[
https://issues.apache.org/jira/browse/SOLR-13898?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16968676#comment-16968676
]
Ben Manes commented on SOLR-13898:
----------------------------------
FYI,
{{ConcurrentHashMap#computeIfAbsent}} is pessimistic and locks more
aggressively than you might expect. In Java 8 it will always lock, even if
present. In Java 9 it will perform a partial prescreen - if the first entry in
the hashbin is the desired entry then return, else lock and scan the bucket.
Caffeine always does a full prescreen, which means a present entry is always
retrieved lock-free. When
[talking|http://jsr166-concurrency.10961.n7.nabble.com/Re-ConcurrentHashMap-computeIfAbsent-td11687.html]
to Doug Lea in Java 8 time-frame, he adopted a pessimistic stance due to
biased locking and safepoints at 32+ cores favors that approach.
You may want to offer a similar optimization in your non-Caffeine
implementations to perform a `get(key)` and fallback to {{computeIfAbsent}}. I
haven't benchmarked at Java 9, but you can see the results in Java 8 were
[dramatic|https://github.com/ben-manes/caffeine/wiki/Benchmarks#compute].
> Non-atomic use of SolrCache get / put
> -------------------------------------
>
> Key: SOLR-13898
> URL: https://issues.apache.org/jira/browse/SOLR-13898
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Affects Versions: 8.3
> Reporter: Andrzej Bialecki
> Assignee: Andrzej Bialecki
> Priority: Major
> Fix For: 8.4
>
> Attachments: SOLR-13898.patch
>
>
> As pointed out by [~ben.manes] in SOLR-13817 Solr code base in many key
> places uses a similar pattern of non-atomic get / put calls to SolrCache-s.
> In multi-threaded environment this leads to cache misses and additional
> unnecessary computations when competing threads both discover a missing
> value, non-atomically compute it and update the cache.
> Some of these places are known performance bottlenecks where efficient
> caching is especially important, such as {{SolrIndexSearcher}},
> {{SolrDocumentFetcher}}, {{UninvertedField}} and join queries .
> I propose to add {{SolrCache.computeIfAbsent(key, mappingFunction)}} that
> will atomically retrieve existing values or compute and update the cache.
> This will require also changing how the {{SolrCache.get(...)}} is used in
> many components.
--
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
