Closed wayblink closed 1 month ago
Clustering compaction is much bigger task than normal compaction. And milvus doesn't have a strict memory control. Hence, we need to control the concurrency of clustering compactions.
No response
what is the memory estimation of a clustering compaction? is it related to segment number * bin log size?
what is the expected memory usage and concurrency?
Is there an existing issue for this?
What would you like to be added?
34495
Clustering compaction is much bigger task than normal compaction. And milvus doesn't have a strict memory control. Hence, we need to control the concurrency of clustering compactions.
Why is this needed?
No response
Anything else?
No response