You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ignite.apache.org by "Vladimir Ozerov (JIRA)" <ji...@apache.org> on 2015/02/10 09:38:36 UTC
[jira] [Commented] (IGNITE-138) Ensure that utility cache is really
executed in utility pool
[ https://issues.apache.org/jira/browse/IGNITE-138?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14313830#comment-14313830 ]
Vladimir Ozerov commented on IGNITE-138:
----------------------------------------
Suspicious places:
1) IgniteTxHandler.processCheckPreparedTxRequest;
2) IgniteTxHandler.sendCheckCommittedResponse.
> Ensure that utility cache is really executed in utility pool
> ------------------------------------------------------------
>
> Key: IGNITE-138
> URL: https://issues.apache.org/jira/browse/IGNITE-138
> Project: Ignite
> Issue Type: Bug
> Components: cache
> Affects Versions: sprint-1
> Reporter: Vladimir Ozerov
> Assignee: Alexey Goncharuk
> Priority: Critical
> Fix For: sprint-1
>
>
> See GridCacheIoManager:
> 1) sendOrderedMessaage() always use system pool.
> 2) safeSend() always use system pool.
> 3) send(ClusterNode, GridCacheMessage) use system pool implicitly.
> With such design there is a great risk that system cache either already process some of it's messages in system pool instead of utility pool, or will do it at some point in future.
> Possible solutions:
> 1) Move GridCacheIoManager from shared context to cache context.
> 2) Force all public methods to specify IO policy explicitly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)