You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@ignite.apache.org by "Roman Kondakov (JIRA)" <ji...@apache.org> on 2019/02/19 14:56:00 UTC

[jira] [Updated] (IGNITE-11349) MVCC: Deadlock in query pool when executing DML over caches with queryParallism > 1

     [ https://issues.apache.org/jira/browse/IGNITE-11349?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Roman Kondakov updated IGNITE-11349:
------------------------------------
    Description: 
Reproducer: {{CacheMvccSelectForUpdateQueryBasicTest}}

There is possible deadlock in {{QUERY_POOL}} when running massive SQL updates using queries without reducer when {{cache.queryParallelism > 1}}. These updates fire two step queries on each data node and both map and reduce phases are executed in the same {{QUERY_POOL}} which can lead to deadlock because reduce phase has blocking operation {{GridReduceQueryExecutor#awaitAllReplies}}.

> MVCC: Deadlock in query pool when executing DML over caches with queryParallism > 1
> -----------------------------------------------------------------------------------
>
>                 Key: IGNITE-11349
>                 URL: https://issues.apache.org/jira/browse/IGNITE-11349
>             Project: Ignite
>          Issue Type: Bug
>          Components: mvcc, sql
>            Reporter: Roman Kondakov
>            Priority: Major
>             Fix For: 2.8
>
>
> Reproducer: {{CacheMvccSelectForUpdateQueryBasicTest}}
> There is possible deadlock in {{QUERY_POOL}} when running massive SQL updates using queries without reducer when {{cache.queryParallelism > 1}}. These updates fire two step queries on each data node and both map and reduce phases are executed in the same {{QUERY_POOL}} which can lead to deadlock because reduce phase has blocking operation {{GridReduceQueryExecutor#awaitAllReplies}}.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)