[ 
https://issues.apache.org/jira/browse/UIMA-3828?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Lou DeGenaro resolved UIMA-3828.
--------------------------------
       Resolution: Fixed
    Fix Version/s:     (was: future-DUCC)
                   2.0.0-Ducc

DUCC 2.0 has changed the handshaking between Job Driver (JD) and Job Process 
(JP) such that work items can no longer become lost.

The new protocol is such that the JP requests work items directly from the JD, 
presenting location right from the start.  In contrast, the 1.x design had the 
JD sending work items to an AMQ queue, not always knowing which JP picked up 
the work item from the queue.

> DUCC job driver (JD) should auto-retry lost work items - improvement
> --------------------------------------------------------------------
>
>                 Key: UIMA-3828
>                 URL: https://issues.apache.org/jira/browse/UIMA-3828
>             Project: UIMA
>          Issue Type: Improvement
>          Components: DUCC
>            Reporter: Lou DeGenaro
>            Assignee: Lou DeGenaro
>            Priority: Minor
>             Fix For: 2.0.0-Ducc
>
>
> Once that all JPs have stopped, perhaps JD should insure queue is purged.
> // The ActiveMQ JMX domain
> > String amqDomain = "org.apache.activemq";
> >
> > // The parameters for an ObjectName
> > Hashtable<String, String> params = new Hashtable<String, String>();
> > params.put("Type", "Queue");
> > params.put("BrokerName", brokerName);
> > params.put("Destination", queueName);
> >
> > // Create an ObjectName
> > ObjectName queueObjectName = ObjectName.getInstance(amqDomain, params);
> >
> > // Create a proxy to the QueueViewMBean
> > QueueViewMBean queueProxy = (QueueViewMBean)
> > broker.getManagementContext().newProxyInstance(queueObjectName,
> > QueueViewMBean.class, true);
> >
> > // Purge the queue
> > queueProxy.purge();
> >
> See Jira 3577.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to