weizijun opened a new issue, #14330: URL: https://github.com/apache/lucene/issues/14330
### Description I found a serious bad case. When I write all the same vector docs, It will cause flush blocked. The cost comes from the `connectComponents` process. When all vectors are the same, there will be a lot of notFullyConnected nodes. It's easy to reproduce. Just us Elasticsearch to write 3 dims docs, We will see that the write threads are blocked. Here are some logs: ``` [[vector_test][0]: Thread-250] HNSW: connectComponents failed on level 1 [[vector_test][0]: Thread-250] HNSW: connectComponents failed on level 2 [[vector_test][0]: Thread-250] HNSW: connectComponents 2622896 ms [[vector_test][0]: Thread-93] HNSW: connectComponents failed on level 1 [[vector_test][0]: Thread-93] HNSW: connectComponents failed on level 2 [[vector_test][0]: Thread-93] HNSW: connectComponents failed on level 3 [[vector_test][0]: Thread-93] HNSW: connectComponents 4436718 ms [[vector_test][0]: Thread-863] HNSW: connectComponents failed on level 1 [[vector_test][0]: Thread-863] HNSW: connectComponents failed on level 2 [[vector_test][0]: Thread-863] HNSW: connectComponents 4025765 ms [write][T#24] HNSW: connectComponents failed on level 1 [write][T#24] HNSW: connectComponents failed on level 2 [write][T#24] HNSW: connectComponents 3632994 ms [write][T#14] HNSW: connectComponents failed on level 1 [write][T#14] HNSW: connectComponents failed on level 2 [write][T#14] HNSW: connectComponents 4111515 ms [write][T#5] HNSW: connectComponents failed on level 1 [write][T#5] HNSW: connectComponents failed on level 2 [write][T#5] HNSW: connectComponents 4186353 ms [refresh][T#2] HNSW: connectComponents failed on level 1 [refresh][T#2] HNSW: connectComponents failed on level 2 [refresh][T#2] HNSW: connectComponents 4224230 ms [flush][T#6] HNSW: connectComponents failed on level 1 [flush][T#6] HNSW: connectComponents failed on level 2 [flush][T#6] HNSW: connectComponents 3971211 ms ``` ### Version and environment details Lucene 9.12.0/10.0.0 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org