Open binbinlv opened 1 month ago
The failure to reset the flushedRowNum
of the writerBuffer resulted in subsequent segments in this buffer continually meeting the pack conditions, leading to the creation of numerous segments. The datanode couldn't load the BloomFilters for so many segments, causing an OOM error.
The OOM issue still exists when major compaction on 50M (768) data: image: v2.4.6-nightly-176034d1-872
Is there an existing issue for this?
Environment
Current Behavior
The datanode is continously restarted for OOM after major compaction for 50M (768) dataset
Expected Behavior
No crash
Steps To Reproduce
Milvus Log
No response
Anything else?
No response