class TestContentFactory(BaseTestCase):
def setUp(self):
super(TestContentFactory, self).setUp()
self.namespace = self.conf['namespace']
self.chunk_size = self.conf['chunk_size']
self.gridconf = {"namespace": self.namespace}
self.content_factory = ContentFactory(self.gridconf)
self.container_name = "TestContentFactory%f" % time.time()
self.blob_client = BlobClient()
self.container_client = ContainerClient(self.gridconf)
self.container_client.container_create(acct=self.account,
ref=self.container_name)
self.container_id = cid_from_name(self.account,
self.container_name).upper()
def tearDown(self):
super(TestContentFactory, self).tearDown()
def test_extract_datasec(self):
self.content_factory.ns_info = {
"data_security": {
"DUPONETWO": "DUP:distance=1|nb_copy=2",
"RAIN": "RAIN:k=6|m=2|algo=liber8tion"
},
"storage_policy": {
"RAIN": "NONE:RAIN:NONE",
"SINGLE": "NONE:NONE:NONE",
"TWOCOPIES": "NONE:DUPONETWO:NONE"
}
}
ds_type, ds_args = self.content_factory._extract_datasec("RAIN")
self.assertEqual(ds_type, "RAIN")
self.assertEqual(ds_args, {
"k": "6",
"m": "2",
"algo": "liber8tion"
})
ds_type, ds_args = self.content_factory._extract_datasec("SINGLE")
self.assertEqual(ds_type, "DUP")
self.assertEqual(ds_args, {
"nb_copy": "1",
"distance": "0"
})
ds_type, ds_args = self.content_factory._extract_datasec("TWOCOPIES")
self.assertEqual(ds_type, "DUP")
self.assertEqual(ds_args, {
"nb_copy": "2",
"distance": "1"
})
self.assertRaises(InconsistentContent,
self.content_factory._extract_datasec,
"UnKnOwN")
def test_get_rain(self):
meta = {
"chunk-method": "plain/rain?algo=liber8tion&k=6&m=2",
"ctime": "1450176946",
"deleted": "False",
"hash": "E952A419957A6E405BFC53EC65483F73",
"hash-method": "md5",
"id": "3FA2C4A1ED2605005335A276890EC458",
"length": "658",
"mime-type": "application/octet-stream",
"name": "tox.ini",
"policy": "RAIN",
"version": "1450176946676289"
}
chunks = [
{
"url": "http://127.0.0.1:6012/A0A0",
"pos": "0.p0", "size": 512,
"hash": "E7D4E4AD460971CA2E3141F2102308D4"},
{
"url": "http://127.0.0.1:6010/A01",
"pos": "0.1", "size": 146,
"hash": "760AB5DA7C51A3654F1CA622687CD6C3"},
{
"url": "http://127.0.0.1:6011/A00",
"pos": "0.0", "size": 512,
"hash": "B1D08B86B8CAA90A2092CCA0DF9201DB"},
{
"url": "http://127.0.0.1:6013/A0A1",
"pos": "0.p1", "size": 512,
"hash": "DA9D7F72AEEA5791565724424CE45C16"}
]
self.content_factory.container_client.content_show = Mock(
return_value=(meta, chunks))
c = self.content_factory.get("xxx_container_id", "xxx_content_id")
self.assertEqual(type(c), RainContent)
self.assertEqual(c.content_id, "3FA2C4A1ED2605005335A276890EC458")
self.assertEqual(c.length, 658)
self.assertEqual(c.path, "tox.ini")
self.assertEqual(c.version, "1450176946676289")
self.assertEqual(c.algo, "liber8tion")
self.assertEqual(c.k, 6)
self.assertEqual(c.m, 2)
self.assertEqual(len(c.chunks), 4)
self.assertEqual(c.chunks[0].raw(), chunks[2])
self.assertEqual(c.chunks[1].raw(), chunks[1])
self.assertEqual(c.chunks[2].raw(), chunks[0])
self.assertEqual(c.chunks[3].raw(), chunks[3])
def test_get_dup(self):
meta = {
"chunk-method": "plain/bytes",
"ctime": "1450176946",
"deleted": "False",
"hash": "E952A419957A6E405BFC53EC65483F73",
"hash-method": "md5",
"id": "3FA2C4A1ED2605005335A276890EC458",
"length": "658",
"mime-type": "application/octet-stream",
"name": "tox.ini",
"policy": "TWOCOPIES",
"version": "1450176946676289"
}
chunks = [
{
"url": "http://127.0.0.1:6010/A0",
"pos": "0", "size": 658,
"hash": "E952A419957A6E405BFC53EC65483F73"},
{
"url": "http://127.0.0.1:6011/A1",
"pos": "0", "size": 658,
"hash": "E952A419957A6E405BFC53EC65483F73"}
]
self.content_factory.container_client.content_show = Mock(
return_value=(meta, chunks))
c = self.content_factory.get("xxx_container_id", "xxx_content_id")
self.assertEqual(type(c), DupContent)
self.assertEqual(c.content_id, "3FA2C4A1ED2605005335A276890EC458")
self.assertEqual(c.length, 658)
self.assertEqual(c.path, "tox.ini")
self.assertEqual(c.version, "1450176946676289")
self.assertEqual(c.nb_copy, 2)
self.assertEqual(c.distance, 1)
self.assertEqual(len(c.chunks), 2)
self.assertEqual(c.chunks[0].raw(), chunks[0])
self.assertEqual(c.chunks[1].raw(), chunks[1])
def test_get_unknown_content(self):
self.assertRaises(ContentNotFound, self.content_factory.get,
self.container_id, "1234")
def test_new_rain(self):
meta = {
"chunk-method": "plain/rain?algo=liber8tion&k=6&m=2",
"ctime": "1450341162",
"deleted": "False",
"hash": "",
"hash-method": "md5",
"id": "F4B1C8DD132705007DE8B43D0709DAA2",
"length": "1000",
"mime-type": "application/octet-stream",
"name": "titi",
"policy": "RAIN",
"version": "1450341162332663"
}
chunks = [
{
"url": "http://127.0.0.1:6010/0_p1",
"pos": "0.p1", "size": 1048576,
"hash": "00000000000000000000000000000000"},
{
"url": "http://127.0.0.1:6011/0_p0",
"pos": "0.p0", "size": 1048576,
"hash": "00000000000000000000000000000000"},
{
"url": "http://127.0.0.1:6016/0_1",
"pos": "0.1", "size": 1048576,
"hash": "00000000000000000000000000000000"},
{
"url": "http://127.0.0.1:6017/0_0",
"pos": "0.0", "size": 1048576,
"hash": "00000000000000000000000000000000"}
]
self.content_factory.container_client.content_prepare = Mock(
return_value=(meta, chunks))
c = self.content_factory.new("xxx_container_id", "titi",
1000, "RAIN")
self.assertEqual(type(c), RainContent)
self.assertEqual(c.content_id, "F4B1C8DD132705007DE8B43D0709DAA2")
self.assertEqual(c.length, 1000)
self.assertEqual(c.path, "titi")
self.assertEqual(c.version, "1450341162332663")
self.assertEqual(c.algo, "liber8tion")
self.assertEqual(c.k, 6)
self.assertEqual(c.m, 2)
self.assertEqual(len(c.chunks), 4)
self.assertEqual(c.chunks[0].raw(), chunks[3])
self.assertEqual(c.chunks[1].raw(), chunks[2])
self.assertEqual(c.chunks[2].raw(), chunks[1])
self.assertEqual(c.chunks[3].raw(), chunks[0])
def _new_content(self, stgpol, data, path="titi"):
old_content = self.content_factory.new(self.container_id, path,
len(data), stgpol)
old_content.upload(StringIO.StringIO(data))
return self.content_factory.get(self.container_id,
old_content.content_id)
def _test_change_policy(self, data_size, old_policy, new_policy):
if (old_policy == "RAIN" or new_policy == "RAIN") \
and len(self.conf['rawx']) < 8:
self.skipTest("RAIN: Need more than 8 rawx to run")
data = random_data(data_size)
obj_type = {
"SINGLE": DupContent,
"TWOCOPIES": DupContent,
"THREECOPIES": DupContent,
"RAIN": RainContent
}
old_content = self._new_content(old_policy, data)
self.assertEqual(type(old_content), obj_type[old_policy])
changed_content = self.content_factory.change_policy(
old_content.container_id, old_content.content_id, new_policy)
self.assertRaises(NotFound, self.container_client.content_show,
self.account,
cid=old_content.container_id,
content=old_content.content_id)
new_content = self.content_factory.get(self.container_id,
changed_content.content_id)
self.assertEqual(type(new_content), obj_type[new_policy])
downloaded_data = "".join(new_content.download())
self.assertEqual(downloaded_data, data)
def test_change_content_0_byte_policy_single_to_rain(self):
self._test_change_policy(0, "SINGLE", "RAIN")
def test_change_content_0_byte_policy_rain_to_twocopies(self):
self._test_change_policy(0, "RAIN", "TWOCOPIES")
def test_change_content_1_byte_policy_single_to_rain(self):
self._test_change_policy(1, "SINGLE", "RAIN")
def test_change_content_chunksize_bytes_policy_twocopies_to_rain(self):
self._test_change_policy(self.chunk_size, "TWOCOPIES", "RAIN")
def test_change_content_2xchunksize_bytes_policy_threecopies_to_rain(self):
self._test_change_policy(self.chunk_size * 2, "THREECOPIES", "RAIN")
def test_change_content_1_byte_policy_rain_to_threecopies(self):
self._test_change_policy(1, "RAIN", "THREECOPIES")
def test_change_content_chunksize_bytes_policy_rain_to_twocopies(self):
self._test_change_policy(self.chunk_size, "RAIN", "TWOCOPIES")
def test_change_content_2xchunksize_bytes_policy_rain_to_single(self):
self._test_change_policy(self.chunk_size * 2, "RAIN", "SINGLE")
def test_change_content_0_byte_policy_twocopies_to_threecopies(self):
self._test_change_policy(0, "TWOCOPIES", "THREECOPIES")
def test_change_content_chunksize_bytes_policy_single_to_twocopies(self):
self._test_change_policy(self.chunk_size, "SINGLE", "TWOCOPIES")
def test_change_content_2xchunksize_bytes_policy_3copies_to_single(self):
self._test_change_policy(self.chunk_size * 2, "THREECOPIES", "SINGLE")
def test_change_content_with_same_policy(self):
data = random_data(10)
old_content = self._new_content("TWOCOPIES", data)
changed_content = self.content_factory.change_policy(
old_content.container_id, old_content.content_id, "TWOCOPIES")
self.assertEqual(old_content.content_id, changed_content.content_id)
def test_change_policy_unknown_content(self):
self.assertRaises(ContentNotFound, self.content_factory.change_policy,
self.container_id, "1234", "SINGLE")
def test_change_policy_unknown_storage_policy(self):
data = random_data(10)
old_content = self._new_content("TWOCOPIES", data)
self.assertRaises(ClientException, self.content_factory.change_policy,
self.container_id, old_content.content_id, "UnKnOwN")
def _test_move_chunk(self, policy):
data = random_data(self.chunk_size)
content = self._new_content(policy, data)
chunk_id = content.chunks.filter(metapos=0)[0].id
chunk_url = content.chunks.filter(metapos=0)[0].url
chunk_meta, chunk_stream = self.blob_client.chunk_get(chunk_url)
chunk_hash = md5_stream(chunk_stream)
new_chunk = content.move_chunk(chunk_id)
content_updated = self.content_factory.get(self.container_id,
content.content_id)
hosts = []
for c in content_updated.chunks.filter(metapos=0):
self.assertThat(hosts, Not(Contains(c.host)))
self.assertNotEquals(c.id, chunk_id)
hosts.append(c.host)
new_chunk_meta, new_chunk_stream = self.blob_client.chunk_get(
new_chunk["url"])
new_chunk_hash = md5_stream(new_chunk_stream)
self.assertEqual(new_chunk_hash, chunk_hash)
del chunk_meta["chunk_id"]
del new_chunk_meta["chunk_id"]
self.assertEqual(new_chunk_meta, chunk_meta)
def test_single_move_chunk(self):
self._test_move_chunk("SINGLE")
def test_twocopies_move_chunk(self):
self._test_move_chunk("TWOCOPIES")
def test_rain_move_chunk(self):
if len(self.conf['rawx']) < 9:
self.skipTest("Need more than 8 rawx")
self._test_move_chunk("RAIN")
def test_move_chunk_not_in_content(self):
data = random_data(self.chunk_size)
content = self._new_content("TWOCOPIES", data)
with ExpectedException(OrphanChunk):
content.move_chunk("1234")
def test_strange_paths(self):
strange_paths = [
"Annual report.txt",
"foo+bar=foobar.txt",
"100%_bug_free.c",
"forward/slash/allowed",
"I\\put\\backslashes\\and$dollar$signs$in$file$names",
"Je suis tombé sur la tête, mais ça va bien.",
"%s%f%u%d%%",
"carriage\rreturn",
"line\nfeed",
"ta\tbu\tla\ttion",
"controlchars",
]
answers = dict()
for cname in strange_paths:
content = self._new_content("SINGLE", "nobody cares", cname)
answers[cname] = content
listing = self.container_client.container_list(self.account,
self.container_name)
obj_set = {k["name"].encode("utf8", "ignore")
for k in listing["objects"]}
try:
# Ensure the saved path is the one we gave the object
for cname in answers:
self.assertEqual(cname, answers[cname].path)
# Ensure all objects appear in listing
for cname in strange_paths:
self.assertIn(cname, obj_set)
finally:
# Cleanup
for cname in answers:
try:
content.delete()
except:
pass
class Checker(object):
def __init__(self, namespace, concurrency=50, error_file=None):
self.pool = GreenPool(concurrency)
self.error_file = error_file
if self.error_file:
f = open(self.error_file, 'a')
self.error_writer = csv.writer(f, delimiter=' ')
conf = {'namespace': namespace}
self.account_client = AccountClient(conf)
self.container_client = ContainerClient(conf)
self.blob_client = BlobClient()
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 = {}
def write_error(self, target):
error = [target.account]
if target.container:
error.append(target.container)
if target.obj:
error.append(target.obj)
if target.chunk:
error.append(target.chunk)
self.error_writer.writerow(error)
def check_chunk(self, target):
chunk = target.chunk
obj_listing = self.check_obj(target)
error = False
if chunk not in obj_listing:
print(' Chunk %s missing in object listing' % target)
error = True
# checksum = None
else:
# TODO check checksum match
# checksum = obj_listing[chunk]['hash']
pass
try:
self.blob_client.chunk_head(chunk)
except exc.NotFound as e:
self.chunk_not_found += 1
error = True
print(' Not found chunk "%s": %s' % (target, str(e)))
except Exception as e:
self.chunk_exceptions += 1
error = True
print(' Exception chunk "%s": %s' % (target, str(e)))
if error and self.error_file:
self.write_error(target)
self.chunks_checked += 1
def check_obj(self, target, recurse=False):
account = target.account
container = target.container
obj = target.obj
if (account, container, obj) in self.running:
self.running[(account, container, obj)].wait()
if (account, container, obj) in self.list_cache:
return self.list_cache[(account, container, obj)]
self.running[(account, container, obj)] = Event()
print('Checking object "%s"' % target)
container_listing = self.check_container(target)
error = False
if obj not in container_listing:
print(' Object %s missing in container listing' % target)
error = True
# checksum = None
else:
# TODO check checksum match
# checksum = container_listing[obj]['hash']
pass
results = []
try:
_, resp = self.container_client.content_show(acct=account,
ref=container,
path=obj)
except exc.NotFound as e:
self.object_not_found += 1
error = True
print(' Not found object "%s": %s' % (target, str(e)))
except Exception as e:
self.object_exceptions += 1
error = True
print(' Exception object "%s": %s' % (target, str(e)))
else:
results = resp
chunk_listing = dict()
for chunk in results:
chunk_listing[chunk['url']] = chunk
self.objects_checked += 1
self.list_cache[(account, container, obj)] = chunk_listing
self.running[(account, container, obj)].send(True)
del self.running[(account, container, obj)]
if recurse:
for chunk in chunk_listing:
t = target.copy()
t.chunk = chunk
self.pool.spawn_n(self.check_chunk, t)
if error and self.error_file:
self.write_error(target)
return chunk_listing
def check_container(self, target, recurse=False):
account = target.account
container = target.container
if (account, container) in self.running:
self.running[(account, container)].wait()
if (account, container) in self.list_cache:
return self.list_cache[(account, container)]
self.running[(account, container)] = Event()
print('Checking container "%s"' % target)
account_listing = self.check_account(target)
error = False
if container not in account_listing:
error = True
print(' Container %s missing in account listing' % target)
marker = None
results = []
while True:
try:
resp = self.container_client.container_list(acct=account,
ref=container,
marker=marker)
except exc.NotFound as e:
self.container_not_found += 1
error = True
print(' Not found container "%s": %s' % (target, str(e)))
break
except Exception as e:
self.container_exceptions += 1
error = True
print(' Exception container "%s": %s' % (target, str(e)))
break
if resp['objects']:
marker = resp['objects'][-1]['name']
else:
break
results.extend(resp['objects'])
container_listing = dict()
for obj in results:
container_listing[obj['name']] = obj
self.containers_checked += 1
self.list_cache[(account, container)] = container_listing
self.running[(account, container)].send(True)
del self.running[(account, container)]
if recurse:
for obj in container_listing:
t = target.copy()
t.obj = obj
self.pool.spawn_n(self.check_obj, t, True)
if error and self.error_file:
self.write_error(target)
return container_listing
def check_account(self, target, recurse=False):
account = target.account
if account in self.running:
self.running[account].wait()
if account in self.list_cache:
return self.list_cache[account]
self.running[account] = Event()
print('Checking account "%s"' % target)
error = False
marker = None
results = []
while True:
try:
resp = self.account_client.containers_list(account,
marker=marker)
except Exception as e:
self.account_exceptions += 1
error = True
print(' Exception account "%s": %s' % (target, str(e)))
break
if resp['listing']:
marker = resp['listing'][-1][0]
else:
break
results.extend(resp['listing'])
containers = dict()
for e in results:
containers[e[0]] = (e[1], e[2])
self.list_cache[account] = containers
self.running[account].send(True)
del self.running[account]
self.accounts_checked += 1
if recurse:
for container in containers:
t = target.copy()
t.container = container
self.pool.spawn_n(self.check_container, t, True)
if error and self.error_file:
self.write_error(target)
return containers
def check(self, target):
if target.chunk and target.obj and target.container:
self.pool.spawn_n(self.check_chunk, target)
elif target.obj and target.container:
self.pool.spawn_n(self.check_obj, target, True)
elif target.container:
self.pool.spawn_n(self.check_container, target, True)
else:
self.pool.spawn_n(self.check_account, target, True)
def wait(self):
self.pool.waitall()
def report(self):
def _report_stat(name, stat):
print("{0:18}: {1}".format(name, stat))
print()
print('Report')
_report_stat("Accounts checked", self.accounts_checked)
if self.account_not_found:
_report_stat("Missing accounts", self.account_not_found)
if self.account_exceptions:
_report_stat("Exceptions", self.account_not_found)
print()
_report_stat("Containers checked", self.containers_checked)
if self.container_not_found:
_report_stat("Missing containers", self.container_not_found)
if self.container_exceptions:
_report_stat("Exceptions", self.container_exceptions)
print()
_report_stat("Objects checked", self.objects_checked)
if self.object_not_found:
_report_stat("Missing objects", self.object_not_found)
if self.object_exceptions:
_report_stat("Exceptions", self.object_exceptions)
print()
_report_stat("Chunks checked", self.chunks_checked)
if self.chunk_not_found:
_report_stat("Missing chunks", self.chunk_not_found)
if self.chunk_exceptions:
_report_stat("Exceptions", self.chunk_exceptions)
class Checker(object):
def __init__(self, namespace, concurrency=50,
error_file=None, rebuild_file=None):
self.pool = GreenPool(concurrency)
self.error_file = error_file
if self.error_file:
f = open(self.error_file, 'a')
self.error_writer = csv.writer(f, delimiter=' ')
self.rebuild_file = rebuild_file
if self.rebuild_file:
fd = open(self.rebuild_file, 'a')
self.rebuild_writer = csv.writer(fd, delimiter='|')
conf = {'namespace': namespace}
self.account_client = AccountClient(conf)
self.container_client = ContainerClient(conf)
self.blob_client = BlobClient()
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 = {}
def write_error(self, target):
error = [target.account]
if target.container:
error.append(target.container)
if target.obj:
error.append(target.obj)
if target.chunk:
error.append(target.chunk)
self.error_writer.writerow(error)
def write_rebuilder_input(self, target, obj_meta, ct_meta):
cid = ct_meta['properties']['sys.name'].split('.', 1)[0]
self.rebuild_writer.writerow((cid, obj_meta['id'], target.chunk))
def _check_chunk_xattr(self, target, obj_meta, xattr_meta):
error = False
# Composed position -> erasure coding
attr_prefix = 'meta' if '.' in obj_meta['pos'] else ''
attr_key = attr_prefix + 'chunk_size'
if str(obj_meta['size']) != xattr_meta.get(attr_key):
print(" Chunk %s '%s' xattr (%s) "
"differs from size in meta2 (%s)" %
(target, attr_key, xattr_meta.get(attr_key),
obj_meta['size']))
error = True
attr_key = attr_prefix + 'chunk_hash'
if obj_meta['hash'] != xattr_meta.get(attr_key):
print(" Chunk %s '%s' xattr (%s) "
"differs from hash in meta2 (%s)" %
(target, attr_key, xattr_meta.get(attr_key),
obj_meta['hash']))
error = True
return error
def check_chunk(self, target):
chunk = target.chunk
obj_listing, obj_meta = self.check_obj(target)
error = False
if chunk not in obj_listing:
print(' Chunk %s missing from object listing' % target)
error = True
db_meta = dict()
else:
db_meta = obj_listing[chunk]
try:
xattr_meta = self.blob_client.chunk_head(chunk)
except exc.NotFound as e:
self.chunk_not_found += 1
error = True
print(' Not found chunk "%s": %s' % (target, str(e)))
except Exception as e:
self.chunk_exceptions += 1
error = True
print(' Exception chunk "%s": %s' % (target, str(e)))
else:
if db_meta:
error = self._check_chunk_xattr(target, db_meta, xattr_meta)
if error:
if self.error_file:
self.write_error(target)
if self.rebuild_file:
self.write_rebuilder_input(
target, obj_meta,
self.list_cache[(target.account, target.container)][1])
self.chunks_checked += 1
def check_obj(self, target, recurse=False):
account = target.account
container = target.container
obj = target.obj
if (account, container, obj) in self.running:
self.running[(account, container, obj)].wait()
if (account, container, obj) in self.list_cache:
return self.list_cache[(account, container, obj)]
self.running[(account, container, obj)] = Event()
print('Checking object "%s"' % target)
container_listing, ct_meta = self.check_container(target)
error = False
if obj not in container_listing:
print(' Object %s missing from container listing' % target)
error = True
# checksum = None
else:
# TODO check checksum match
# checksum = container_listing[obj]['hash']
pass
results = []
meta = dict()
try:
meta, results = self.container_client.content_show(
acct=account, ref=container, path=obj)
except exc.NotFound as e:
self.object_not_found += 1
error = True
print(' Not found object "%s": %s' % (target, str(e)))
except Exception as e:
self.object_exceptions += 1
error = True
print(' Exception object "%s": %s' % (target, str(e)))
chunk_listing = dict()
for chunk in results:
chunk_listing[chunk['url']] = chunk
self.objects_checked += 1
self.list_cache[(account, container, obj)] = (chunk_listing, meta)
self.running[(account, container, obj)].send(True)
del self.running[(account, container, obj)]
if recurse:
for chunk in chunk_listing:
t = target.copy()
t.chunk = chunk
self.pool.spawn_n(self.check_chunk, t)
if error and self.error_file:
self.write_error(target)
return chunk_listing, meta
def check_container(self, target, recurse=False):
account = target.account
container = target.container
if (account, container) in self.running:
self.running[(account, container)].wait()
if (account, container) in self.list_cache:
return self.list_cache[(account, container)]
self.running[(account, container)] = Event()
print('Checking container "%s"' % target)
account_listing = self.check_account(target)
error = False
if container not in account_listing:
error = True
print(' Container %s missing from account listing' % target)
marker = None
results = []
ct_meta = dict()
while True:
try:
resp = self.container_client.container_list(
acct=account, ref=container, marker=marker)
except exc.NotFound as e:
self.container_not_found += 1
error = True
print(' Not found container "%s": %s' % (target, str(e)))
break
except Exception as e:
self.container_exceptions += 1
error = True
print(' Exception container "%s": %s' % (target, str(e)))
break
if resp['objects']:
marker = resp['objects'][-1]['name']
results.extend(resp['objects'])
else:
ct_meta = resp
ct_meta.pop('objects')
break
container_listing = dict()
for obj in results:
container_listing[obj['name']] = obj
self.containers_checked += 1
self.list_cache[(account, container)] = container_listing, ct_meta
self.running[(account, container)].send(True)
del self.running[(account, container)]
if recurse:
for obj in container_listing:
t = target.copy()
t.obj = obj
self.pool.spawn_n(self.check_obj, t, True)
if error and self.error_file:
self.write_error(target)
return container_listing, ct_meta
def check_account(self, target, recurse=False):
account = target.account
if account in self.running:
self.running[account].wait()
if account in self.list_cache:
return self.list_cache[account]
self.running[account] = Event()
print('Checking account "%s"' % target)
error = False
marker = None
results = []
while True:
try:
resp = self.account_client.containers_list(
account, marker=marker)
except Exception as e:
self.account_exceptions += 1
error = True
print(' Exception account "%s": %s' % (target, str(e)))
break
if resp['listing']:
marker = resp['listing'][-1][0]
else:
break
results.extend(resp['listing'])
containers = dict()
for e in results:
containers[e[0]] = (e[1], e[2])
self.list_cache[account] = containers
self.running[account].send(True)
del self.running[account]
self.accounts_checked += 1
if recurse:
for container in containers:
t = target.copy()
t.container = container
self.pool.spawn_n(self.check_container, t, True)
if error and self.error_file:
self.write_error(target)
return containers
def check(self, target):
if target.chunk and target.obj and target.container:
self.pool.spawn_n(self.check_chunk, target)
elif target.obj and target.container:
self.pool.spawn_n(self.check_obj, target, True)
elif target.container:
self.pool.spawn_n(self.check_container, target, True)
else:
self.pool.spawn_n(self.check_account, target, True)
def wait(self):
self.pool.waitall()
def report(self):
def _report_stat(name, stat):
print("{0:18}: {1}".format(name, stat))
print()
print('Report')
_report_stat("Accounts checked", self.accounts_checked)
if self.account_not_found:
_report_stat("Missing accounts", self.account_not_found)
if self.account_exceptions:
_report_stat("Exceptions", self.account_not_found)
print()
_report_stat("Containers checked", self.containers_checked)
if self.container_not_found:
_report_stat("Missing containers", self.container_not_found)
if self.container_exceptions:
_report_stat("Exceptions", self.container_exceptions)
print()
_report_stat("Objects checked", self.objects_checked)
if self.object_not_found:
_report_stat("Missing objects", self.object_not_found)
if self.object_exceptions:
_report_stat("Exceptions", self.object_exceptions)
print()
_report_stat("Chunks checked", self.chunks_checked)
if self.chunk_not_found:
_report_stat("Missing chunks", self.chunk_not_found)
if self.chunk_exceptions:
_report_stat("Exceptions", self.chunk_exceptions)
class TestContentFactory(BaseTestCase):
def setUp(self):
super(TestContentFactory, self).setUp()
self.namespace = self.conf['namespace']
self.chunk_size = self.conf['chunk_size']
self.gridconf = {"namespace": self.namespace}
self.content_factory = ContentFactory(self.gridconf)
self.container_name = "TestContentFactory%f" % time.time()
self.blob_client = BlobClient()
self.container_client = ContainerClient(self.gridconf)
self.container_client.container_create(acct=self.account,
ref=self.container_name)
self.container_id = cid_from_name(self.account,
self.container_name).upper()
def tearDown(self):
super(TestContentFactory, self).tearDown()
def test_extract_datasec(self):
self.content_factory.ns_info = {
"data_security": {
"DUPONETWO": "DUP:distance=1|nb_copy=2",
"RAIN": "RAIN:k=6|m=2|algo=liber8tion"
},
"storage_policy": {
"RAIN": "NONE:RAIN:NONE",
"SINGLE": "NONE:NONE:NONE",
"TWOCOPIES": "NONE:DUPONETWO:NONE"
}
}
ds_type, ds_args = self.content_factory._extract_datasec("RAIN")
self.assertEqual(ds_type, "RAIN")
self.assertEqual(ds_args, {"k": "6", "m": "2", "algo": "liber8tion"})
ds_type, ds_args = self.content_factory._extract_datasec("SINGLE")
self.assertEqual(ds_type, "DUP")
self.assertEqual(ds_args, {"nb_copy": "1", "distance": "0"})
ds_type, ds_args = self.content_factory._extract_datasec("TWOCOPIES")
self.assertEqual(ds_type, "DUP")
self.assertEqual(ds_args, {"nb_copy": "2", "distance": "1"})
self.assertRaises(InconsistentContent,
self.content_factory._extract_datasec, "UnKnOwN")
def test_get_rain(self):
meta = {
"chunk-method": "plain/rain?algo=liber8tion&k=6&m=2",
"ctime": "1450176946",
"deleted": "False",
"hash": "E952A419957A6E405BFC53EC65483F73",
"hash-method": "md5",
"id": "3FA2C4A1ED2605005335A276890EC458",
"length": "658",
"mime-type": "application/octet-stream",
"name": "tox.ini",
"policy": "RAIN",
"version": "1450176946676289"
}
chunks = [{
"url": "http://127.0.0.1:6012/A0A0",
"pos": "0.p0",
"size": 512,
"hash": "E7D4E4AD460971CA2E3141F2102308D4"
}, {
"url": "http://127.0.0.1:6010/A01",
"pos": "0.1",
"size": 146,
"hash": "760AB5DA7C51A3654F1CA622687CD6C3"
}, {
"url": "http://127.0.0.1:6011/A00",
"pos": "0.0",
"size": 512,
"hash": "B1D08B86B8CAA90A2092CCA0DF9201DB"
}, {
"url": "http://127.0.0.1:6013/A0A1",
"pos": "0.p1",
"size": 512,
"hash": "DA9D7F72AEEA5791565724424CE45C16"
}]
self.content_factory.container_client.content_show = Mock(
return_value=(meta, chunks))
c = self.content_factory.get("xxx_container_id", "xxx_content_id")
self.assertEqual(type(c), RainContent)
self.assertEqual(c.content_id, "3FA2C4A1ED2605005335A276890EC458")
self.assertEqual(c.length, 658)
self.assertEqual(c.path, "tox.ini")
self.assertEqual(c.version, "1450176946676289")
self.assertEqual(c.algo, "liber8tion")
self.assertEqual(c.k, 6)
self.assertEqual(c.m, 2)
self.assertEqual(len(c.chunks), 4)
self.assertEqual(c.chunks[0].raw(), chunks[2])
self.assertEqual(c.chunks[1].raw(), chunks[1])
self.assertEqual(c.chunks[2].raw(), chunks[0])
self.assertEqual(c.chunks[3].raw(), chunks[3])
def test_get_dup(self):
meta = {
"chunk-method": "plain/bytes",
"ctime": "1450176946",
"deleted": "False",
"hash": "E952A419957A6E405BFC53EC65483F73",
"hash-method": "md5",
"id": "3FA2C4A1ED2605005335A276890EC458",
"length": "658",
"mime-type": "application/octet-stream",
"name": "tox.ini",
"policy": "TWOCOPIES",
"version": "1450176946676289"
}
chunks = [{
"url": "http://127.0.0.1:6010/A0",
"pos": "0",
"size": 658,
"hash": "E952A419957A6E405BFC53EC65483F73"
}, {
"url": "http://127.0.0.1:6011/A1",
"pos": "0",
"size": 658,
"hash": "E952A419957A6E405BFC53EC65483F73"
}]
self.content_factory.container_client.content_show = Mock(
return_value=(meta, chunks))
c = self.content_factory.get("xxx_container_id", "xxx_content_id")
self.assertEqual(type(c), DupContent)
self.assertEqual(c.content_id, "3FA2C4A1ED2605005335A276890EC458")
self.assertEqual(c.length, 658)
self.assertEqual(c.path, "tox.ini")
self.assertEqual(c.version, "1450176946676289")
self.assertEqual(c.nb_copy, 2)
self.assertEqual(c.distance, 1)
self.assertEqual(len(c.chunks), 2)
self.assertEqual(c.chunks[0].raw(), chunks[0])
self.assertEqual(c.chunks[1].raw(), chunks[1])
def test_get_unknown_content(self):
self.assertRaises(ContentNotFound, self.content_factory.get,
self.container_id, "1234")
def test_new_rain(self):
meta = {
"chunk-method": "plain/rain?algo=liber8tion&k=6&m=2",
"ctime": "1450341162",
"deleted": "False",
"hash": "",
"hash-method": "md5",
"id": "F4B1C8DD132705007DE8B43D0709DAA2",
"length": "1000",
"mime-type": "application/octet-stream",
"name": "titi",
"policy": "RAIN",
"version": "1450341162332663"
}
chunks = [{
"url": "http://127.0.0.1:6010/0_p1",
"pos": "0.p1",
"size": 1048576,
"hash": "00000000000000000000000000000000"
}, {
"url": "http://127.0.0.1:6011/0_p0",
"pos": "0.p0",
"size": 1048576,
"hash": "00000000000000000000000000000000"
}, {
"url": "http://127.0.0.1:6016/0_1",
"pos": "0.1",
"size": 1048576,
"hash": "00000000000000000000000000000000"
}, {
"url": "http://127.0.0.1:6017/0_0",
"pos": "0.0",
"size": 1048576,
"hash": "00000000000000000000000000000000"
}]
self.content_factory.container_client.content_prepare = Mock(
return_value=(meta, chunks))
c = self.content_factory.new("xxx_container_id", "titi", 1000, "RAIN")
self.assertEqual(type(c), RainContent)
self.assertEqual(c.content_id, "F4B1C8DD132705007DE8B43D0709DAA2")
self.assertEqual(c.length, 1000)
self.assertEqual(c.path, "titi")
self.assertEqual(c.version, "1450341162332663")
self.assertEqual(c.algo, "liber8tion")
self.assertEqual(c.k, 6)
self.assertEqual(c.m, 2)
self.assertEqual(len(c.chunks), 4)
self.assertEqual(c.chunks[0].raw(), chunks[3])
self.assertEqual(c.chunks[1].raw(), chunks[2])
self.assertEqual(c.chunks[2].raw(), chunks[1])
self.assertEqual(c.chunks[3].raw(), chunks[0])
def _new_content(self, stgpol, data, path="titi"):
old_content = self.content_factory.new(self.container_id, path,
len(data), stgpol)
old_content.upload(StringIO.StringIO(data))
return self.content_factory.get(self.container_id,
old_content.content_id)
def _test_change_policy(self, data_size, old_policy, new_policy):
if (old_policy == "RAIN" or new_policy == "RAIN") \
and len(self.conf['rawx']) < 8:
self.skipTest("RAIN: Need more than 8 rawx to run")
data = random_data(data_size)
obj_type = {
"SINGLE": DupContent,
"TWOCOPIES": DupContent,
"THREECOPIES": DupContent,
"RAIN": RainContent
}
old_content = self._new_content(old_policy, data)
self.assertEqual(type(old_content), obj_type[old_policy])
changed_content = self.content_factory.change_policy(
old_content.container_id, old_content.content_id, new_policy)
self.assertRaises(NotFound,
self.container_client.content_show,
self.account,
cid=old_content.container_id,
content=old_content.content_id)
new_content = self.content_factory.get(self.container_id,
changed_content.content_id)
self.assertEqual(type(new_content), obj_type[new_policy])
downloaded_data = "".join(new_content.download())
self.assertEqual(downloaded_data, data)
def test_change_content_0_byte_policy_single_to_rain(self):
self._test_change_policy(0, "SINGLE", "RAIN")
def test_change_content_0_byte_policy_rain_to_twocopies(self):
self._test_change_policy(0, "RAIN", "TWOCOPIES")
def test_change_content_1_byte_policy_single_to_rain(self):
self._test_change_policy(1, "SINGLE", "RAIN")
def test_change_content_chunksize_bytes_policy_twocopies_to_rain(self):
self._test_change_policy(self.chunk_size, "TWOCOPIES", "RAIN")
def test_change_content_2xchunksize_bytes_policy_threecopies_to_rain(self):
self._test_change_policy(self.chunk_size * 2, "THREECOPIES", "RAIN")
def test_change_content_1_byte_policy_rain_to_threecopies(self):
self._test_change_policy(1, "RAIN", "THREECOPIES")
def test_change_content_chunksize_bytes_policy_rain_to_twocopies(self):
self._test_change_policy(self.chunk_size, "RAIN", "TWOCOPIES")
def test_change_content_2xchunksize_bytes_policy_rain_to_single(self):
self._test_change_policy(self.chunk_size * 2, "RAIN", "SINGLE")
def test_change_content_0_byte_policy_twocopies_to_threecopies(self):
self._test_change_policy(0, "TWOCOPIES", "THREECOPIES")
def test_change_content_chunksize_bytes_policy_single_to_twocopies(self):
self._test_change_policy(self.chunk_size, "SINGLE", "TWOCOPIES")
def test_change_content_2xchunksize_bytes_policy_3copies_to_single(self):
self._test_change_policy(self.chunk_size * 2, "THREECOPIES", "SINGLE")
def test_change_content_with_same_policy(self):
data = random_data(10)
old_content = self._new_content("TWOCOPIES", data)
changed_content = self.content_factory.change_policy(
old_content.container_id, old_content.content_id, "TWOCOPIES")
self.assertEqual(old_content.content_id, changed_content.content_id)
def test_change_policy_unknown_content(self):
self.assertRaises(ContentNotFound, self.content_factory.change_policy,
self.container_id, "1234", "SINGLE")
def test_change_policy_unknown_storage_policy(self):
data = random_data(10)
old_content = self._new_content("TWOCOPIES", data)
self.assertRaises(ClientException, self.content_factory.change_policy,
self.container_id, old_content.content_id, "UnKnOwN")
def _test_move_chunk(self, policy):
data = random_data(self.chunk_size)
content = self._new_content(policy, data)
chunk_id = content.chunks.filter(metapos=0)[0].id
chunk_url = content.chunks.filter(metapos=0)[0].url
chunk_meta, chunk_stream = self.blob_client.chunk_get(chunk_url)
chunk_hash = md5_stream(chunk_stream)
new_chunk = content.move_chunk(chunk_id)
content_updated = self.content_factory.get(self.container_id,
content.content_id)
hosts = []
for c in content_updated.chunks.filter(metapos=0):
self.assertThat(hosts, Not(Contains(c.host)))
self.assertNotEquals(c.id, chunk_id)
hosts.append(c.host)
new_chunk_meta, new_chunk_stream = self.blob_client.chunk_get(
new_chunk["url"])
new_chunk_hash = md5_stream(new_chunk_stream)
self.assertEqual(new_chunk_hash, chunk_hash)
del chunk_meta["chunk_id"]
del new_chunk_meta["chunk_id"]
self.assertEqual(new_chunk_meta, chunk_meta)
def test_single_move_chunk(self):
self._test_move_chunk("SINGLE")
def test_twocopies_move_chunk(self):
self._test_move_chunk("TWOCOPIES")
def test_rain_move_chunk(self):
if len(self.conf['rawx']) < 9:
self.skipTest("Need more than 8 rawx")
self._test_move_chunk("RAIN")
def test_move_chunk_not_in_content(self):
data = random_data(self.chunk_size)
content = self._new_content("TWOCOPIES", data)
with ExpectedException(OrphanChunk):
content.move_chunk("1234")
def test_strange_paths(self):
strange_paths = [
"Annual report.txt",
"foo+bar=foobar.txt",
"100%_bug_free.c",
"forward/slash/allowed",
"I\\put\\backslashes\\and$dollar$signs$in$file$names",
"Je suis tombé sur la tête, mais ça va bien.",
"%s%f%u%d%%",
"carriage\rreturn",
"line\nfeed",
"ta\tbu\tla\ttion",
"controlchars",
]
answers = dict()
for cname in strange_paths:
content = self._new_content("SINGLE", "nobody cares", cname)
answers[cname] = content
listing = self.container_client.container_list(self.account,
self.container_name)
obj_set = {
k["name"].encode("utf8", "ignore")
for k in listing["objects"]
}
try:
# Ensure the saved path is the one we gave the object
for cname in answers:
self.assertEqual(cname, answers[cname].path)
# Ensure all objects appear in listing
for cname in strange_paths:
self.assertIn(cname, obj_set)
finally:
# Cleanup
for cname in answers:
try:
content.delete()
except:
pass
class StorageTiererWorker(object):
def __init__(self, conf, logger):
self.conf = conf
self.logger = logger
self.account = conf[CONF_ACCOUNT]
self.container_client = ContainerClient(self.conf)
self.account_client = AccountClient(self.conf)
self.content_factory = ContentFactory(self.conf)
self.passes = 0
self.errors = 0
self.last_reported = 0
self.contents_run_time = 0
self.total_contents_processed = 0
self.report_interval = int_value(conf.get('report_interval'), 3600)
self.max_contents_per_second = int_value(
conf.get('contents_per_second'), 30)
self.container_fetch_limit = int_value(
conf.get('container_fetch_limit'), 100)
self.content_fetch_limit = int_value(conf.get('content_fetch_limit'),
100)
self.outdated_threshold = int_value(conf.get(CONF_OUTDATED_THRESHOLD),
9999999999)
self.new_policy = conf.get(CONF_NEW_POLICY)
def _list_containers(self):
container = None
while True:
resp = self.account_client.containers_list(
self.account,
marker=container,
limit=self.container_fetch_limit)
if len(resp["listing"]) == 0:
break
for container, _, _, _ in resp["listing"]:
yield container
def _list_contents(self):
for container in self._list_containers():
marker = None
while True:
try:
resp = self.container_client.container_list(
acct=self.account,
ref=container,
limit=self.content_fetch_limit,
marker=marker)
except NotFound:
self.logger.warn("Container %s in account "
"but not found" % container)
break
if len(resp["objects"]) == 0:
break
for obj in resp["objects"]:
marker = obj["name"]
if obj["mtime"] > time.time() - self.outdated_threshold:
continue
if obj["policy"] == self.new_policy:
continue
container_id = cid_from_name(self.account, container)
yield (container_id, obj["content"])
def run(self):
start_time = report_time = time.time()
total_errors = 0
for (container_id, content_id) in self._list_contents():
self.safe_change_policy(container_id, content_id)
self.contents_run_time = ratelimit(self.contents_run_time,
self.max_contents_per_second)
self.total_contents_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 '
'%(content_rate).2f ' % {
'elapsed': elapsed,
'errors': total_errors + self.errors,
'content_rate': self.total_contents_processed / elapsed
})
def safe_change_policy(self, container_id, content_id):
try:
self.change_policy(container_id, content_id)
except Exception:
self.errors += 1
self.logger.exception(
"ERROR while changing policy for content "
"%s/%s", (container_id, content_id))
self.passes += 1
def change_policy(self, container_id, content_id):
self.logger.info("Changing policy for content %s/%s" %
(container_id, content_id))
self.content_factory.change_policy(container_id, content_id,
self.new_policy)
class StorageTiererWorker(object):
def __init__(self, conf, logger):
self.conf = conf
self.logger = logger
self.account = conf[CONF_ACCOUNT]
self.container_client = ContainerClient(self.conf)
self.account_client = AccountClient(self.conf)
self.content_factory = ContentFactory(self.conf)
self.passes = 0
self.errors = 0
self.last_reported = 0
self.contents_run_time = 0
self.total_contents_processed = 0
self.report_interval = int_value(
conf.get('report_interval'), 3600)
self.max_contents_per_second = int_value(
conf.get('contents_per_second'), 30)
self.container_fetch_limit = int_value(
conf.get('container_fetch_limit'), 100)
self.content_fetch_limit = int_value(
conf.get('content_fetch_limit'), 100)
self.outdated_threshold = int_value(
conf.get(CONF_OUTDATED_THRESHOLD), 9999999999)
self.new_policy = conf.get(CONF_NEW_POLICY)
def _list_containers(self):
container = None
while True:
resp = self.account_client.containers_list(
self.account, marker=container,
limit=self.container_fetch_limit)
if len(resp["listing"]) == 0:
break
for container, _, _, _ in resp["listing"]:
yield container
def _list_contents(self):
for container in self._list_containers():
marker = None
while True:
try:
resp = self.container_client.container_list(
acct=self.account, ref=container,
limit=self.content_fetch_limit, marker=marker)
except NotFound:
self.logger.warn("Container %s in account "
"but not found" % container)
break
if len(resp["objects"]) == 0:
break
for obj in resp["objects"]:
marker = obj["name"]
if obj["mtime"] > time.time() - self.outdated_threshold:
continue
if obj["policy"] == self.new_policy:
continue
container_id = cid_from_name(self.account, container)
yield (container_id, obj["content"])
def run(self):
start_time = report_time = time.time()
total_errors = 0
for (container_id, content_id) in self._list_contents():
self.safe_change_policy(container_id, content_id)
self.contents_run_time = ratelimit(
self.contents_run_time,
self.max_contents_per_second
)
self.total_contents_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 '
'%(content_rate).2f ' % {
'elapsed': elapsed,
'errors': total_errors + self.errors,
'content_rate': self.total_contents_processed / elapsed
}
)
def safe_change_policy(self, container_id, content_id):
try:
self.change_policy(container_id, content_id)
except Exception:
self.errors += 1
self.logger.exception("ERROR while changing policy for content "
"%s/%s", (container_id, content_id))
self.passes += 1
def change_policy(self, container_id, content_id):
self.logger.info("Changing policy for content %s/%s"
% (container_id, content_id))
self.content_factory.change_policy(
container_id, content_id, self.new_policy)
