[
https://issues.apache.org/jira/browse/IGNITE-2335?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vladimir Ozerov updated IGNITE-2335:
------------------------------------
Issue Type: Task (was: Sub-task)
Parent: (was: IGNITE-2232)
> Review ReentrantReadWriteLock usages.
> -------------------------------------
>
> Key: IGNITE-2335
> URL: https://issues.apache.org/jira/browse/IGNITE-2335
> Project: Ignite
> Issue Type: Task
> Components: general
> Affects Versions: ignite-1.4
> Reporter: Vladimir Ozerov
> Assignee: Vladimir Ozerov
> Fix For: 1.6
>
>
> We have about 40-50 ReentrantReadWriteLock lock usages all over the project.
> This class is known to perform poorly under high readers contention. One
> typical example - IGNITE-2333.
> We need to do the following:
> 1) If lock has "busy" semantics (read.tryLock + read.unlock + write.lock),
> then change it to normal busy lock.
> 2) If lock appears to be on a hot path, it should be either replaced with
> correctly stripped/padded counterpart, or with simple ReentrantLock if
> protected operations are very small (e.g. single put into map).
> 3) Do not change anything otherwise.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
