQuotasInserts(quotas)
row = namedtuple('Row', ['Database', 'Directory', 'Quota', 'Remaining', 'QuotaHR', 'RemainingHR'])
toPrint = []
for (directory, quota, remainingQuota) in quotas:
dbName = directory.replace(".db", "").replace(self.conf.get(Config.HIVE_WAREHOUSE_DIR), "")
quotaHR = bytes2human(quota) if quota != 'none' else quota
remainingQuotaHR = bytes2human(remainingQuota) if remainingQuota != 'inf' else remainingQuota
toPrint.append(row(dbName, directory, quota, remainingQuota, quotaHR, remainingQuotaHR))
pprinttable(toPrint)
def __printDBQuotasInserts(self, quotas):
for (directory, quota, remainingQuota) in quotas:
row = HiveDBQuotaRow()
row.database = directory.replace(".db", "").replace(self.conf.get(Config.HIVE_WAREHOUSE_DIR), "")
row.dir = directory
if not quota == 'none':
row.quota = int(quota)
row.quotaRemaining = int(remainingQuota)
row.quotaUsed = row.quota - row.quotaRemaining
else:
row.quota = None
row.quotaRemaining = None
row.quotaUsed = None
self.pgUtil.writeInsert(row)
def getDatabaseSize(self, dbDir):
sizes = self.hdfsUtil.getDirSizes([dbDir])
sum = 0
for (dir, size) in sizes:
sum += size
return (dbDir, sum)
def printTopKLargestDatabases(self, hivedbdirs):
k = self.conf.get(Config.REPORTER_K)
printInfo("Getting top %s largest Hive databases" % (k))
dbSizes = []
for dbDir in hivedbdirs:
tDbSize = self.getDatabaseSize(dbDir)
if not tDbSize is None:
dbSizes.append(tDbSize)
if len(dbSizes) == 0:
printInfo("No Hive databases found in HDFS")
return
dbSizes.sort(key=operator.itemgetter(1), reverse=True)
if len(dbSizes) > k:
dbSizes = dbSizes[0:k]
self.__printTopKLargestDatabases(dbSizes)
# print sizes
row = namedtuple('Row', ['Database', 'Size', 'SizeHR'])
toPrint = []
for (db, size) in dbSizes:
sizeHR = bytes2human(size)
toPrint.append(row(db, str(size), str(sizeHR)))
pprinttable(toPrint)
def __printTopKLargestDatabases(self, dbSizes):
for (db, size) in dbSizes:
row = HiveDBSizeRow()
row.database = db
row.size = size
self.pgUtil.writeInsert(row)
def setDatabaseQuota(self, db, quota):
if db == self.conf.get(Config.HIVE_DB_BLACKLIST).split():
printError("Database %s is in the blacklist. Remove to set quota" % (db))
return
printInfo("Setting quota for %s to %s bytes" % (db, quota))
self.hdfsUtil.setSpaceQuotas([self.getDbPathFromName(db)], quota)
def clearDatabaseQuota(self, db):
printInfo("Clearing quota for database %s" % (db))
self.hdfsUtil.clearSpaceQuotas([self.getDbPathFromName(db)])
def getDbNameFromPath(self, dir):
return dir.replace(self.conf.get(Config.HIVE_WAREHOUSE_DIR), "").replace(".db", "")
def getDbPathFromName(self, db):
return "%s%s.db" % (self.conf.get(Config.HIVE_WAREHOUSE_DIR), db);