def _deindex_chunk(self, chunk):
rdir = RdirClient(self.conf, pool_manager=self.conscience.pool_manager)
url = chunk['url']
volume_id = url.split('/', 3)[2]
chunk_id = url.split('/', 3)[3]
rdir.chunk_delete(volume_id, self.container_id,
self.content_id, chunk_id)
class ChunkOperator(object):
"""
Execute maintenance operations on chunks.
"""
def __init__(self, conf, logger=None):
self.conf = conf
self.logger = logger or get_logger(conf)
self.rdir_client = RdirClient(conf, logger=self.logger)
self.content_factory = ContentFactory(conf, logger=self.logger)
def rebuild(self, container_id, content_id, chunk_id_or_pos,
rawx_id=None, try_chunk_delete=False, allow_same_rawx=True):
"""
Try to find the chunk in the metadata of the specified object,
then rebuild it.
"""
try:
content = self.content_factory.get(container_id, content_id)
except ContentNotFound:
raise OrphanChunk('Content not found: possible orphan chunk')
chunk_size = 0
chunk_pos = None
if len(chunk_id_or_pos) < 32:
chunk_pos = chunk_id_or_pos
chunk_id = None
metapos = int(chunk_pos.split('.', 1)[0])
chunk_size = content.chunks.filter(metapos=metapos).all()[0].size
else:
if '/' in chunk_id_or_pos:
chunk_id = chunk_id_or_pos.rsplit('/', 1)[-1]
else:
chunk_id = chunk_id_or_pos
chunk = content.chunks.filter(id=chunk_id).one()
if chunk is None:
raise OrphanChunk(
'Chunk not found in content: possible orphan chunk')
elif rawx_id and chunk.host != rawx_id:
raise ValueError('Chunk does not belong to this rawx')
chunk_size = chunk.size
content.rebuild_chunk(
chunk_id, allow_same_rawx=allow_same_rawx,
chunk_pos=chunk_pos)
if try_chunk_delete:
try:
content.blob_client.chunk_delete(chunk.url)
self.logger.info("Chunk %s deleted", chunk.url)
except NotFound as exc:
self.logger.debug("Chunk %s: %s", chunk.url, exc)
# This call does not raise exception if chunk is not referenced
if chunk_id is not None:
self.rdir_client.chunk_delete(
chunk.host, container_id, content_id, chunk_id)
return chunk_size
def __init__(self,
conf,
logger,
volume,
try_chunk_delete=False,
beanstalkd_addr=None,
**kwargs):
super(BlobRebuilder, self).__init__(conf, logger, volume, **kwargs)
# rdir
self.rdir_client = RdirClient(conf, logger=self.logger)
self.rdir_fetch_limit = int_value(conf.get('rdir_fetch_limit'), 100)
# rawx
self.try_chunk_delete = try_chunk_delete
# beanstalk
if beanstalkd_addr:
self.beanstalkd_listener = BeanstalkdListener(
beanstalkd_addr,
conf.get('beanstalkd_tube', DEFAULT_REBUILDER_TUBE),
self.logger, **kwargs)
else:
self.beanstalkd_listener = None
# counters
self.bytes_processed = 0
self.total_bytes_processed = 0
self.total_expected_chunks = None
# distributed
self.distributed = False
def __init__(self, conf, **kwargs):
super(BlobIndexer, self).__init__(conf)
self.logger = get_logger(conf)
volume = conf.get('volume')
if not volume:
raise exc.ConfigurationException('No volume specified for indexer')
self.volume = volume
self.passes = 0
self.errors = 0
self.successes = 0
self.last_reported = 0
self.total_since_last_reported = 0
self.chunks_run_time = 0
self.interval = int_value(
conf.get('interval'), 300)
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
pm = get_pool_manager(pool_connections=10)
self.index_client = RdirClient(conf, logger=self.logger,
pool_manager=pm)
self.namespace, self.volume_id = check_volume(self.volume)
self.convert_chunks = true_value(conf.get('convert_chunks'))
if self.convert_chunks:
converter_conf = self.conf.copy()
converter_conf['no_backup'] = True
self.converter = BlobConverter(converter_conf, logger=self.logger,
pool_manager=pm)
else:
self.converter = None
def setUp(self):
super(TestIndexerCrawler, self).setUp()
self.namespace = self.conf['namespace']
self.gridconf = {"namespace": self.namespace}
self.rdir_client = RdirClient(self.gridconf)
def __init__(self,
conf,
logger,
volume,
input_file=None,
try_chunk_delete=False,
beanstalkd_addr=None):
self.conf = conf
self.logger = logger or get_logger(conf)
self.volume = volume
self.run_time = 0
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.bytes_running_time = 0
self.bytes_processed = 0
self.total_bytes_processed = 0
self.total_chunks_processed = 0
self.dry_run = true_value(conf.get('dry_run', False))
self.report_interval = int_value(conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(conf.get('chunks_per_second'),
30)
self.max_bytes_per_second = int_value(conf.get('bytes_per_second'),
10000000)
self.rdir_fetch_limit = int_value(conf.get('rdir_fetch_limit'), 100)
self.allow_same_rawx = true_value(conf.get('allow_same_rawx'))
self.input_file = input_file
self.rdir_client = RdirClient(conf, logger=self.logger)
self.content_factory = ContentFactory(conf)
self.try_chunk_delete = try_chunk_delete
self.beanstalkd_addr = beanstalkd_addr
self.beanstalkd_tube = conf.get('beanstalkd_tube', 'rebuild')
self.beanstalk = None
class Harasser(object):
def __init__(self, ns, max_containers=256, max_contents=256):
conf = {'namespace': ns}
self.cs = ConscienceClient(conf)
self.rdir = RdirClient(conf)
self.rawx_list = [x['addr'] for x in self.cs.all_services('rawx')]
self.sent = set()
self.max_containers = max_containers
self.max_contents = max_contents
def harass_put(self, loops=None):
if loops is None:
loops = random.randint(1000, 2000)
print "Pushing %d fake chunks" % loops
loop = loops
count_start_container = random.randrange(2**20)
count_start_content = random.randrange(2**20)
start = time.time()
nb_rawx = len(self.rawx_list)
while loop > 0:
args = {'mtime': int(start)}
# vol_id = random.choice(self.rawx_list)
# container_id = "%064X" % (random.randrange(self.max_containers))
# content_id = "%032X" % (random.randrange(self.max_contents))
vol_id = self.rawx_list[loop % nb_rawx]
container_id = "%064X" % (loop + count_start_container)
content_id = "%032X" % (loop + count_start_content)
chunk_id = "http://%s/%064X" \
% (vol_id, random.randrange(2**128))
self.rdir.chunk_push(
vol_id, container_id, content_id, chunk_id, **args)
self.sent.add((vol_id, container_id, content_id, chunk_id))
loop -= 1
end = time.time()
print "%d pushed in %.3fs, %d req/s" \
% (loops, end-start, loops/(end-start))
def harass_del(self, min_loops=0):
min_loops = min(min_loops, len(self.sent))
loops = random.randint(min_loops, len(self.sent))
print "Removing %d fake chunks" % loops
loop = loops
start = time.time()
while loop > 0:
args = self.sent.pop()
self.rdir.chunk_delete(*args)
loop -= 1
end = time.time()
print "%d removed in %.3fs, %d req/s" \
% (loops, end-start, loops/(end-start))
def __call__(self):
try:
while True:
self.harass_put()
self.harass_del()
except KeyboardInterrupt:
print "Cleaning..."
self.harass_del(len(self.sent))
class TestRdirClient(BaseTestCase):
def setUp(self):
super(TestRdirClient, self).setUp()
self.namespace = self.conf['namespace']
self.rdir_client = RdirClient({'namespace': self.namespace})
self.rdir_client._get_rdir_addr = Mock(return_value="0.1.2.3:4567")
def tearDown(self):
super(TestRdirClient, self).tearDown()
del self.rdir_client
def test_fetch_one_req_post(self):
self.rdir_client._direct_request = Mock(
side_effect=[
(
Mock(),
[
["container1|content1|chunk1", {'mtime': 10}],
["container2|content2|chunk2", {'mtime': 20}]
]
)
])
gen = self.rdir_client.chunk_fetch("volume", limit=2)
self.assertEqual(gen.next(),
("container1", "content1", "chunk1", {'mtime': 10}))
self.assertEqual(gen.next(),
("container2", "content2", "chunk2", {'mtime': 20}))
self.assertRaises(StopIteration, gen.next)
self.assertEqual(self.rdir_client._direct_request.call_count, 2)
def test_fetch_multi_req(self):
self.rdir_client._direct_request = Mock(
side_effect=[
(
Mock(),
[
["container1|content1|chunk1", {'mtime': 10}],
["container2|content2|chunk2", {'mtime': 20}]
]
),
(
Mock(),
[
["container3|content3|chunk3", {'mtime': 30}]
]
)
])
gen = self.rdir_client.chunk_fetch("volume", limit=2)
# print(gen.next())
# print(gen.next())
# print(gen.next())
self.assertEqual(gen.next(),
("container1", "content1", "chunk1", {'mtime': 10}))
self.assertEqual(gen.next(),
("container2", "content2", "chunk2", {'mtime': 20}))
self.assertEqual(gen.next(),
("container3", "content3", "chunk3", {'mtime': 30}))
self.assertRaises(StopIteration, gen.next)
self.assertEqual(self.rdir_client._direct_request.call_count, 3)
def __init__(self,
namespace,
concurrency=50,
error_file=None,
rebuild_file=None,
check_xattr=True,
limit_listings=0,
request_attempts=1,
logger=None,
verbose=False,
check_hash=False,
**_kwargs):
self.pool = GreenPool(concurrency)
self.error_file = error_file
self.check_xattr = bool(check_xattr)
self.check_hash = bool(check_hash)
self.logger = logger or get_logger(
{'namespace': namespace}, name='integrity', verbose=verbose)
# Optimisation for when we are only checking one object
# or one container.
# 0 -> do not limit
# 1 -> limit account listings (list of containers)
# 2 -> limit container listings (list of objects)
self.limit_listings = limit_listings
if self.error_file:
outfile = open(self.error_file, 'a')
self.error_writer = csv.writer(outfile, delimiter=' ')
self.rebuild_file = rebuild_file
if self.rebuild_file:
self.fd = open(self.rebuild_file, 'a')
self.rebuild_writer = csv.writer(self.fd, delimiter='|')
self.api = ObjectStorageApi(namespace,
logger=self.logger,
max_retries=request_attempts - 1,
request_attempts=request_attempts)
self.rdir_client = RdirClient({"namespace": namespace},
logger=self.logger)
self.accounts_checked = 0
self.containers_checked = 0
self.objects_checked = 0
self.chunks_checked = 0
self.account_not_found = 0
self.container_not_found = 0
self.object_not_found = 0
self.chunk_not_found = 0
self.account_exceptions = 0
self.container_exceptions = 0
self.object_exceptions = 0
self.chunk_exceptions = 0
self.list_cache = {}
self.running = {}
self.running_lock = Semaphore(1)
self.result_queue = Queue(concurrency)
self.run_time = 0
def setUp(self):
super(TestMeta2Indexing, self).setUp()
self.rdir_client = RdirClient(self.conf)
self.directory_client = DirectoryClient(self.conf)
self.container_client = ContainerClient(self.conf)
self.containers = [random_str(14) for _ in range(0, randint(1, 10))]
self.containers_svcs = {}
self.event_agent_name = 'event-agent-1'
def setUp(self):
super(TestRdirMeta2Client, self).setUp()
self.namespace = "dummy"
self.volid = "e29b4c56-8522-4118-82ea"
self.container_url = "OPENIO/testing/test1"
self.container_id = "random833999id"
self.mtime = 2874884.47
self.rdir_client = RdirClient({'namespace': self.namespace},
endpoint='127.0.0.0:6000')
def test_rebuild_chunk(self):
# push a new content
content = TestContent(self.conf, self.account,
self.container_name, "mycontent", "TWOCOPIES")
data = "azerty"
content.add_chunk(data, pos='0', rawx=0)
content.add_chunk(data, pos='0', rawx=1)
self._push_content(content)
# rebuild the first rawx
rebuilder = BlobRebuilderWorker(self.gridconf, None,
self.conf['rawx'][0]['addr'])
rebuilder.chunk_rebuild(content.container_id, content.content_id,
content.chunks[0].id)
# check meta2 information
_, res = self.container_client.content_show(acct=content.account,
ref=content.container_name,
content=content.content_id)
new_chunk_info = None
for c in res:
if (c['url'] != content.chunks[0].url and
c['url'] != content.chunks[1].url):
new_chunk_info = c
new_chunk_id = new_chunk_info['url'].split('/')[-1]
self.assertEqual(new_chunk_info['hash'], content.chunks[0].hash)
self.assertEqual(new_chunk_info['pos'], content.chunks[0].pos)
self.assertEqual(new_chunk_info['size'], content.chunks[0].size)
# check chunk information
meta, stream = self.blob_client.chunk_get(new_chunk_info['url'])
self.assertEqual(meta['content_size'], str(content.chunks[0].size))
self.assertEqual(meta['content_path'], content.content_name)
self.assertEqual(meta['content_cid'], content.container_id)
self.assertEqual(meta['content_id'], content.content_id)
self.assertEqual(meta['chunk_id'], new_chunk_id)
self.assertEqual(meta['chunk_pos'], content.chunks[0].pos)
self.assertEqual(meta['content_version'], content.version)
self.assertEqual(meta['chunk_hash'], content.chunks[0].hash)
self.assertEqual(stream.next(), content.chunks[0].data)
# check rtime flag in rdir
rdir_client = RdirClient(self.gridconf)
res = rdir_client.chunk_fetch(self.conf['rawx'][0]['addr'])
key = (content.container_id, content.content_id, content.chunks[0].id)
for i_container, i_content, i_chunk, i_value in res:
if (i_container, i_content, i_chunk) == key:
check_value = i_value
self.assertIsNotNone(check_value.get('rtime'))
def test_rebuild_chunk(self):
# push a new content
content = TestContent(self.conf, self.account, self.container_name,
"mycontent", "TWOCOPIES")
data = "azerty"
content.add_chunk(data, pos='0', rawx=0)
content.add_chunk(data, pos='0', rawx=1)
self._push_content(content)
# rebuild the first rawx
rebuilder = BlobRebuilderWorker(self.gridconf, None,
self.conf['rawx'][0]['addr'])
rebuilder.chunk_rebuild(content.container_id, content.content_id,
content.chunks[0].id)
# check meta2 information
_, res = self.container_client.content_show(acct=content.account,
ref=content.container_name,
content=content.content_id)
new_chunk_info = None
for c in res:
if (c['url'] != content.chunks[0].url
and c['url'] != content.chunks[1].url):
new_chunk_info = c
new_chunk_id = new_chunk_info['url'].split('/')[-1]
self.assertEqual(new_chunk_info['hash'], content.chunks[0].hash)
self.assertEqual(new_chunk_info['pos'], content.chunks[0].pos)
self.assertEqual(new_chunk_info['size'], content.chunks[0].size)
# check chunk information
meta, stream = self.blob_client.chunk_get(new_chunk_info['url'])
self.assertEqual(meta['content_size'], str(content.chunks[0].size))
self.assertEqual(meta['content_path'], content.content_name)
self.assertEqual(meta['content_cid'], content.container_id)
self.assertEqual(meta['content_id'], content.content_id)
self.assertEqual(meta['chunk_id'], new_chunk_id)
self.assertEqual(meta['chunk_pos'], content.chunks[0].pos)
self.assertEqual(meta['content_version'], content.version)
self.assertEqual(meta['chunk_hash'], content.chunks[0].hash)
self.assertEqual(stream.next(), content.chunks[0].data)
# check rtime flag in rdir
rdir_client = RdirClient(self.gridconf)
res = rdir_client.chunk_fetch(self.conf['rawx'][0]['addr'])
key = (content.container_id, content.content_id, content.chunks[0].id)
for i_container, i_content, i_chunk, i_value in res:
if (i_container, i_content, i_chunk) == key:
check_value = i_value
self.assertIsNotNone(check_value.get('rtime'))
class TestRdirClient(BaseTestCase):
def setUp(self):
super(TestRdirClient, self).setUp()
self.namespace = self.conf["namespace"]
self.rdir_client = RdirClient({"namespace": self.namespace})
self.rdir_client._get_rdir_addr = Mock(return_value="0.1.2.3:4567")
self.container_id_1 = random_id(64)
self.container_id_2 = random_id(64)
self.container_id_3 = random_id(64)
self.content_id_1 = random_id(32)
self.content_id_2 = random_id(32)
self.content_id_3 = random_id(32)
self.chunk_id_1 = random_id(64)
self.chunk_id_2 = random_id(64)
self.chunk_id_3 = random_id(64)
def tearDown(self):
super(TestRdirClient, self).tearDown()
del self.rdir_client
def test_fetch_one_req_post(self):
self.rdir_client._direct_request = Mock(
side_effect=[
(
Mock(),
[
["%s|%s|%s" % (self.container_id_1, self.content_id_1, self.chunk_id_1), {"mtime": 10}],
["%s|%s|%s" % (self.container_id_2, self.content_id_2, self.chunk_id_2), {"mtime": 20}],
],
)
]
)
gen = self.rdir_client.chunk_fetch("volume", limit=2)
self.assertEqual(gen.next(), (self.container_id_1, self.content_id_1, self.chunk_id_1, {"mtime": 10}))
self.assertEqual(gen.next(), (self.container_id_2, self.content_id_2, self.chunk_id_2, {"mtime": 20}))
self.assertRaises(StopIteration, gen.next)
self.assertEqual(self.rdir_client._direct_request.call_count, 2)
def test_fetch_multi_req(self):
self.rdir_client._direct_request = Mock(
side_effect=[
(
Mock(),
[
["%s|%s|%s" % (self.container_id_1, self.content_id_1, self.chunk_id_1), {"mtime": 10}],
["%s|%s|%s" % (self.container_id_2, self.content_id_2, self.chunk_id_2), {"mtime": 20}],
],
),
(Mock(), [["%s|%s|%s" % (self.container_id_3, self.content_id_3, self.chunk_id_3), {"mtime": 30}]]),
]
)
gen = self.rdir_client.chunk_fetch("volume", limit=2)
self.assertEqual(gen.next(), (self.container_id_1, self.content_id_1, self.chunk_id_1, {"mtime": 10}))
self.assertEqual(gen.next(), (self.container_id_2, self.content_id_2, self.chunk_id_2, {"mtime": 20}))
self.assertEqual(gen.next(), (self.container_id_3, self.content_id_3, self.chunk_id_3, {"mtime": 30}))
self.assertRaises(StopIteration, gen.next)
self.assertEqual(self.rdir_client._direct_request.call_count, 3)
def setUp(self):
super(TestIndexerCrawler, self).setUp()
self.namespace = self.conf['namespace']
self.rawx_conf = self.conf['services']['rawx'][0]
self.conf = {"namespace": self.namespace,
"volume": self.rawx_conf['path']}
self.rdir_client = RdirClient(self.conf)
def setUp(self):
super(TestRdirClient, self).setUp()
self.rawx_conf = random.choice(self.conf['services']['rawx'])
self.rawx_id = self.rawx_conf.get('service_id', self.rawx_conf['addr'])
self.rdir = RdirClient(self.conf)
self.rdir.admin_clear(self.rawx_id, clear_all=True)
self._push_chunks()
self.rdir._direct_request = Mock(side_effect=self.rdir._direct_request)
def __init__(self, conf, logger, volume, try_chunk_delete=False, **kwargs):
super(BlobRebuilderWorker, self).__init__(conf, logger, **kwargs)
self.volume = volume
self.bytes_processed = 0
self.total_bytes_processed = 0
self.dry_run = true_value(conf.get('dry_run', False))
self.allow_same_rawx = true_value(conf.get('allow_same_rawx'))
self.rdir_client = RdirClient(conf, logger=self.logger)
self.content_factory = ContentFactory(conf, logger=self.logger)
self.try_chunk_delete = try_chunk_delete
def __init__(self, conf, input_file=None, service_id=None, **kwargs):
super(Meta2Rebuilder, self).__init__(conf, **kwargs)
# input
self.input_file = input_file
self.meta2_id = service_id
# rawx/rdir
self.rdir_client = RdirClient(self.conf, logger=self.logger)
self.rdir_fetch_limit = int_value(self.conf.get('rdir_fetch_limit'),
self.DEFAULT_RDIR_FETCH_LIMIT)
class TestRdirClient(BaseTestCase):
def setUp(self):
super(TestRdirClient, self).setUp()
self.namespace = self.conf['namespace']
self.rdir_client = RdirClient({'namespace': self.namespace})
self.rdir_client._get_rdir_addr = Mock(return_value="0.1.2.3:4567")
def tearDown(self):
super(TestRdirClient, self).tearDown()
del self.rdir_client
def test_fetch_one_req_post(self):
self.rdir_client._direct_request = Mock(
side_effect=[(Mock(),
[["container1|content1|chunk1", {
'mtime': 10
}], ["container2|content2|chunk2", {
'mtime': 20
}]])])
gen = self.rdir_client.chunk_fetch("volume", limit=2)
self.assertEqual(gen.next(), ("container1", "content1", "chunk1", {
'mtime': 10
}))
self.assertEqual(gen.next(), ("container2", "content2", "chunk2", {
'mtime': 20
}))
self.assertRaises(StopIteration, gen.next)
self.assertEqual(self.rdir_client._direct_request.call_count, 2)
def test_fetch_multi_req(self):
self.rdir_client._direct_request = Mock(side_effect=[(
Mock(), [["container1|content1|chunk1", {
'mtime': 10
}], ["container2|content2|chunk2", {
'mtime': 20
}]]), (Mock(), [["container3|content3|chunk3", {
'mtime': 30
}]])])
gen = self.rdir_client.chunk_fetch("volume", limit=2)
# print(gen.next())
# print(gen.next())
# print(gen.next())
self.assertEqual(gen.next(), ("container1", "content1", "chunk1", {
'mtime': 10
}))
self.assertEqual(gen.next(), ("container2", "content2", "chunk2", {
'mtime': 20
}))
self.assertEqual(gen.next(), ("container3", "content3", "chunk3", {
'mtime': 30
}))
self.assertRaises(StopIteration, gen.next)
self.assertEqual(self.rdir_client._direct_request.call_count, 3)
def tier_content(self, config, stats, control):
def _set(lock_, field, value):
lock_.acquire()
field.value = value
lock_.release()
def _add(lock_, field, value):
lock_.acquire()
field.value += value
lock_.release()
lock = control.get('lock')
try:
src = config.get('src')
del config['src']
self.client.lock_score(dict(type="rawx", addr=src))
api = ObjectStorageApi(config["namespace"])
rdir_client = RdirClient({'namespace': config["namespace"]})
self.log.info("Starting tierer on %s with policy %s" % (src, config["policy"]))
policies = dict()
for part in config["policy"].split(','):
policies[part.split(':')[0]] = part.split(':')[1]
self.log.info("Parsed policy: " + part.split(':')[0] + " " + part.split(':')[1])
for marker in config["markers"]:
req = dict(
start_after=marker,
limit=1000,
)
_, resp_body = rdir_client._rdir_request(src, 'POST', 'fetch', json=req)
for (key, value) in resp_body:
_, _, chunk = key.split('|')
res = requests.head("http://" + src + "/" + chunk)
policy = res.headers.get("x-oio-chunk-meta-content-storage-policy", "")
if policy not in policies.keys():
_add(lock, stats.get("skip"), 1)
continue
path = res.headers.get("x-oio-chunk-meta-full-path", "///")
path_parts = path.split('/')
if len(path_parts) < 3:
_add(lock, stats.get("skip"), 1)
continue
try:
api.object_change_policy(unquote(path_parts[0]), unquote(path_parts[1]), unquote(path_parts[2]), policies[policy])
_add(lock, stats.get("success"), 1)
except Exception as e:
self.log.info("Operation failed %s: %s (%s)" % (path, format_exc(e), policies[policy]))
_add(lock, stats.get("fail"), 1)
except Exception as e:
self.log.error("Tierer failed with %s" % format_exc(e))
_set(lock, control.get('status'), 2)
_set(lock, control.get('end'), int(time.time()))
def __init__(self, ns, max_containers=256, max_contents=256):
conf = {'namespace': ns}
self.cs = ConscienceClient(conf)
self.rdir = RdirClient(conf)
self.rawx_list = [x['addr'] for x in self.cs.all_services('rawx')]
self.sent = set()
self.max_containers = max_containers
self.max_contents = max_contents
self.pushed_count = 0
self.pushed_time = 0
self.removed_count = 0
self.removed_time = 0
def setUp(self):
super(TestRdirClient, self).setUp()
self.namespace = self.conf['namespace']
self.rdir_client = RdirClient({'namespace': self.namespace})
self.rdir_client._get_rdir_addr = Mock(return_value="0.1.2.3:4567")
self.container_id_1 = random_id(64)
self.container_id_2 = random_id(64)
self.container_id_3 = random_id(64)
self.content_id_1 = random_id(32)
self.content_id_2 = random_id(32)
self.content_id_3 = random_id(32)
self.chunk_id_1 = random_id(64)
self.chunk_id_2 = random_id(64)
self.chunk_id_3 = random_id(64)
def __init__(self, volume_path, conf, pool_manager=None):
"""
Initializes an Indexing worker for indexing meta2 databases.
Possible values of conf relating to this worker are:
- interval: (int) in sec time between two full scans. Default: half an
hour.
- report_interval: (int) in sec, time between two reports: Default: 300
- scanned_per_second: (int) maximum number of indexed databases /s.
- try_removing_faulty_indexes : In the event where we encounter a
database that's not supposed to be handled by this volume, attempt
to remove it from this volume rdir index if it exists
WARNING: The decision is based off of a proxy response, that could
be affected by cache inconsistencies for example, use at your own
risk. Default: False
:param volume_path: The volume path to be indexed
:param conf: The configuration to be passed to the needed services
:param pool_manager: A connection pool manager. If none is given, a
new one with a default size of 10 will be created.
"""
self.logger = get_logger(conf)
self._stop = False
self.volume = volume_path
self.success_nb = 0
self.failed_nb = 0
self.full_scan_nb = 0
self.last_report_time = 0
self.last_scan_time = 0
self.last_index_time = 0
self.start_time = 0
self.indexed_since_last_report = 0
self.scans_interval = int_value(
conf.get('interval'), 1800)
self.report_interval = int_value(
conf.get('report_interval'), 300)
self.max_indexed_per_second = int_value(
conf.get('scanned_per_second'), 3000)
self.namespace, self.volume_id = check_volume_for_service_type(
self.volume, "meta2")
self.attempt_bad_index_removal = boolean_value(
conf.get('try_removing_faulty_indexes', False)
)
if not pool_manager:
pool_manager = get_pool_manager(pool_connections=10)
self.index_client = RdirClient(conf, logger=self.logger,
pool_manager=pool_manager)
self.dir_client = DirectoryClient(conf, logger=self.logger,
pool_manager=pool_manager)
def init(self):
eventlet.monkey_patch(os=False)
self.session = requests.Session()
self.cs = ConscienceClient(self.conf)
self.rdir = RdirClient(self.conf)
self._acct_addr = None
self.acct_update = 0
self.graceful_timeout = 1
self.acct_refresh_interval = int_value(
self.conf.get('acct_refresh_interval'), 60)
self.concurrency = int_value(self.conf.get('concurrency'), 1000)
self.acct_update = true_value(self.conf.get('acct_update', True))
self.rdir_update = true_value(self.conf.get('rdir_update', True))
super(EventWorker, self).init()
def test_rdir_repartition(self):
client = RdirClient({'namespace': self.ns})
all_rawx = client.assign_all_rawx()
by_rdir = dict()
total = 0
for rawx in all_rawx:
count = by_rdir.get(rawx['rdir']['addr'], 0)
total += 1
by_rdir[rawx['rdir']['addr']] = count + 1
avg = total / float(len(by_rdir))
print "Ideal number of bases per rdir: ", avg
print "Current repartition: ", by_rdir
for count in by_rdir.itervalues():
self.assertLessEqual(count, avg + 1)
def __init__(self, conf, input_file=None, service_id=None, **kwargs):
super(BlobRebuilder, self).__init__(conf, **kwargs)
# counters
self.bytes_processed = 0
self.total_bytes_processed = 0
# input
self.input_file = input_file
self.rawx_id = service_id
# rawx/rdir
self.rdir_client = RdirClient(self.conf, logger=self.logger)
self.rdir_fetch_limit = int_value(self.conf.get('rdir_fetch_limit'),
self.DEFAULT_RDIR_FETCH_LIMIT)
def setUp(self):
super(TestRdirClient, self).setUp()
self.namespace = "dummy"
self.rdir_client = RdirClient({'namespace': self.namespace},
endpoint='127.0.0.0:6000')
self.rdir_client._get_rdir_addr = Mock(return_value="0.1.2.3:4567")
self.container_id_1 = random_id(64)
self.container_id_2 = random_id(64)
self.container_id_3 = random_id(64)
self.content_id_1 = random_id(32)
self.content_id_2 = random_id(32)
self.content_id_3 = random_id(32)
self.chunk_id_1 = random_id(64)
self.chunk_id_2 = random_id(64)
self.chunk_id_3 = random_id(64)
def __init__(self, conf, logger, volume):
self.conf = conf
self.logger = logger
self.volume = volume
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.total_chunks_processed = 0
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
self.index_client = RdirClient(conf)
self.namespace, self.volume_id = check_volume(self.volume)
def init(self):
self.concurrency = int_value(self.conf.get('concurrency'), 10)
self.tube = self.conf.get("tube", DEFAULT_TUBE)
acct_refresh_interval = int_value(
self.conf.get('acct_refresh_interval'), 3600)
self.app_env['account_client'] = AccountClient(
self.conf,
logger=self.logger,
refresh_delay=acct_refresh_interval,
pool_connections=3, # 1 account, 1 proxy, 1 extra
)
self.app_env['rdir_client'] = RdirClient(
self.conf,
logger=self.logger,
pool_maxsize=self.concurrency, # 1 cnx per greenthread per host
)
if 'handlers_conf' not in self.conf:
raise ValueError("'handlers_conf' path not defined in conf")
self.handlers = loadhandlers(self.conf.get('handlers_conf'),
global_conf=self.conf,
app=self)
for opt in ('acct_update', 'rdir_update', 'retries_per_second',
'batch_size'):
if opt in self.conf:
self.logger.warn('Deprecated option: %s', opt)
super(EventWorker, self).init()
def __init__(self, conf, logger, volume):
self.conf = conf
self.logger = logger or get_logger(conf)
self.volume = volume
self.run_time = 0
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.bytes_running_time = 0
self.bytes_processed = 0
self.total_bytes_processed = 0
self.total_chunks_processed = 0
self.dry_run = true_value(
conf.get('dry_run', False))
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
self.max_bytes_per_second = int_value(
conf.get('bytes_per_second'), 10000000)
self.rdir_fetch_limit = int_value(
conf.get('rdir_fetch_limit'), 100)
self.rdir_client = RdirClient(conf)
self.content_factory = ContentFactory(conf)
def setUp(self):
super(TestIndexerCrawler, self).setUp()
self.namespace = self.conf['namespace']
self.gridconf = {"namespace": self.namespace}
self.rdir_client = RdirClient(self.gridconf)
def __init__(self,
conf,
logger,
volume,
try_chunk_delete=False,
beanstalkd_addr=None,
**kwargs):
super(BlobRebuilder, self).__init__(conf, logger, **kwargs)
self.volume = volume
self.rdir_client = RdirClient(conf, logger=self.logger)
self.try_chunk_delete = try_chunk_delete
self.beanstalkd_addr = beanstalkd_addr
self.beanstalkd_tube = conf.get('beanstalkd_tube',
DEFAULT_REBUILDER_TUBE)
self.beanstalk = None
self.rdir_fetch_limit = int_value(conf.get('rdir_fetch_limit'), 100)
def __init__(self, conf, name, context, **kwargs):
self.conf = conf
self.name = name
verbose = kwargs.pop('verbose', False)
self.logger = get_logger(self.conf, verbose=verbose)
self.init_zmq(context)
self.cs = ConscienceClient(self.conf)
self.rdir = RdirClient(self.conf)
self._acct_addr = None
self.acct_update = 0
self.acct_refresh_interval = int_value(
conf.get('acct_refresh_interval'), 60)
self.acct_update = true_value(conf.get('acct_update', True))
self.rdir_update = true_value(conf.get('rdir_update', True))
self.session = requests.Session()
self.failed = False
def __init__(self, ns, max_containers=256, max_contents=256):
conf = {'namespace': ns}
self.cs = ConscienceClient(conf)
self.rdir = RdirClient(conf)
self.rawx_list = [x['addr'] for x in self.cs.all_services('rawx')]
self.sent = set()
self.max_containers = max_containers
self.max_contents = max_contents
class Meta2DecommissionJob(XcuteRdirJob):
JOB_TYPE = 'meta2-decommission'
TASK_CLASS = Meta2DecommissionTask
@classmethod
def sanitize_params(cls, job_params):
sanitized_job_params, _ = super(Meta2DecommissionJob,
cls).sanitize_params(job_params)
src = job_params.get('service_id')
if not src:
raise ValueError('Missing service ID')
sanitized_job_params['service_id'] = src
sanitized_job_params['dst'] = job_params.get('dst')
return sanitized_job_params, 'meta2/%s' % src
def __init__(self, conf, logger=None):
super(Meta2DecommissionJob, self).__init__(conf, logger=logger)
self.rdir_client = RdirClient(conf, logger=logger)
def get_tasks(self, job_params, marker=None):
containers = self._containers_from_rdir(job_params, marker)
for marker, container_id in containers:
yield marker, dict(container_id=container_id)
def get_total_tasks(self, job_params, marker=None):
containers = self._containers_from_rdir(job_params, marker)
i = 0
for i, (marker, _) in enumerate(containers, 1):
if i % 1000 == 0:
yield marker, 1000
remaining = i % 1000
if remaining == 0:
return
yield marker, remaining
def _containers_from_rdir(self, job_params, marker):
service_id = job_params['service_id']
rdir_fetch_limit = job_params['rdir_fetch_limit']
rdir_timeout = job_params['rdir_timeout']
containers = self.rdir_client.meta2_index_fetch_all(
service_id,
marker=marker,
timeout=rdir_timeout,
limit=rdir_fetch_limit)
for container_info in containers:
container_url = container_info['container_url']
container_id = container_info['container_id']
yield container_url, container_id
def __init__(self, conf, **kwargs):
super(BlobIndexer, self).__init__(conf)
self.logger = get_logger(conf)
volume = conf.get('volume')
if not volume:
raise exc.ConfigurationException('No volume specified for indexer')
self.volume = volume
self.passes = 0
self.errors = 0
self.successes = 0
self.last_reported = 0
self.chunks_run_time = 0
self.interval = int_value(conf.get('interval'), 300)
self.report_interval = int_value(conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(conf.get('chunks_per_second'),
30)
self.index_client = RdirClient(conf, logger=self.logger)
self.namespace, self.volume_id = check_volume(self.volume)
def setUp(self):
super(TestIndexerCrawler, self).setUp()
self.namespace = self.conf['namespace']
self.rawx_conf = self.conf['services']['rawx'][0]
self.conf = {"namespace": self.namespace,
"volume": self.rawx_conf['path']}
self.rdir_client = RdirClient(self.conf)
def test_link_rdir_to_zero_scored_rawx(self):
client = RdirClient({'namespace': self.ns})
disp = RdirDispatcher({'namespace': self.ns})
# Register a service, with score locked to zero
new_rawx = self._srv('rawx', {'tag.loc': 'whatever'})
new_rawx['score'] = 0
self._register_srv(new_rawx)
self._reload_proxy()
all_rawx = disp.assign_all_rawx()
all_rawx_keys = [x['addr'] for x in all_rawx]
self.assertIn(new_rawx['addr'], all_rawx_keys)
rdir_addr = client._get_rdir_addr(new_rawx['addr'])
self.assertIsNotNone(rdir_addr)
try:
self.api.unlink('_RDIR', new_rawx['addr'], 'rdir')
self.api.delete('_RDIR', new_rawx['addr'])
# self._flush_cs('rawx')
except Exception:
pass
def setUp(self):
super(TestBlobIndexer, self).setUp()
self.rdir_client = RdirClient(self.conf)
self.blob_client = BlobClient(self.conf)
_, self.rawx_path, rawx_addr, _ = \
self.get_service_url('rawx')
services = self.conscience.all_services('rawx')
self.rawx_id = None
for rawx in services:
if rawx_addr == rawx['addr']:
self.rawx_id = rawx['tags'].get('tag.service_id', None)
if self.rawx_id is None:
self.rawx_id = rawx_addr
conf = self.conf.copy()
conf['volume'] = self.rawx_path
self.blob_indexer = BlobIndexer(conf)
# clear rawx/rdir
chunk_files = paths_gen(self.rawx_path)
for chunk_file in chunk_files:
os.remove(chunk_file)
self.rdir_client.admin_clear(self.rawx_id, clear_all=True)
def __init__(self, conf, logger, volume):
self.conf = conf
self.logger = logger
self.volume = volume
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.total_chunks_processed = 0
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
self.index_client = RdirClient(conf)
def setUp(self):
super(TestRdirClient, self).setUp()
self.namespace = self.conf["namespace"]
self.rdir_client = RdirClient({"namespace": self.namespace})
self.rdir_client._get_rdir_addr = Mock(return_value="0.1.2.3:4567")
self.container_id_1 = random_id(64)
self.container_id_2 = random_id(64)
self.container_id_3 = random_id(64)
self.content_id_1 = random_id(32)
self.content_id_2 = random_id(32)
self.content_id_3 = random_id(32)
self.chunk_id_1 = random_id(64)
self.chunk_id_2 = random_id(64)
self.chunk_id_3 = random_id(64)
def init(self):
eventlet.monkey_patch(os=False)
self.session = requests.Session()
self.cs = ConscienceClient(self.conf)
self.rdir = RdirClient(self.conf)
self._acct_addr = None
self.acct_update = 0
self.graceful_timeout = 1
self.acct_refresh_interval = int_value(
self.conf.get('acct_refresh_interval'), 60
)
self.concurrency = int_value(self.conf.get('concurrency'), 1000)
self.acct_update = true_value(self.conf.get('acct_update', True))
self.rdir_update = true_value(self.conf.get('rdir_update', True))
super(EventWorker, self).init()
def __init__(self, conf, name, context, **kwargs):
self.conf = conf
self.name = name
verbose = kwargs.pop("verbose", False)
self.logger = get_logger(self.conf, verbose=verbose)
self.init_zmq(context)
self.cs = ConscienceClient(self.conf)
self.rdir = RdirClient(self.conf)
self._acct_addr = None
self.acct_update = 0
self.acct_refresh_interval = int_value(conf.get("acct_refresh_interval"), 60)
self.acct_update = true_value(conf.get("acct_update", True))
self.rdir_update = true_value(conf.get("rdir_update", True))
self.session = requests.Session()
self.failed = False
def __init__(self, conf, **kwargs):
super(BlobIndexer, self).__init__(conf)
self.logger = get_logger(conf)
volume = conf.get('volume')
if not volume:
raise exc.ConfigurationException('No volume specified for indexer')
self.volume = volume
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.total_chunks_processed = 0
self.interval = int_value(
conf.get('interval'), 300)
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
self.index_client = RdirClient(conf)
self.namespace, self.volume_id = check_volume(self.volume)
class BlobIndexer(Daemon):
def __init__(self, conf, **kwargs):
super(BlobIndexer, self).__init__(conf)
self.logger = get_logger(conf)
volume = conf.get('volume')
if not volume:
raise exc.ConfigurationException('No volume specified for indexer')
self.volume = volume
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.total_chunks_processed = 0
self.interval = int_value(
conf.get('interval'), 300)
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
self.index_client = RdirClient(conf)
self.namespace, self.volume_id = check_volume(self.volume)
def index_pass(self):
start_time = report_time = time.time()
total_errors = 0
paths = paths_gen(self.volume)
for path in paths:
self.safe_update_index(path)
self.chunks_run_time = ratelimit(
self.chunks_run_time,
self.max_chunks_per_second
)
self.total_chunks_processed += 1
now = time.time()
if now - self.last_reported >= self.report_interval:
self.logger.info(
'%(start_time)s '
'%(passes)d '
'%(errors)d '
'%(c_rate).2f '
'%(total).2f ' % {
'start_time': time.ctime(report_time),
'passes': self.passes,
'errors': self.errors,
'c_rate': self.passes / (now - report_time),
'total': (now - start_time)
}
)
report_time = now
total_errors += self.errors
self.passes = 0
self.errors = 0
self.last_reported = now
elapsed = (time.time() - start_time) or 0.000001
self.logger.info(
'%(elapsed).02f '
'%(errors)d '
'%(chunk_rate).2f ' % {
'elapsed': elapsed,
'errors': total_errors + self.errors,
'chunk_rate': self.total_chunks_processed / elapsed
}
)
if elapsed < self.interval:
time.sleep(self.interval - elapsed)
def safe_update_index(self, path):
try:
self.logger.debug('Updating index: %s' % path)
self.update_index(path)
except Exception:
self.errors += 1
self.logger.exception('ERROR while updating index for chunk %s',
path)
self.passes += 1
def update_index(self, path):
with open(path) as f:
try:
meta = read_chunk_metadata(f)
except exc.MissingAttribute as e:
raise exc.FaultyChunk(
'Missing extended attribute %s' % e)
data = {
'content_version': meta['content_version'],
'content_nbchunks': meta['content_chunksnb'],
'content_path': meta['content_path'],
'content_size': meta['content_size'],
'chunk_hash': meta['chunk_hash'],
'chunk_position': meta['chunk_pos'],
'chunk_size': meta['chunk_size'],
'mtime': int(time.time())
}
self.index_client.chunk_push(self.volume_id,
meta['content_cid'],
meta['content_id'],
meta['chunk_id'],
**data)
def run(self, *args, **kwargs):
time.sleep(random() * self.interval)
while True:
try:
self.index_pass()
except Exception as e:
self.logger.exception('ERROR during indexing: %s' % e)
class TestIndexerCrawler(BaseTestCase):
def setUp(self):
super(TestIndexerCrawler, self).setUp()
self.namespace = self.conf['namespace']
self.gridconf = {"namespace": self.namespace}
self.rdir_client = RdirClient(self.gridconf)
def tearDown(self):
super(TestIndexerCrawler, self).tearDown()
def _create_chunk(self, rawx_path):
container_id = generate_id(64)
content_id = generate_id(64)
chunk_id = generate_id(64)
chunk_dir = "%s/%s" % (rawx_path, chunk_id[0:2])
if not os.path.isdir(chunk_dir):
os.makedirs(chunk_dir)
chunk_path = "%s/%s" % (chunk_dir, chunk_id)
with open(chunk_path, "w") as f:
f.write("toto")
xattr.setxattr(chunk_path, 'user.grid.chunk.hash', 32 * '0')
xattr.setxattr(chunk_path, 'user.grid.chunk.id', chunk_id)
xattr.setxattr(chunk_path, 'user.grid.chunk.position', '0')
xattr.setxattr(chunk_path, 'user.grid.chunk.size', '4')
xattr.setxattr(chunk_path, 'user.grid.content.container', container_id)
xattr.setxattr(chunk_path, 'user.grid.content.id', content_id)
xattr.setxattr(chunk_path, 'user.grid.content.nbchunk', '1')
xattr.setxattr(chunk_path, 'user.grid.content.path', 'toto')
xattr.setxattr(chunk_path, 'user.grid.content.size', '4')
xattr.setxattr(chunk_path, 'user.grid.content.mime_type',
'application/octet-stream')
xattr.setxattr(chunk_path, 'user.grid.content.storage_policy',
'TESTPOLICY')
xattr.setxattr(chunk_path, 'user.grid.content.chunk_method',
'bytes')
xattr.setxattr(chunk_path, 'user.grid.content.version', '0')
return chunk_path, container_id, content_id, chunk_id
def _rdir_get(self, rawx_addr, container_id, content_id, chunk_id):
data = self.rdir_client.chunk_fetch(rawx_addr)
key = (container_id, content_id, chunk_id)
for i_container, i_content, i_chunk, i_value in data:
if (i_container, i_content, i_chunk) == key:
return i_value
return None
def test_index_chunk(self):
rawx_conf = self.conf['rawx'][0]
# create a fake chunk
chunk_path, container_id, content_id, chunk_id = self._create_chunk(
rawx_conf['path'])
# index the chunk
indexer = BlobIndexerWorker(self.gridconf, None, rawx_conf['path'])
with mock.patch('oio.blob.indexer.time.time',
mock.MagicMock(return_value=1234)):
indexer.update_index(chunk_path)
# check rdir
check_value = self._rdir_get(rawx_conf['addr'], container_id,
content_id, chunk_id)
self.assertIsNotNone(check_value)
self.assertEqual(check_value['content_nbchunks'], 1)
self.assertEqual(check_value['chunk_hash'], 32 * '0')
self.assertEqual(check_value['content_size'], 4)
self.assertEqual(check_value['content_path'], 'toto')
self.assertEqual(check_value['chunk_position'], '0')
self.assertEqual(check_value['chunk_size'], 4)
self.assertEqual(check_value['mtime'], 1234)
self.assertEqual(check_value['content_version'], 0)
# index a chunk already indexed
with mock.patch('oio.blob.indexer.time.time',
mock.MagicMock(return_value=4567)):
indexer.update_index(chunk_path)
# check rdir
check_value = self._rdir_get(rawx_conf['addr'], container_id,
content_id, chunk_id)
self.assertIsNotNone(check_value)
self.assertEqual(check_value['mtime'], 4567)
def test_index_chunk_missing_xattr(self):
rawx_conf = self.conf['rawx'][0]
# create a fake chunk
chunk_path, container_id, content_id, chunk_id = self._create_chunk(
rawx_conf['path'])
# remove mandatory xattr
xattr.removexattr(chunk_path, 'user.grid.chunk.hash')
# try to index the chunk
indexer = BlobIndexerWorker(self.gridconf, None, rawx_conf['path'])
self.assertRaises(FaultyChunk, indexer.update_index, chunk_path)
class BlobRebuilderWorker(object):
def __init__(self, conf, logger, volume):
self.conf = conf
self.logger = logger or get_logger(conf)
self.volume = volume
self.run_time = 0
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.bytes_running_time = 0
self.bytes_processed = 0
self.total_bytes_processed = 0
self.total_chunks_processed = 0
self.dry_run = true_value(
conf.get('dry_run', False))
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
self.max_bytes_per_second = int_value(
conf.get('bytes_per_second'), 10000000)
self.rdir_fetch_limit = int_value(
conf.get('rdir_fetch_limit'), 100)
self.blob_client = BlobClient()
self.container_client = ContainerClient(conf)
self.rdir_client = RdirClient(conf)
def rebuilder_pass_with_lock(self):
self.rdir_client.admin_lock(self.volume,
"rebuilder on %s" % gethostname())
try:
self.rebuilder_pass()
finally:
self.rdir_client.admin_unlock(self.volume)
def rebuilder_pass(self):
start_time = report_time = time.time()
total_errors = 0
rebuilder_time = 0
chunks = self.rdir_client.chunk_fetch(self.volume,
limit=self.rdir_fetch_limit,
rebuild=True)
for container_id, content_id, chunk_id, data in chunks:
loop_time = time.time()
if self.dry_run:
self.dryrun_chunk_rebuild(container_id, content_id, chunk_id)
else:
self.safe_chunk_rebuild(container_id, content_id, chunk_id)
self.chunks_run_time = ratelimit(
self.chunks_run_time,
self.max_chunks_per_second
)
self.total_chunks_processed += 1
now = time.time()
if now - self.last_reported >= self.report_interval:
self.logger.info(
'%(start_time)s '
'%(passes)d '
'%(errors)d '
'%(c_rate).2f '
'%(b_rate).2f '
'%(total).2f '
'%(rebuilder_time).2f'
'%(rebuilder_rate).2f' % {
'start_time': time.ctime(report_time),
'passes': self.passes,
'errors': self.errors,
'c_rate': self.passes / (now - report_time),
'b_rate': self.bytes_processed / (now - report_time),
'total': (now - start_time),
'rebuilder_time': rebuilder_time,
'rebuilder_rate': rebuilder_time / (now - start_time)
}
)
report_time = now
total_errors += self.errors
self.passes = 0
self.bytes_processed = 0
self.last_reported = now
rebuilder_time += (now - loop_time)
elapsed = (time.time() - start_time) or 0.000001
self.logger.info(
'%(elapsed).02f '
'%(errors)d '
'%(chunk_rate).2f '
'%(bytes_rate).2f '
'%(rebuilder_time).2f '
'%(rebuilder_rate).2f' % {
'elapsed': elapsed,
'errors': total_errors + self.errors,
'chunk_rate': self.total_chunks_processed / elapsed,
'bytes_rate': self.total_bytes_processed / elapsed,
'rebuilder_time': rebuilder_time,
'rebuilder_rate': rebuilder_time / elapsed
}
)
def dryrun_chunk_rebuild(self, container_id, content_id, chunk_id):
self.logger.info("[dryrun] Rebuilding "
"container %s, content %s, chunk %s"
% (container_id, content_id, chunk_id))
self.passes += 1
def safe_chunk_rebuild(self, container_id, content_id, chunk_id):
self.logger.info('Rebuilding (container %s, content %s, chunk %s)'
% (container_id, content_id, chunk_id))
try:
self.chunk_rebuild(container_id, content_id, chunk_id)
except Exception as e:
self.errors += 1
self.logger.error('ERROR while rebuilding chunk %s|%s|%s) : %s',
container_id, content_id, chunk_id, e)
self.passes += 1
def _meta2_get_chunks_at_pos(self, container_id, content_id, chunk_id):
current_chunk_url = 'http://%s/%s' % (self.volume, chunk_id)
try:
data = self.container_client.content_show(
cid=container_id, content=content_id)
except exc.NotFound:
raise exc.OrphanChunk('Content not found')
current_chunk = None
for c in data:
if c['url'] == current_chunk_url:
current_chunk = c
break
if not current_chunk:
raise exc.OrphanChunk('Chunk not found in content')
duplicate_chunks = []
for c in data:
if c['pos'] == current_chunk['pos'] \
and c['url'] != current_chunk['url']:
duplicate_chunks.append(c)
if len(duplicate_chunks) == 0:
raise exc.UnrecoverableContent('No copy of missing chunk')
return current_chunk, duplicate_chunks
def _meta2_get_spare_chunk(self, container_id, content_id, notin, broken):
spare_data = {'notin': notin,
'broken': [broken],
'size': 0}
try:
spare_resp = self.container_client.content_spare(
cid=container_id, content=content_id, data=spare_data)
except ClientException as e:
raise exc.SpareChunkException('No spare chunk (%s)' % e.message)
return spare_resp['chunks'][0]
def _meta2_replace_chunk(self, container_id, content_id,
current_chunk, new_chunk):
old = [{'type': 'chunk',
'id': current_chunk['url'],
'hash': current_chunk['hash'],
'size': current_chunk['size'],
'pos': current_chunk['pos'],
'content': content_id}]
new = [{'type': 'chunk',
'id': new_chunk['id'],
'hash': current_chunk['hash'],
'size': current_chunk['size'],
'pos': current_chunk['pos'],
'content': content_id}]
update_data = {'old': old, 'new': new}
self.container_client.container_raw_update(
cid=container_id, data=update_data)
# TODO rain support
def chunk_rebuild(self, container_id, content_id, chunk_id):
current_chunk, duplicate_chunks = self._meta2_get_chunks_at_pos(
container_id, content_id, chunk_id)
spare_chunk = self._meta2_get_spare_chunk(
container_id, content_id, duplicate_chunks, current_chunk)
uploaded = False
for src in duplicate_chunks:
try:
self.blob_client.chunk_copy(src['url'], spare_chunk['id'])
self.logger.debug('copy chunk from %s to %s',
src['url'], spare_chunk['id'])
uploaded = True
break
except Exception as e:
self.logger.debug('Failed to copy chunk from %s to %s: %s',
src['url'], spare_chunk['id'], type(e))
if not uploaded:
raise exc.UnrecoverableContent('No copy available '
'of missing chunk')
self._meta2_replace_chunk(container_id, content_id,
current_chunk, spare_chunk)
self.rdir_client.chunk_push(self.volume, container_id, content_id,
chunk_id, rtime=int(time.time()))
self.bytes_processed += current_chunk['size']
self.total_bytes_processed += current_chunk['size']
class EventWorker(Worker):
def init(self):
eventlet.monkey_patch(os=False)
self.session = requests.Session()
self.cs = ConscienceClient(self.conf)
self.rdir = RdirClient(self.conf)
self._acct_addr = None
self.acct_update = 0
self.graceful_timeout = 1
self.acct_refresh_interval = int_value(
self.conf.get('acct_refresh_interval'), 60
)
self.concurrency = int_value(self.conf.get('concurrency'), 1000)
self.acct_update = true_value(self.conf.get('acct_update', True))
self.rdir_update = true_value(self.conf.get('rdir_update', True))
super(EventWorker, self).init()
def notify(self):
"""TODO"""
pass
def safe_decode_job(self, job):
try:
return json.loads(job)
except Exception as e:
self.logger.warn('ERROR decoding job "%s"', str(e.message))
return None
def run(self):
queue_url = self.conf.get('queue_url', 'tcp://127.0.0.1:11300')
self.beanstalk = Beanstalk.from_url(queue_url)
gt = eventlet.spawn(
self.handle)
while self.alive:
self.notify()
try:
eventlet.sleep(1.0)
except AssertionError:
self.alive = False
break
self.notify()
try:
with Timeout(self.graceful_timeout) as t:
gt.kill(StopServe())
gt.wait()
except Timeout as te:
if te != t:
raise
gt.kill()
def handle(self):
try:
while True:
job_id, data = self.beanstalk.reserve()
try:
event = self.safe_decode_job(data)
if event:
self.process_event(event)
self.beanstalk.delete(job_id)
except Exception:
self.logger.exception("ERROR handling event %s", job_id)
except StopServe:
self.logger.info('Stopping event handler')
def process_event(self, event):
handler = self.get_handler(event)
if not handler:
self.logger.warn("ERROR no handler found for event")
# mark as success
return True
success = True
try:
handler(event)
except Exception:
success = False
finally:
return success
def get_handler(self, event):
event_type = event.get('event')
if not event_type:
return None
if event_type == EventType.CONTAINER_PUT:
return self.handle_container_put
elif event_type == EventType.CONTAINER_DESTROY:
return self.handle_container_destroy
elif event_type == EventType.CONTAINER_UPDATE:
return self.handle_container_update
elif event_type == EventType.OBJECT_PUT:
return self.handle_object_put
elif event_type == EventType.OBJECT_DELETE:
return self.handle_object_delete
elif event_type == EventType.REFERENCE_UPDATE:
return self.handle_reference_update
elif event_type == EventType.CHUNK_PUT:
return self.handle_chunk_put
elif event_type == EventType.CHUNK_DELETE:
return self.handle_chunk_delete
elif event_type == EventType.PING:
return self.handle_ping
else:
return None
@property
def acct_addr(self):
if not self._acct_addr or self.acct_refresh():
try:
acct_instance = self.cs.next_instance(ACCOUNT_SERVICE)
self._acct_addr = acct_instance.get('addr')
self.acct_update = time.time()
except Exception:
self.logger.warn('Unable to find account instance')
return self._acct_addr
def acct_refresh(self):
return (time.time() - self.acct_update) > self.acct_refresh_interval
def handle_container_put(self, event):
"""
Handle container creation.
:param event:
"""
self.logger.debug('worker handle container put')
if not self.acct_update:
return
uri = 'http://%s/v1.0/account/container/update' % self.acct_addr
mtime = event.get('when')
data = event.get('data')
name = data.get('url').get('user')
account = data.get('url').get('account')
event = {'mtime': mtime, 'name': name}
self.session.post(uri, params={'id': account}, json=event)
def handle_container_update(self, event):
"""
Handle container update.
:param event:
"""
self.logger.debug('worker handle container update')
if not self.acct_update:
return
uri = 'http://%s/v1.0/account/container/update' % self.acct_addr
mtime = event.get('when')
data = event.get('data')
name = event.get('url').get('user')
account = event.get('url').get('account')
bytes_count = data.get('bytes-count', 0)
object_count = data.get('object-count', 0)
event = {
'mtime': mtime,
'name': name,
'bytes': bytes_count,
'objects': object_count
}
self.session.post(uri, params={'id': account}, json=event)
def handle_container_destroy(self, event):
"""
Handle container destroy.
:param event:
"""
self.logger.debug('worker handle container destroy')
if not self.acct_update:
return
uri = 'http://%s/v1.0/account/container/update' % self.acct_addr
dtime = event.get('when')
data = event.get('data')
name = data.get('url').get('user')
account = data.get('url').get('account')
event = {'dtime': dtime, 'name': name}
self.session.post(uri, params={'id': account}, data=json.dumps(event))
def handle_object_delete(self, event):
"""
Handle object deletion.
Delete the chunks of the object.
:param event:
"""
self.logger.debug('worker handle object delete')
pile = GreenPile(PARALLEL_CHUNKS_DELETE)
chunks = []
for item in event.get('data'):
if item.get('type') == 'chunks':
chunks.append(item)
if not len(chunks):
self.logger.warn('No chunks found in event data')
return
def delete_chunk(chunk):
resp = None
try:
with Timeout(CHUNK_TIMEOUT):
resp = self.session.delete(chunk['id'])
except (Exception, Timeout) as e:
self.logger.warn('error while deleting chunk %s "%s"',
chunk['id'], str(e.message))
return resp
for chunk in chunks:
pile.spawn(delete_chunk, chunk)
resps = [resp for resp in pile if resp]
for resp in resps:
if resp.status_code == 204:
self.logger.debug('deleted chunk %s' % resp.url)
else:
self.logger.warn('failed to delete chunk %s' % resp.url)
def handle_object_put(self, event):
"""
Handle object creation.
TODO
:param event:
"""
self.logger.debug('worker handle object put')
def handle_reference_update(self, event):
"""
Handle reference update.
TODO
:param event
"""
self.logger.debug('worker handle reference update')
def handle_chunk_put(self, event):
"""
Handle chunk creation.
:param event
"""
if not self.rdir_update:
self.logger.debug('worker skip chunk creation')
return
self.logger.debug('worker handle chunk creation')
when = event.get('when')
data = event.get('data')
volume_id = data.get('volume_id')
del data['volume_id']
container_id = data.get('container_id')
del data['container_id']
content_id = data.get('content_id')
del data['content_id']
chunk_id = data.get('chunk_id')
del data['chunk_id']
data['mtime'] = when
self.rdir.chunk_push(volume_id, container_id, content_id, chunk_id,
**data)
def handle_chunk_delete(self, event):
"""
Handle chunk deletion.
:param event
"""
if not self.rdir_update:
self.logger.debug('worker skip chunk deletion')
return
self.logger.debug('worker handle chunk deletion')
data = event.get('data')
volume_id = data.get('volume_id')
container_id = data.get('container_id')
content_id = data.get('content_id')
chunk_id = data.get('chunk_id')
self.rdir.chunk_delete(volume_id, container_id, content_id, chunk_id)
def handle_ping(self, event):
"""
Handle ping
:param event
"""
self.logger.debug('worker handle ping')
def setUp(self):
super(TestRdirClient, self).setUp()
self.rdir_client = RdirClient({'namespace': "NS"})
self.rdir_client._get_rdir_addr = Mock(return_value="0.1.2.3:4567")
class BlobRebuilderWorker(object):
def __init__(self, conf, logger, volume):
self.conf = conf
self.logger = logger or get_logger(conf)
self.volume = volume
self.run_time = 0
self.passes = 0
self.errors = 0
self.last_reported = 0
self.chunks_run_time = 0
self.bytes_running_time = 0
self.bytes_processed = 0
self.total_bytes_processed = 0
self.total_chunks_processed = 0
self.dry_run = true_value(
conf.get('dry_run', False))
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_chunks_per_second = int_value(
conf.get('chunks_per_second'), 30)
self.max_bytes_per_second = int_value(
conf.get('bytes_per_second'), 10000000)
self.rdir_fetch_limit = int_value(
conf.get('rdir_fetch_limit'), 100)
self.rdir_client = RdirClient(conf)
self.content_factory = ContentFactory(conf)
def rebuilder_pass_with_lock(self):
self.rdir_client.admin_lock(self.volume,
"rebuilder on %s" % gethostname())
try:
self.rebuilder_pass()
finally:
self.rdir_client.admin_unlock(self.volume)
def rebuilder_pass(self):
start_time = report_time = time.time()
total_errors = 0
rebuilder_time = 0
chunks = self.rdir_client.chunk_fetch(self.volume,
limit=self.rdir_fetch_limit,
rebuild=True)
for container_id, content_id, chunk_id, data in chunks:
loop_time = time.time()
if self.dry_run:
self.dryrun_chunk_rebuild(container_id, content_id, chunk_id)
else:
self.safe_chunk_rebuild(container_id, content_id, chunk_id)
self.chunks_run_time = ratelimit(
self.chunks_run_time,
self.max_chunks_per_second
)
self.total_chunks_processed += 1
now = time.time()
if now - self.last_reported >= self.report_interval:
self.logger.info(
'%(start_time)s '
'%(passes)d '
'%(errors)d '
'%(c_rate).2f '
'%(b_rate).2f '
'%(total).2f '
'%(rebuilder_time).2f'
'%(rebuilder_rate).2f' % {
'start_time': time.ctime(report_time),
'passes': self.passes,
'errors': self.errors,
'c_rate': self.passes / (now - report_time),
'b_rate': self.bytes_processed / (now - report_time),
'total': (now - start_time),
'rebuilder_time': rebuilder_time,
'rebuilder_rate': rebuilder_time / (now - start_time)
}
)
report_time = now
total_errors += self.errors
self.passes = 0
self.bytes_processed = 0
self.last_reported = now
rebuilder_time += (now - loop_time)
elapsed = (time.time() - start_time) or 0.000001
self.logger.info(
'%(elapsed).02f '
'%(errors)d '
'%(chunk_rate).2f '
'%(bytes_rate).2f '
'%(rebuilder_time).2f '
'%(rebuilder_rate).2f' % {
'elapsed': elapsed,
'errors': total_errors + self.errors,
'chunk_rate': self.total_chunks_processed / elapsed,
'bytes_rate': self.total_bytes_processed / elapsed,
'rebuilder_time': rebuilder_time,
'rebuilder_rate': rebuilder_time / elapsed
}
)
def dryrun_chunk_rebuild(self, container_id, content_id, chunk_id):
self.logger.info("[dryrun] Rebuilding "
"container %s, content %s, chunk %s",
container_id, content_id, chunk_id)
self.passes += 1
def safe_chunk_rebuild(self, container_id, content_id, chunk_id):
try:
self.chunk_rebuild(container_id, content_id, chunk_id)
except Exception as e:
self.errors += 1
self.logger.error('ERROR while rebuilding chunk %s|%s|%s) : %s',
container_id, content_id, chunk_id, e)
self.passes += 1
def chunk_rebuild(self, container_id, content_id, chunk_id):
self.logger.info('Rebuilding (container %s, content %s, chunk %s)',
container_id, content_id, chunk_id)
try:
content = self.content_factory.get(container_id, content_id)
except ContentNotFound:
raise exc.OrphanChunk('Content not found')
chunk = content.chunks.filter(id=chunk_id).one()
if chunk is None:
raise OrphanChunk("Chunk not found in content")
chunk_size = chunk.size
content.rebuild_chunk(chunk_id)
self.rdir_client.chunk_push(self.volume, container_id, content_id,
chunk_id, rtime=int(time.time()))
self.bytes_processed += chunk_size
self.total_bytes_processed += chunk_size
class EventWorker(object):
def __init__(self, conf, name, context, **kwargs):
self.conf = conf
self.name = name
verbose = kwargs.pop('verbose', False)
self.logger = get_logger(self.conf, verbose=verbose)
self.init_zmq(context)
self.cs = ConscienceClient(self.conf)
self.rdir = RdirClient(self.conf)
self._acct_addr = None
self.acct_update = 0
self.acct_refresh_interval = int_value(
conf.get('acct_refresh_interval'), 60
)
self.acct_update = true_value(
conf.get('acct_update', True))
self.session = requests.Session()
self.failed = False
def start(self):
self.logger.info('worker "%s" starting', self.name)
self.running = True
self.run()
def stop(self):
self.logger.info('worker "%s" stopping', self.name)
self.running = False
def init_zmq(self, context):
socket = context.socket(zmq.REP)
socket.connect('inproc://event-front')
self.socket = socket
def safe_ack(self, msg):
try:
self.socket.send_multipart(msg)
except Exception:
self.logger.warn('Unable to ack event')
def run(self):
try:
while self.running:
msg = self.socket.recv_multipart()
self.logger.debug("msg received: %s" % msg)
event = decode_msg(msg)
success = self.process_event(event)
f = "0" if success else ""
self.safe_ack([msg[0], f])
except Exception as e:
self.logger.warn('ERROR in worker "%s"', e)
self.failed = True
raise e
finally:
self.logger.info('worker "%s" stopped', self.name)
def process_event(self, event):
handler = self.get_handler(event)
if not handler:
self.logger.warn("No handler found")
# mark as success
return True
success = True
try:
handler(event)
except Exception:
success = False
finally:
return success
def get_handler(self, event):
event_type = event.get('event')
if not event_type:
return None
if event_type == EventType.CONTAINER_PUT:
return self.handle_container_put
elif event_type == EventType.CONTAINER_DESTROY:
return self.handle_container_destroy
elif event_type == EventType.CONTAINER_UPDATE:
return self.handle_container_update
elif event_type == EventType.OBJECT_PUT:
return self.handle_object_put
elif event_type == EventType.OBJECT_DELETE:
return self.handle_object_delete
elif event_type == EventType.REFERENCE_UPDATE:
return self.handle_reference_update
elif event_type == EventType.CHUNK_PUT:
return self.handle_chunk_put
elif event_type == EventType.CHUNK_DELETE:
return self.handle_chunk_delete
elif event_type == EventType.PING:
return self.handle_ping
else:
return None
@property
def acct_addr(self):
if not self._acct_addr or self.acct_refresh():
try:
acct_instance = self.cs.next_instance(ACCOUNT_SERVICE)
self._acct_addr = acct_instance.get('addr')
self.acct_update = time.time()
except Exception:
self.logger.warn('Unable to find account instance')
return self._acct_addr
def acct_refresh(self):
return (time.time() - self.acct_update) > self.acct_refresh_interval
def handle_container_put(self, event):
"""
Handle container creation.
:param event:
"""
self.logger.debug('worker "%s" handle container put', self.name)
if not self.acct_update:
return
uri = 'http://%s/v1.0/account/container/update' % self.acct_addr
mtime = event.get('when')
data = event.get('data')
name = data.get('url').get('user')
account = data.get('url').get('account')
event = {'mtime': mtime, 'name': name}
self.session.post(uri, params={'id': account}, data=json.dumps(event))
def handle_container_update(self, event):
"""
Handle container update.
:param event:
"""
self.logger.debug('worker "%s" handle container update', self.name)
if not self.acct_update:
return
uri = 'http://%s/v1.0/account/container/update' % self.acct_addr
mtime = event.get('when')
data = event.get('data')
name = event.get('url').get('user')
account = event.get('url').get('account')
bytes_count = data.get('bytes-count', 0)
object_count = data.get('object-count', 0)
event = {
'mtime': mtime,
'name': name,
'bytes': bytes_count,
'objects': object_count
}
self.session.post(uri, params={'id': account}, data=json.dumps(event))
def handle_container_destroy(self, event):
"""
Handle container destroy.
:param event:
"""
self.logger.debug('worker "%s" handle container destroy', self.name)
if not self.acct_update:
return
uri = 'http://%s/v1.0/account/container/update' % self.acct_addr
dtime = event.get('when')
data = event.get('data')
name = data.get('url').get('user')
account = data.get('url').get('account')
event = {'dtime': dtime, 'name': name}
self.session.post(uri, params={'id': account}, data=json.dumps(event))
def handle_object_delete(self, event):
"""
Handle object deletion.
Delete the chunks of the object.
:param event:
"""
self.logger.debug('worker "%s" handle object delete', self.name)
pile = GreenPile(PARALLEL_CHUNKS_DELETE)
chunks = []
for item in event.get('data'):
if item.get('type') == 'chunks':
chunks.append(item)
if not len(chunks):
self.logger.warn('No chunks found in event data')
return
def delete_chunk(chunk):
resp = None
try:
with Timeout(CHUNK_TIMEOUT):
resp = self.session.delete(chunk['id'])
except (Exception, Timeout) as e:
self.logger.exception(e)
return resp
for chunk in chunks:
pile.spawn(delete_chunk, chunk)
resps = [resp for resp in pile if resp]
for resp in resps:
if resp.status_code == 204:
self.logger.info('deleted chunk %s' % resp.url)
else:
self.logger.warn('failed to delete chunk %s' % resp.url)
def handle_object_put(self, event):
"""
Handle object creation.
TODO
:param event:
"""
self.logger.debug('worker "%s" handle object put', self.name)
def handle_reference_update(self, event):
"""
Handle reference update.
TODO
:param event
"""
self.logger.debug('worker "%s" handle reference update', self.name)
def handle_chunk_put(self, event):
"""
Handle chunk creation.
:param event
"""
self.logger.debug('worker "%s" handle chunk creation', self.name)
when = event.get('when')
data = event.get('data')
volume_id = data.get('volume_id')
del data['volume_id']
container_id = data.get('container_id')
del data['container_id']
content_id = data.get('content_id')
del data['content_id']
chunk_id = data.get('chunk_id')
del data['chunk_id']
data['mtime'] = when
self.rdir.chunk_push(volume_id, container_id, content_id, chunk_id,
**data)
def handle_chunk_delete(self, event):
"""
Handle chunk deletion.
:param event
"""
self.logger.debug('worker "%s" handle chunk deletion', self.name)
data = event.get('data')
volume_id = data.get('volume_id')
container_id = data.get('container_id')
content_id = data.get('content_id')
chunk_id = data.get('chunk_id')
self.rdir.chunk_delete(volume_id, container_id, content_id, chunk_id)
def handle_ping(self, event):
"""
Handle ping
:param event
"""
self.logger.debug('worker "%s" handle ping', self.name)
class TestIndexerCrawler(BaseTestCase):
def setUp(self):
super(TestIndexerCrawler, self).setUp()
self.namespace = self.conf['namespace']
self.rawx_conf = self.conf['services']['rawx'][0]
self.conf = {"namespace": self.namespace,
"volume": self.rawx_conf['path']}
self.rdir_client = RdirClient(self.conf)
def tearDown(self):
super(TestIndexerCrawler, self).tearDown()
def _create_chunk(self, rawx_path, alias="toto"):
container_id = random_id(64)
content_id = random_id(32)
chunk_id = random_id(64)
chunk_dir = "%s/%s" % (rawx_path, chunk_id[0:3])
if not os.path.isdir(chunk_dir):
os.makedirs(chunk_dir)
chunk_path = "%s/%s" % (chunk_dir, chunk_id)
with open(chunk_path, "w") as f:
f.write("toto")
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['chunk_hash'], 32 * '0')
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['chunk_id'], chunk_id)
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['chunk_pos'], '0')
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['chunk_size'], '4')
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['container_id'],
container_id)
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['content_id'], content_id)
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['content_path'], alias)
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['content_policy'],
'TESTPOLICY')
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['content_chunkmethod'],
'plain/nb_copy=3')
xattr.setxattr(
chunk_path, 'user.' + chunk_xattr_keys['content_version'], '0')
return chunk_path, container_id, content_id, chunk_id
def _rdir_get(self, rawx_addr, container_id, content_id, chunk_id):
data = self.rdir_client.chunk_fetch(rawx_addr)
key = (container_id, content_id, chunk_id)
for i_container, i_content, i_chunk, i_value in data:
if (i_container, i_content, i_chunk) == key:
return i_value
return None
def _test_index_chunk(self, alias="toto"):
# create a fake chunk
chunk_path, container_id, content_id, chunk_id = self._create_chunk(
self.rawx_conf['path'], alias)
# index the chunk
indexer = BlobIndexer(self.conf)
with mock.patch('oio.blob.indexer.time.time',
mock.MagicMock(return_value=1234)):
indexer.update_index(chunk_path)
# check rdir
check_value = self._rdir_get(self.rawx_conf['addr'], container_id,
content_id, chunk_id)
self.assertIsNotNone(check_value)
self.assertEqual(check_value['mtime'], 1234)
# index a chunk already indexed
with mock.patch('oio.blob.indexer.time.time',
mock.MagicMock(return_value=4567)):
indexer.update_index(chunk_path)
# check rdir
check_value = self._rdir_get(self.rawx_conf['addr'], container_id,
content_id, chunk_id)
self.assertIsNotNone(check_value)
self.assertEqual(check_value['mtime'], 4567)
def test_index_chunk(self):
return self._test_index_chunk()
def test_index_unicode_chunk(self):
return self._test_index_chunk('a%%%s%d%xàç"\r\n{0}€ 1+1=2/\\$\t_')
def test_index_chunk_missing_xattr(self):
# create a fake chunk
chunk_path, container_id, content_id, chunk_id = self._create_chunk(
self.rawx_conf['path'])
# remove mandatory xattr
xattr.removexattr(
chunk_path, 'user.' + chunk_xattr_keys['container_id'])
# try to index the chunk
indexer = BlobIndexer(self.conf)
self.assertRaises(FaultyChunk, indexer.update_index, chunk_path)
class TestIndexerCrawler(BaseTestCase):
def setUp(self):
super(TestIndexerCrawler, self).setUp()
self.namespace = self.conf["namespace"]
self.gridconf = {"namespace": self.namespace}
self.rdir_client = RdirClient(self.gridconf)
def tearDown(self):
super(TestIndexerCrawler, self).tearDown()
def _create_chunk(self, rawx_path):
container_id = generate_id(64)
content_id = generate_id(64)
chunk_id = generate_id(64)
chunk_dir = "%s/%s" % (rawx_path, chunk_id[0:2])
if not os.path.isdir(chunk_dir):
os.makedirs(chunk_dir)
chunk_path = "%s/%s" % (chunk_dir, chunk_id)
with open(chunk_path, "w") as f:
f.write("toto")
xattr.setxattr(chunk_path, "user.grid.chunk.hash", 32 * "0")
xattr.setxattr(chunk_path, "user.grid.chunk.id", chunk_id)
xattr.setxattr(chunk_path, "user.grid.chunk.position", "0")
xattr.setxattr(chunk_path, "user.grid.chunk.size", "4")
xattr.setxattr(chunk_path, "user.grid.content.container", container_id)
xattr.setxattr(chunk_path, "user.grid.content.id", content_id)
xattr.setxattr(chunk_path, "user.grid.content.nbchunk", "1")
xattr.setxattr(chunk_path, "user.grid.content.path", "toto")
xattr.setxattr(chunk_path, "user.grid.content.size", "4")
xattr.setxattr(chunk_path, "user.grid.content.version", "0")
return chunk_path, container_id, content_id, chunk_id
def _rdir_get(self, rawx_addr, container_id, content_id, chunk_id):
data = self.rdir_client.chunk_fetch(rawx_addr)
key = (container_id, content_id, chunk_id)
for i_container, i_content, i_chunk, i_value in data:
if (i_container, i_content, i_chunk) == key:
return i_value
return None
def test_index_chunk(self):
rawx_conf = self.conf["rawx"][0]
# create a fake chunk
chunk_path, container_id, content_id, chunk_id = self._create_chunk(rawx_conf["path"])
# index the chunk
indexer = BlobIndexerWorker(self.gridconf, None, rawx_conf["path"])
with mock.patch("oio.blob.indexer.time.time", mock.MagicMock(return_value=1234)):
indexer.update_index(chunk_path)
# check rdir
check_value = self._rdir_get(rawx_conf["addr"], container_id, content_id, chunk_id)
self.assertIsNotNone(check_value)
self.assertEqual(check_value["content_nbchunks"], 1)
self.assertEqual(check_value["chunk_hash"], 32 * "0")
self.assertEqual(check_value["content_size"], 4)
self.assertEqual(check_value["content_path"], "toto")
self.assertEqual(check_value["chunk_position"], "0")
self.assertEqual(check_value["chunk_size"], 4)
self.assertEqual(check_value["mtime"], 1234)
self.assertEqual(check_value["content_version"], 0)
# index a chunk already indexed
with mock.patch("oio.blob.indexer.time.time", mock.MagicMock(return_value=4567)):
indexer.update_index(chunk_path)
# check rdir
check_value = self._rdir_get(rawx_conf["addr"], container_id, content_id, chunk_id)
self.assertIsNotNone(check_value)
self.assertEqual(check_value["mtime"], 4567)
def test_index_chunk_missing_xattr(self):
rawx_conf = self.conf["rawx"][0]
# create a fake chunk
chunk_path, container_id, content_id, chunk_id = self._create_chunk(rawx_conf["path"])
# remove mandatory xattr
xattr.removexattr(chunk_path, "user.grid.chunk.hash")
# try to index the chunk
indexer = BlobIndexerWorker(self.gridconf, None, rawx_conf["path"])
self.assertRaises(FaultyChunk, indexer.update_index, chunk_path)
