def del_vectors(collection_name, ids):
milvus = Milvus()
try:
milvus.connect(MILVUS_ADDR, MILVUS_PORT)
milvus.delete_by_id(collection_name=collection_name, id_array=ids)
except Exception as e:
raise MilvusError("There has some error when delete vectors", e)
for seg in par.segments_stat:
print(
"\t\tsegment name: {}, vector count: {}, index: {}, storage size {:.3f} MB"
.format(seg.segment_name, seg.count, seg.index_name,
seg.data_size / 1024 / 1024))
# obtain vector ids from segment, then
# get vector by specifying vector id
segment0 = info.partitions_stat[0].segments_stat[0]
status, ids = client.get_vector_ids(collection_name, segment0.segment_name)
if not status.OK():
print("Cannot obtain vector ids from segment {}. exiting ....".format(
segment0.segment_name))
sys.exit(1)
# obtain first vector
status, vector = client.get_vector_by_id(collection_name, ids[0])
if not status.OK():
print("Cannot obtain vector. exiting ....")
sys.exit(1)
# delete top 10 vectors
status = client.delete_by_id(collection_name, ids[:10])
if status.OK():
print("Delete top 10 vectors successfully")
else:
print("Error occurred when try to delete top 10 vectors. Reason: ",
status.message)
client.drop_collection(collection_name)
class MilvusClient(object):
def __init__(self, collection_name=None, ip=None, port=None, timeout=60):
self._collection_name = collection_name
try:
i = 1
start_time = time.time()
if not ip:
self._milvus = Milvus(host=SERVER_HOST_DEFAULT,
port=SERVER_PORT_DEFAULT)
else:
# retry connect for remote server
while time.time() < start_time + timeout:
try:
self._milvus = Milvus(host=ip, port=port)
if self._milvus.server_status():
logger.debug(
"Try connect times: %d, %s" %
(i, round(time.time() - start_time, 2)))
break
except Exception as e:
logger.debug("Milvus connect failed")
i = i + 1
except Exception as e:
raise e
def __str__(self):
return 'Milvus collection %s' % self._collection_name
def check_status(self, status):
if not status.OK():
logger.error(status.message)
raise Exception("Status not ok")
def check_result_ids(self, result):
for index, item in enumerate(result):
if item[0].distance >= epsilon:
logger.error(index)
logger.error(item[0].distance)
raise Exception("Distance wrong")
def create_collection(self, collection_name, dimension, index_file_size,
metric_type):
if not self._collection_name:
self._collection_name = collection_name
if metric_type == "l2":
metric_type = MetricType.L2
elif metric_type == "ip":
metric_type = MetricType.IP
elif metric_type == "jaccard":
metric_type = MetricType.JACCARD
elif metric_type == "hamming":
metric_type = MetricType.HAMMING
elif metric_type == "sub":
metric_type = MetricType.SUBSTRUCTURE
elif metric_type == "super":
metric_type = MetricType.SUPERSTRUCTURE
else:
logger.error("Not supported metric_type: %s" % metric_type)
create_param = {
'collection_name': collection_name,
'dimension': dimension,
'index_file_size': index_file_size,
"metric_type": metric_type
}
status = self._milvus.create_collection(create_param)
self.check_status(status)
@time_wrapper
def insert(self, X, ids=None):
status, result = self._milvus.add_vectors(self._collection_name, X,
ids)
self.check_status(status)
return status, result
@time_wrapper
def delete_vectors(self, ids):
status = self._milvus.delete_by_id(self._collection_name, ids)
self.check_status(status)
@time_wrapper
def flush(self):
status = self._milvus.flush([self._collection_name])
self.check_status(status)
@time_wrapper
def compact(self):
status = self._milvus.compact(self._collection_name)
self.check_status(status)
@time_wrapper
def create_index(self, index_type, index_param=None):
index_type = INDEX_MAP[index_type]
logger.info(
"Building index start, collection_name: %s, index_type: %s" %
(self._collection_name, index_type))
if index_param:
logger.info(index_param)
status = self._milvus.create_index(self._collection_name, index_type,
index_param)
self.check_status(status)
def describe_index(self):
status, result = self._milvus.describe_index(self._collection_name)
self.check_status(status)
index_type = None
for k, v in INDEX_MAP.items():
if result._index_type == v:
index_type = k
break
return {"index_type": index_type, "index_param": result._params}
def drop_index(self):
logger.info("Drop index: %s" % self._collection_name)
return self._milvus.drop_index(self._collection_name)
@time_wrapper
def query(self, X, top_k, search_param=None):
status, result = self._milvus.search_vectors(self._collection_name,
top_k,
query_records=X,
params=search_param)
self.check_status(status)
return result
@time_wrapper
def query_ids(self, top_k, ids, search_param=None):
status, result = self._milvus.search_by_ids(self._collection_name,
ids,
top_k,
params=search_param)
self.check_result_ids(result)
return result
def count(self):
return self._milvus.count_collection(self._collection_name)[1]
def delete(self, timeout=120):
timeout = int(timeout)
logger.info("Start delete collection: %s" % self._collection_name)
self._milvus.drop_collection(self._collection_name)
i = 0
while i < timeout:
if self.count():
time.sleep(1)
i = i + 1
continue
else:
break
if i >= timeout:
logger.error("Delete collection timeout")
def describe(self):
return self._milvus.describe_collection(self._collection_name)
def show_collections(self):
return self._milvus.show_collections()
def exists_collection(self, collection_name=None):
if collection_name is None:
collection_name = self._collection_name
status, res = self._milvus.has_collection(collection_name)
# self.check_status(status)
return res
@time_wrapper
def preload_collection(self):
status = self._milvus.preload_collection(self._collection_name,
timeout=3000)
self.check_status(status)
return status
def get_server_version(self):
status, res = self._milvus.server_version()
return res
def get_server_mode(self):
return self.cmd("mode")
def get_server_commit(self):
return self.cmd("build_commit_id")
def get_server_config(self):
return json.loads(self.cmd("get_config *"))
def get_mem_info(self):
result = json.loads(self.cmd("get_system_info"))
result_human = {
# unit: Gb
"memory_used":
round(int(result["memory_used"]) / (1024 * 1024 * 1024), 2)
}
return result_human
def cmd(self, command):
status, res = self._milvus._cmd(command)
logger.info("Server command: %s, result: %s" % (command, res))
self.check_status(status)
return res