You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ode.apache.org by "Sean Ahn (JIRA)" <ji...@apache.org> on 2009/06/30 02:30:47 UTC
[jira] Created: (ODE-632) Reduce db deadlocks by getting rid of
delete statements that use index
Reduce db deadlocks by getting rid of delete statements that use index
----------------------------------------------------------------------
Key: ODE-632
URL: https://issues.apache.org/jira/browse/ODE-632
Project: ODE
Issue Type: Improvement
Components: BPEL Runtime
Reporter: Sean Ahn
Assignee: Sean Ahn
Under real work load, the on-the-fly instance cleanup functionality dead-locks a lot due to the excessive sub-select statements that depends on index. Some dbms like mysql locks the whole index that are used in the nested select statements.
Split the delete queries into object id selection and deletion on the ids. Since we already batch up instance cleanup by number of instances(default is 10 instances), this is not a problem in scalability with big dataset.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (ODE-632) Reduce db deadlocks by getting rid of
delete statements that use index
Posted by "Sean Ahn (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/ODE-632?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sean Ahn resolved ODE-632.
--------------------------
Resolution: Fixed
Most of the data clean up hql queries have been changed to collect ids for the objects. The DAO classes use the id collections in deleting the objects. Thus, indexes are not used directly with the 'delete' sql statements.
> Reduce db deadlocks by getting rid of delete statements that use index
> ----------------------------------------------------------------------
>
> Key: ODE-632
> URL: https://issues.apache.org/jira/browse/ODE-632
> Project: ODE
> Issue Type: Improvement
> Components: BPEL Runtime
> Reporter: Sean Ahn
> Assignee: Sean Ahn
>
> Under real work load, the on-the-fly instance cleanup functionality dead-locks a lot due to the excessive sub-select statements that depends on index. Some dbms like mysql locks the whole index that are used in the nested select statements.
> Split the delete queries into object id selection and deletion on the ids. Since we already batch up instance cleanup by number of instances(default is 10 instances), this is not a problem in scalability with big dataset.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.