jackrabbit-oak-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tommaso Teofili (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (OAK-6710) CommitMitigated merge policy should not reduce performance significantly
Date Wed, 27 Sep 2017 07:46:00 GMT

    [ https://issues.apache.org/jira/browse/OAK-6710?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16182145#comment-16182145

Tommaso Teofili commented on OAK-6710:

thanks [~catholicon] for reporting the issue, I think the fixed values for max commit rate
MB and Docs [1,2] have to be adjusted.
Ideally we can come out with an algorithm which can adapt those numbers to an average commit
rate depending on the system load (over time), alternatively we can simply adjust them and
at first make them larger.
The performance problem is mostly due to not enough merges being done and hence having too
many segments (slowing down searches and indexing tasks).

[1] : https://github.com/apache/jackrabbit-oak/blob/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/writer/CommitMitigatingTieredMergePolicy.java#L50
[2] : https://github.com/apache/jackrabbit-oak/blob/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/writer/CommitMitigatingTieredMergePolicy.java#L51

> CommitMitigated merge policy should not reduce performance significantly
> ------------------------------------------------------------------------
>                 Key: OAK-6710
>                 URL: https://issues.apache.org/jira/browse/OAK-6710
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: lucene
>    Affects Versions: 1.7.7
>            Reporter: Vikas Saurabh
>            Assignee: Tommaso Teofili
>              Labels: performance, scalability
>             Fix For: 1.8
> While running performance tests (internally) using latest unstable releases having {{CommitMitigatedTieredMergePolicy}}
we observed significant drop in performance (OAK-6704).
> The policy, although bad for performance, showed dramatic drop in churn created in data
store size. So, clearly, OAK-5192 did what it intended to do.
> Opening this issue to factor in drop in performance and then set the default back to

This message was sent by Atlassian JIRA

View raw message