[
https://issues.apache.org/jira/browse/LUCENE-8962?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17028727#comment-17028727
]
Michael Froh commented on LUCENE-8962:
--------------------------------------
bq. Yeah I think you are right! That would be a nice simplification. Probably
this can just be folded into the existing MergePolicy API as a different
MergeTrigger. Though then I wonder why e.g. forceMerge or expungeDeletes are
not also simply different triggers ... Michael Froh what do you think?
As I was first writing this, I added a {{MergeTrigger.COMMIT}} value and used
that, rather than adding a dedicated method.
Then I realized that any time I've ever written a custom implementation of
{{MergePolicy.findMerges()}}, I've ignored the {{MergeTrigger}} value, because
I didn't really care what triggered the merge -- I just wanted to define the
{{MergeSpecification}}. Even {{TieredMergePolicy.findMerges()}}} doesn't look
at the {{MergeTrigger}} parameter.
If I had made {{IndexWriter}} call {{findMerges}} with a
{{MergeTrigger.COMMIT}} trigger, anyone with a similar {{MergePolicy}} would
have probably ended up running (and blocking on) some pretty expensive merges
on commit. The best way I could think of to be backwards compatible with the
"old" behavior by default was to add a no-op method to the base class.
Looking through the history, it looks like {{forceMerge}} and
{{expungeDeletes}} predate {{MergeTrigger}}, so that could explain them.
I really like the idea of controlling this with a {{MergeTrigger}}, but I'm
concerned about breaking existing {{MergePolicy}} implementations that ignore
the {{MergeTrigger}} (which I suspect may be most of them).
> Can we merge small segments during refresh, for faster searching?
> -----------------------------------------------------------------
>
> Key: LUCENE-8962
> URL: https://issues.apache.org/jira/browse/LUCENE-8962
> Project: Lucene - Core
> Issue Type: Improvement
> Components: core/index
> Reporter: Michael McCandless
> Priority: Major
> Attachments: LUCENE-8962_demo.png
>
> Time Spent: 4h 40m
> Remaining Estimate: 0h
>
> With near-real-time search we ask {{IndexWriter}} to write all in-memory
> segments to disk and open an {{IndexReader}} to search them, and this is
> typically a quick operation.
> However, when you use many threads for concurrent indexing, {{IndexWriter}}
> will accumulate write many small segments during {{refresh}} and this then
> adds search-time cost as searching must visit all of these tiny segments.
> The merge policy would normally quickly coalesce these small segments if
> given a little time ... so, could we somehow improve {{IndexWriter'}}s
> refresh to optionally kick off merge policy to merge segments below some
> threshold before opening the near-real-time reader? It'd be a bit tricky
> because while we are waiting for merges, indexing may continue, and new
> segments may be flushed, but those new segments shouldn't be included in the
> point-in-time segments returned by refresh ...
> One could almost do this on top of Lucene today, with a custom merge policy,
> and some hackity logic to have the merge policy target small segments just
> written by refresh, but it's tricky to then open a near-real-time reader,
> excluding newly flushed but including newly merged segments since the refresh
> originally finished ...
> I'm not yet sure how best to solve this, so I wanted to open an issue for
> discussion!
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
