You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Stanilovsky Evgeny (Jira)" <ji...@apache.org> on 2020/12/01 10:54:00 UTC

[jira] [Commented] (IGNITE-13775) U.ReentrantReadWriteLockTracer improper realization.

    [ https://issues.apache.org/jira/browse/IGNITE-13775?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17241442#comment-17241442 ] 

Stanilovsky Evgeny commented on IGNITE-13775:
---------------------------------------------

I have benchmarked current approach, 1 client, 3 servers, persistent=on, backups=1, left - ignite-2.9 right current PR, looks like all ok here. 
!image-2020-12-01-13-51-39-048.png!

> U.ReentrantReadWriteLockTracer improper realization.
> ----------------------------------------------------
>
>                 Key: IGNITE-13775
>                 URL: https://issues.apache.org/jira/browse/IGNITE-13775
>             Project: Ignite
>          Issue Type: Improvement
>          Components: general
>    Affects Versions: 2.9
>            Reporter: Stanilovsky Evgeny
>            Assignee: Stanilovsky Evgeny
>            Priority: Major
>         Attachments: image-2020-12-01-13-51-39-048.png
>
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> ReentrantReadWriteLockTracer accepts ReentrantReadWriteLock as a delegate and stores delegates for readLock and writeLock. But ReentrantReadWriteLock#isWriteLockedByCurrentThread uses sync object to evaluate the result instead of writeLock, and ReentrantReadWriteLockTracer has it's own sync object.
> As a result, if ReentrantReadWriteLockTracer is used to create checkpoint lock (when IGNITE_PDS_LOG_CP_READ_LOCK_HOLDERS=true), GridCacheDatabaseSharedManager#checkpointLockIsHeldByThread doesn't work correctly: it returns false when checkpoint lock is acquired.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)