ode-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Rick.Todo (JIRA)" <j...@apache.org>
Subject [jira] Updated: (ODE-681) Implement immediate transaction retries in addition to the presistent retries
Date Wed, 15 Dec 2010 09:28:02 GMT

     [ https://issues.apache.org/jira/browse/ODE-681?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Rick.Todo updated ODE-681:
--------------------------

    Attachment: retryForDeadlock.patch

with commited code, job did not be retried when the failure was due to deadlocks.
if an exception occurs at "txm.commit()", "execTransaction" retruned with no retry.
see attached patch.

> Implement immediate transaction retries in addition to the presistent retries
> -----------------------------------------------------------------------------
>
>                 Key: ODE-681
>                 URL: https://issues.apache.org/jira/browse/ODE-681
>             Project: ODE
>          Issue Type: Improvement
>          Components: BPEL Runtime
>            Reporter: Sean Ahn
>            Assignee: Sean Ahn
>             Fix For: 1.3.4
>
>         Attachments: retryForDeadlock.patch
>
>
> Currently, the ode engine re-schedules a job into the ODE_JOB table when the job fails.
The job will be picked up later by the job scheduler on the same node. For 2 reasons, we want
to pre-pend an immediate transaction retry logic before the persistent retries.
> 1. The current way of scheduling a job involves deleting of the job and inserting/selecting
of a new job for a retry. If the system is under heavy load, and if the failure was due to
deadlocks from database overload, you do not want to put on even more load.
> 2. The interval between retries could be pretty long. Cases like database deadlocks can
be resolved by re-trying the same transaction in a relatively short time.
> Repeat the same transaction by configurable number of tries with a configuration interval.
If all tries still fail, then schedule the job for persistent retries.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message