[
https://issues.apache.org/jira/browse/LUCENE-10125?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17420843#comment-17420843
]
Uwe Schindler commented on LUCENE-10125:
----------------------------------------
BTW, the original code in BufferedOutputStream also uses same logic: It just
checks the remaining buffer size and calls flush() if it is not large enough.
Only difference is that for writeBytes() it splits the byte array writes, we
can't do. I think here it is perfectly fine to do an upfront check that buffer
size must be >=8. I checked the code:
- We only use 8192 as buffer size in FSIndexOutput, as this is also some buffer
allocation boundary in the JDK
- We have one test using 8192, too
Nothing else calls it. Any consumer outside lucene needs to adapt code for 9.0
anyways, so I see no risk. I updated Javadocs to require a limit of 8 bytes for
the buffer size.
> 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
> Attachments: LUCENE-10125_hack.patch
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> 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]
