[
https://issues.apache.org/jira/browse/LUCENE-10125?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17420722#comment-17420722
]
Adrien Grand commented on LUCENE-10125:
---------------------------------------
I tried to do a git bisect but I haven't been successful at finding the commit
that introduced the regression by looking at indexing rates, because they are
not reproducible enough.
So I started looking at the InfoStream and between these two commits I'm seeing:
- 15% **faster** flushes for points, likely thanks to LUCENE-9932.
- 15% slower merges of postings (likely flushes too) but this is unlikely the
root cause since the benchmark spends only 10% of the time it spends on points
or doc values on postings. This is likely due to LUCENE-9850.
- 10% slower merges of stored fields but this is unlikely the root cause since
the benchmark spends much more time merging points and doc values.
- 15% slower merges of points - likely the root cause of the slowdown.
I'll try to bisect based on the merge times for points now.
> Investigate indexing throughput regression on NYC Taxis between 2021-04-12
> and 2021-05-24
> -----------------------------------------------------------------------------------------
>
> Key: LUCENE-10125
> URL: https://issues.apache.org/jira/browse/LUCENE-10125
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Priority: Minor
>
> There's been a significant drop in indexing throughput between 2021-04-12 and
> 2021-05-24 on the NYC Taxis benchmark. Unfortunately several suspects have
> been merged during that period of time so we might need to git bisect to
> figure out which one is responsible for the regression. Interestingly the
> sorted index looks less affected than the non-sorted indexes.
> https://home.apache.org/~mikemccand/lucenebench/sparseResults.html#index_throughput
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
