def test_hbase_crashstorage_basic_error(self):
mock_logging = mock.Mock()
required_config = HBaseCrashStorage.get_required_config()
required_config.add_option('logger', default=mock_logging)
config_manager = ConfigurationManager(
[required_config],
app_name='testapp',
app_version='1.0',
app_description='app description',
values_source_list=[{
'logger': mock_logging,
'hbase_timeout': 100,
'hbase_host': commonconfig.hbaseHost.default,
'hbase_port': commonconfig.hbasePort.default,
}])
with config_manager.context() as config:
hbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
with mock.patch(hbaseclient_) as hclient:
klass = hclient.HBaseConnectionForCrashReports
def retry_raiser(*args, **kwargs):
raise SomeThriftError('try again')
klass.put_json_dump.side_effect = ValueError('crap!')
crashstorage = HBaseCrashStorage(config)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
# Note, we're not expect it to raise an error
self.assertRaises(ValueError, crashstorage.save_raw_crash,
json.loads(raw), raw, "abc123")
def test_hbase_crashstorage_success_after_retries(self):
cshbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
cchbaseclient_ = \
'socorro.external.hbase.connection_context.hbase_client'
with nested(mock.patch(cshbaseclient_),
mock.patch(cchbaseclient_)) as (cshclient, cchclient):
fake_hbase_client_connection = mock.MagicMock()
cshclient.HBaseConnectionForCrashReports.return_value = \
fake_hbase_client_connection
fake_put_json_method = mock.MagicMock()
cshclient.HBaseConnectionForCrashReports.put_json_dump = \
fake_put_json_method
cchclient.HBaseConnectionForCrashReports.return_value = \
fake_hbase_client_connection
fake_hbase_client_connection.hbaseThriftExceptions = \
(SomeThriftError,)
_attempts = [SomeThriftError, SomeThriftError]
def retry_raiser_iterator(*args, **kwargs):
try:
raise _attempts.pop(0)
except IndexError:
return None
fake_put_json_method.side_effect = retry_raiser_iterator
config = DotDict({
'logger':
mock.MagicMock(),
'hbase_timeout':
0,
'hbase_host':
'somehost',
'hbase_port':
9090,
'number_of_retries':
2,
'hbase_connection_pool_class':
HBaseConnectionContextPooled,
'transaction_executor_class':
TransactionExecutorWithLimitedBackoff,
'backoff_delays': [0, 0, 0],
'redactor_class':
Redactor,
'forbidden_keys':
Redactor.required_config.forbidden_keys.default,
'executor_identity':
lambda: 'dwight' # bogus thread id
})
crashstorage = HBaseCrashStorage(config)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
crashstorage.save_raw_crash(json.loads(raw), raw, "abc123")
eq_(fake_put_json_method.call_count, 3)
def test_hbase_crashstorage_success_after_retries(self):
mock_logging = mock.Mock()
required_config = HBaseCrashStorage.required_config
required_config.add_option('logger', default=mock_logging)
config_manager = ConfigurationManager(
[required_config],
app_name='testapp',
app_version='1.0',
app_description='app description',
values_source_list=[{
'logger': mock_logging,
'hbase_timeout': 100,
'hbase_host': commonconfig.hbaseHost.default,
'hbase_port': commonconfig.hbasePort.default,
'transaction_executor_class':
TransactionExecutorWithLimitedBackoff,
'backoff_delays': [0, 0, 0]
}]
)
with config_manager.context() as config:
hbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
with mock.patch(hbaseclient_) as hclient:
class SomeThriftError(Exception):
pass
klass = hclient.HBaseConnectionForCrashReports
klass.return_value = klass
klass.operational_exceptions = (SomeThriftError,)
_attempts = [SomeThriftError, SomeThriftError]
def retry_raiser_iterator(*args, **kwargs):
try:
raise _attempts.pop(0)
except IndexError:
return klass
klass.put_json_dump.side_effect = retry_raiser_iterator
crashstorage = HBaseCrashStorage(config)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
crashstorage.save_raw_crash(json.loads(raw), raw, "abc123")
self.assertEqual(klass.put_json_dump.call_count, 3)
def test_hbase_crashstorage_success_after_retries(self):
mock_logging = mock.Mock()
required_config = HBaseCrashStorage.required_config
required_config.add_option('logger', default=mock_logging)
config_manager = ConfigurationManager(
[required_config],
app_name='testapp',
app_version='1.0',
app_description='app description',
values_source_list=[{
'logger': mock_logging,
'hbase_timeout': 100,
'hbase_host': commonconfig.hbaseHost.default,
'hbase_port': commonconfig.hbasePort.default,
'transaction_executor_class':
TransactionExecutorWithLimitedBackoff,
'backoff_delays': [0, 0, 0]
}])
with config_manager.context() as config:
hbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
with mock.patch(hbaseclient_) as hclient:
class SomeThriftError(Exception):
pass
klass = hclient.HBaseConnectionForCrashReports
klass.return_value = klass
klass.operational_exceptions = (SomeThriftError, )
_attempts = [SomeThriftError, SomeThriftError]
def retry_raiser_iterator(*args, **kwargs):
try:
raise _attempts.pop(0)
except IndexError:
return klass
klass.put_json_dump.side_effect = retry_raiser_iterator
crashstorage = HBaseCrashStorage(config)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
crashstorage.save_raw_crash(json.loads(raw), raw, "abc123")
self.assertEqual(klass.put_json_dump.call_count, 3)
def test_hbase_crashstorage_success_after_retries(self):
cshbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
cchbaseclient_ = \
'socorro.external.hbase.connection_context.hbase_client'
with nested(mock.patch(cshbaseclient_),
mock.patch(cchbaseclient_)) as (cshclient, cchclient):
fake_hbase_client_connection = mock.MagicMock()
cshclient.HBaseConnectionForCrashReports.return_value = \
fake_hbase_client_connection
fake_put_json_method = mock.MagicMock()
cshclient.HBaseConnectionForCrashReports.put_json_dump = \
fake_put_json_method
cchclient.HBaseConnectionForCrashReports.return_value = \
fake_hbase_client_connection
fake_hbase_client_connection.hbaseThriftExceptions = \
(SomeThriftError,)
_attempts = [SomeThriftError, SomeThriftError]
def retry_raiser_iterator(*args, **kwargs):
try:
raise _attempts.pop(0)
except IndexError:
return None
fake_put_json_method.side_effect = retry_raiser_iterator
config = DotDict({
'logger': mock.MagicMock(),
'hbase_timeout': 0,
'hbase_host': 'somehost',
'hbase_port': 9090,
'number_of_retries': 2,
'hbase_connection_pool_class':
HBaseConnectionContextPooled,
'transaction_executor_class':
TransactionExecutorWithLimitedBackoff,
'backoff_delays': [0, 0, 0],
'redactor_class': Redactor,
'forbidden_keys':
Redactor.required_config.forbidden_keys.default,
'executor_identity': lambda: 'dwight' # bogus thread id
})
crashstorage = HBaseCrashStorage(config)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
crashstorage.save_raw_crash(json.loads(raw), raw, "abc123")
eq_(fake_put_json_method.call_count, 3)
def test_hbase_crashstorage_error_after_retries(self):
cshbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
cchbaseclient_ = \
'socorro.external.hbase.connection_context.hbase_client'
with nested(mock.patch(cshbaseclient_),
mock.patch(cchbaseclient_)) as (cshclient, cchclient):
fake_hbase_client_connection = mock.MagicMock()
cshclient.HBaseConnectionForCrashReports.return_value = \
fake_hbase_client_connection
fake_put_json_method = mock.MagicMock()
cshclient.HBaseConnectionForCrashReports.put_json_dump = \
fake_put_json_method
cchclient.HBaseConnectionForCrashReports.return_value = \
fake_hbase_client_connection
fake_hbase_client_connection.hbaseThriftExceptions = \
(SomeThriftError,)
fake_put_json_method.side_effect = SomeThriftError('try again')
config = DotDict({
'logger':
mock.MagicMock(),
'hbase_timeout':
0,
'hbase_host':
'somehost',
'hbase_port':
9090,
'number_of_retries':
2,
'hbase_connection_pool_class':
HBaseConnectionContextPooled,
'transaction_executor_class':
TransactionExecutorWithLimitedBackoff,
'backoff_delays': [0, 0, 0],
'redactor_class':
Redactor,
'forbidden_keys':
Redactor.required_config.forbidden_keys.default,
})
crashstorage = HBaseCrashStorage(config)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
self.assertRaises(SomeThriftError, crashstorage.save_raw_crash,
json.loads(raw), raw, {})
self.assertEqual(fake_put_json_method.call_count, 3)
def __init__(self, config, quit_check=None):
super(CrashStorageSystemForHBase, self).__init__(config)
# new_config is an adapter to allow the modern configman enabled
# file system crash storage classes to use the old style configuration.
new_config = DotDict()
new_config.logger = config.logger
new_config.hbase_connection_pool_class = HBaseConnectionContextPooled
new_config.number_of_retries = 2
new_config.hbase_host = config.hbaseHost
new_config.hbase_port = config.hbasePort
new_config.hbase_timeout = config.hbaseTimeout
new_config.forbidden_keys = ['email', 'url', 'user_id',
'exploitability']
new_config.transaction_executor_class = \
TransactionExecutorWithLimitedBackoff
new_config.backoff_delays = [10, 30, 60, 120, 300]
new_config.wait_log_interval = 5
self.crash_storage = HBaseCrashStorage(new_config, quit_check)
def test_hbase_crashstorage_basic_error(self):
mock_logging = mock.Mock()
required_config = HBaseCrashStorage.get_required_config()
required_config.add_option('logger', default=mock_logging)
config_manager = ConfigurationManager(
[required_config],
app_name='testapp',
app_version='1.0',
app_description='app description',
values_source_list=[{
'logger': mock_logging,
'hbase_timeout': 100,
'hbase_host': commonconfig.hbaseHost.default,
'hbase_port': commonconfig.hbasePort.default,
}],
argv_source=[]
)
with config_manager.context() as config:
config.executor_identity = lambda: 'dwight' # bogus thread id
hbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
with mock.patch(hbaseclient_) as hclient:
klass = hclient.HBaseConnectionForCrashReports
def retry_raiser(*args, **kwargs):
raise SomeThriftError('try again')
klass.put_json_dump.side_effect = ValueError('crap!')
crashstorage = HBaseCrashStorage(config)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
# Note, we're not expect it to raise an error
assert_raises(ValueError,
crashstorage.save_raw_crash,
json.loads(raw),
raw,
"abc123"
)
def test_basic_hbase_crashstorage(self):
mock_logging = mock.Mock()
required_config = HBaseCrashStorage.required_config
required_config.add_option('logger', default=mock_logging)
config_manager = ConfigurationManager(
[required_config],
app_name='testapp',
app_version='1.0',
app_description='app description',
values_source_list=[{
'logger':
mock_logging,
'hbase_timeout':
100,
'hbase_host':
commonconfig.hbaseHost.default,
'hbase_port':
commonconfig.hbasePort.default,
}])
with config_manager.context() as config:
crashstorage = HBaseCrashStorage(config)
self.assertEqual(list(crashstorage.new_crashes()), [])
crash_id = '86b58ff2-9708-487d-bfc4-9dac32121214'
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
fake_raw_dump_1 = 'peter is a swede'
fake_raw_dump_2 = 'lars is a norseman'
fake_raw_dump_3 = 'adrian is a frenchman'
fake_dumps = {
'upload_file_minidump': fake_raw_dump_1,
'lars': fake_raw_dump_2,
'adrian': fake_raw_dump_3
}
crashstorage.save_raw_crash(json.loads(raw), fake_dumps,
crash_id)
assert config.logger.info.called
assert config.logger.info.call_count > 1
msg_tmpl, msg_arg = config.logger.info.call_args_list[1][0]
# ie logging.info(<template>, <arg>)
msg = msg_tmpl % msg_arg
self.assertTrue('saved' in msg)
self.assertTrue(crash_id in msg)
raw_crash = crashstorage.get_raw_crash(crash_id)
assert isinstance(raw_crash, dict)
self.assertEqual(raw_crash['name'], 'Peter')
dump = crashstorage.get_raw_dump(crash_id)
assert isinstance(dump, basestring)
self.assertTrue('peter is a swede' in dump)
dumps = crashstorage.get_raw_dumps(crash_id)
assert isinstance(dumps, dict)
self.assertTrue('upload_file_minidump' in dumps)
self.assertTrue('lars' in dumps)
self.assertTrue('adrian' in dumps)
self.assertEqual(dumps['upload_file_minidump'],
fake_dumps['upload_file_minidump'])
self.assertEqual(dumps['lars'], fake_dumps['lars'])
self.assertEqual(dumps['adrian'], fake_dumps['adrian'])
# hasn't been processed yet
self.assertRaises(CrashIDNotFound, crashstorage.get_processed,
crash_id)
pro = ('{"name":"Peter",'
'"uuid":"86b58ff2-9708-487d-bfc4-9dac32121214", '
'"submitted_timestamp":"%d", '
'"completeddatetime": "%d"}' %
(time.time(), time.time()))
crashstorage.save_processed(json.loads(pro))
data = crashstorage.get_processed(crash_id)
self.assertEqual(data['name'], u'Peter')
hb_connection = crashstorage.hbaseConnectionPool.connection()
self.assertTrue(hb_connection.transport.isOpen())
crashstorage.close()
self.assertFalse(hb_connection.transport.isOpen())
def test_hbase_crashstorage_puts_and_gets(self):
mock_logging = mock.Mock()
required_config = HBaseCrashStorage.required_config
required_config.add_option('logger', default=mock_logging)
config_manager = ConfigurationManager(
[required_config],
app_name='testapp',
app_version='1.0',
app_description='app description',
values_source_list=[{
'logger': mock_logging,
'hbase_timeout': 100,
'hbase_host': commonconfig.hbaseHost.default,
'hbase_port': commonconfig.hbasePort.default,
'transaction_executor_class':
TransactionExecutorWithLimitedBackoff,
'backoff_delays': [0, 0, 0]
}])
with config_manager.context() as config:
hbaseclient_ = 'socorro.external.hbase.crashstorage.hbase_client'
with mock.patch(hbaseclient_) as hclient:
# test save_raw_crash
raw_crash = {
"name": "Peter",
"email": "*****@*****.**",
"url": "http://embarassing.xxx",
"submitted_timestamp": "2012-05-04T15:10:00",
"user_id": "000-00-0000",
}
fake_binary_dump = "this a bogus binary dump"
expected_raw_crash = raw_crash
expected_dump = fake_binary_dump
expected_dump_2 = fake_binary_dump + " number 2"
# saves us from loooong lines
klass = hclient.HBaseConnectionForCrashReports
crashstorage = HBaseCrashStorage(config)
crashstorage.save_raw_crash(raw_crash, fake_binary_dump,
"abc123")
self.assertEqual(klass.put_json_dump.call_count, 1)
a = klass.put_json_dump.call_args
self.assertEqual(len(a[0]), 4)
#self.assertEqual(a[0][1], "abc123")
self.assertEqual(a[0][2], expected_raw_crash)
self.assertEqual(a[0][3], expected_dump)
self.assertEqual(a[1], {'number_of_retries': 0})
# test save_processed
processed_crash = {
"name": "Peter",
"uuid": "abc123",
"email": "*****@*****.**",
"url": "http://embarassing.xxx",
"user_id": "000-00-0000",
}
expected_processed_crash = {
"name": "Peter",
"uuid": "abc123",
}
crashstorage = HBaseCrashStorage(config)
crashstorage.save_processed(processed_crash)
self.assertEqual(klass.put_processed_json.call_count, 1)
a = klass.put_processed_json.call_args
self.assertEqual(len(a[0]), 3)
self.assertEqual(a[0][1], "abc123")
self.assertEqual(a[0][2], expected_processed_crash)
self.assertEqual(a[1], {'number_of_retries': 0})
# test get_raw_crash
m = mock.Mock(return_value=raw_crash)
klass.get_json = m
r = crashstorage.get_raw_crash("abc123")
self.assertTrue(isinstance(r, DotDict))
a = klass.get_json.call_args
self.assertEqual(len(a[0]), 2)
self.assertEqual(a[0][1], "abc123")
self.assertEqual(klass.get_json.call_count, 1)
self.assertEqual(r, expected_raw_crash)
# test get_raw_dump
m = mock.Mock(return_value=fake_binary_dump)
klass.get_dump = m
r = crashstorage.get_raw_dump("abc123")
a = klass.get_dump.call_args
self.assertEqual(len(a[0]), 3)
self.assertEqual(a[0][1], "abc123")
self.assertEqual(klass.get_dump.call_count, 1)
self.assertEqual(r, expected_dump)
# test get_raw_dumps
m = mock.Mock(
return_value={'upload_file_minidump': fake_binary_dump})
klass.get_dumps = m
r = crashstorage.get_raw_dumps("abc123")
a = klass.get_dumps.call_args
self.assertEqual(len(a[0]), 2)
self.assertEqual(a[0][1], "abc123")
self.assertEqual(klass.get_dumps.call_count, 1)
self.assertEqual(r, {'upload_file_minidump': expected_dump})
# test get_raw_dumps 2
m = mock.Mock(
return_value={
'upload_file_minidump': fake_binary_dump,
'aux_1': expected_dump_2
})
klass.get_dumps = m
r = crashstorage.get_raw_dumps("abc123")
a = klass.get_dumps.call_args
self.assertEqual(len(a[0]), 2)
self.assertEqual(a[0][1], "abc123")
self.assertEqual(klass.get_dumps.call_count, 1)
self.assertEqual(
r, {
'upload_file_minidump': fake_binary_dump,
'aux_1': expected_dump_2
})
# test get_processed
m = mock.Mock(return_value=expected_processed_crash)
klass.get_processed_json = m
r = crashstorage.get_processed("abc123")
self.assertTrue(isinstance(r, DotDict))
a = klass.get_processed_json.call_args
self.assertEqual(len(a[0]), 2)
self.assertEqual(a[0][1], "abc123")
self.assertEqual(klass.get_processed_json.call_count, 1)
self.assertEqual(r, expected_processed_crash)
def test_basic_hbase_crashstorage(self):
mock_logging = mock.Mock()
required_config = HBaseCrashStorage.required_config
required_config.add_option('logger', default=mock_logging)
config_manager = ConfigurationManager(
[required_config],
app_name='testapp',
app_version='1.0',
app_description='app description',
values_source_list=[{
'logger':
mock_logging,
'hbase_timeout':
100,
'hbase_host':
commonconfig.hbaseHost.default,
'hbase_port':
commonconfig.hbasePort.default,
}])
with config_manager.context() as config:
crashstorage = HBaseCrashStorage(config)
self.assertEqual(list(crashstorage.new_crashes()), [])
# data doesn't contain an 'ooid' key
#raw = '{"name": "Peter"}'
#self.assertRaises(
#CrashIDNotFound,
#crashstorage.save_raw_crash,
#json.loads(raw),
#raw
#)
#raw = '{"name":"Peter","ooid":"abc123"}'
#self.assertRaises(
#ValueError, # missing the 'submitted_timestamp' key
#crashstorage.save_raw_crash,
#json.loads(raw),
#raw
#)
raw = ('{"name":"Peter", '
'"submitted_timestamp":"%d"}' % time.time())
crashstorage.save_raw_crash(json.loads(raw), raw, "abc123")
assert config.logger.info.called
assert config.logger.info.call_count > 1
msg_tmpl, msg_arg = config.logger.info.call_args_list[1][0]
# ie logging.info(<template>, <arg>)
msg = msg_tmpl % msg_arg
self.assertTrue('saved' in msg)
self.assertTrue('abc123' in msg)
meta = crashstorage.get_raw_crash('abc123')
assert isinstance(meta, dict)
self.assertEqual(meta['name'], 'Peter')
dump = crashstorage.get_raw_dump('abc123')
assert isinstance(dump, basestring)
self.assertTrue('"name":"Peter"' in dump)
# hasn't been processed yet
self.assertRaises(CrashIDNotFound,
crashstorage.get_processed_crash, 'abc123')
pro = ('{"name":"Peter","uuid":"abc123", '
'"submitted_timestamp":"%d", '
'"completeddatetime": "%d"}' %
(time.time(), time.time()))
crashstorage.save_processed(json.loads(pro))
data = crashstorage.get_processed_crash('abc123')
self.assertEqual(data['name'], u'Peter')
assert crashstorage.hbaseConnection.transport.isOpen()
crashstorage.close()
transport = crashstorage.hbaseConnection.transport
self.assertTrue(not transport.isOpen())
