Hi,
just a short follow up on this topic:
The Lucene developers found the source of the problem and maybe will fix this
issue:
https://github.com/apache/lucene/issues/14180
Best regards
Andreas
Hi David,
Thank you for giving me to a good starting point for a test setup.
I have written a Unit Test that should match our usecase but it cannot yet
reproduce the error.
My guess is that the error may also depend on the status of the segment files.
I will give it another try in the coming
Hi Andreas,
The test org.apache.solr.CursorPagingTest is a single-core test for
cursorMark. Alternatively, perhaps you might want to modify a KNN query
test like org.apache.solr.search.neural.KnnQParserTest.
BTW, the correct capitalization of "Solr" is as I typed it, not as you are
typing it.
~
Hi,
I was aware of the fact, that the tie breaker "sort by id" would only effect
the search after the results of the KNN algorithm come in (which may be
non-deterministic), but I guessed it to be a best practice to add it to all
queries since e.g. searching with a cursor mark requires the tie
We have seen this, and when we tested against Solr 9.8 (currently being
released), the error went away.
It turned out to be a weird thing about the search executor, but we
couldn't necessarily narrow down why it happened.
Anyway, please test out with the new Solr 9.8.0 release when it is
available
I'm not sure the KNN search supports multiple sort conditions.
I should do some deep dive in the Lucene code but I don't have time in the
foreseeable short future.
I can imagine that anyway it would only support 're-ranking' the retrieved
topK by the additional score condition, this can not affect
Hi,
I want to inform you about a behavior change in SolR 9.6 (Lucene 9.10) vs. SolR
9.7 (Lucene 9.11) for vector searches.
We heavily rely on vector searches for embeddings in combination with filter
queries on the parent documents.
Our queries in general looked like this:
select?q={ knn f=vector