为什么同样结构的两个集合，一个数据量是500w，一个是100w，但是500w的data_size却小于100w的？

[yuguosen]

版本：0.11

API：res=client.get_collection_stats("xxxx_500w")
"data_size": 519732797,
	"partition_count": 2,
	"partitions": [{
		"data_size": 519732797,
		"id": 44,
		"row_count": 5000016,
		"segment_count": 1101,

API：res=client.get_collection_stats("xxxx_100w")
"data_size": 533510720,
	"partition_count": 2,
	"partitions": [{
		"data_size": 533510720,
		"id": 36,
		"row_count": 1000032,
		"segment_count": 221,

我接着尝试遍历这个dict，然后将每个segment中的data_size相加，结果是500w的data_size之和为2667405656，而100w的data_size
之和为533510720这是怎么回事呢？

[yhmo]

Segment data size including: internal ID array, bloom-filter file, vector array, index file. If the collection data size is not equal to sum of segments data size, it could be a bug of this api.