camel-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roland Knight (JIRA)" <j...@apache.org>
Subject [jira] Reopened: (CAMEL-2692) Multithreading bug: getBody sporadically returns null
Date Tue, 11 May 2010 22:43:42 GMT

     [ https://issues.apache.org/activemq/browse/CAMEL-2692?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Roland Knight reopened CAMEL-2692:
----------------------------------


Claus, the synchronize you added didn't fix the problem. Seems that importNode requires the
Document to be synchronized (yuck). Changing your fix in XmlConverter.toDOMDocument to:
                Document doc = createDocument();
                // import node must no occur concurrent on the same node
                // so we need to synchronize on it
                synchronized (node.getOwnerDocument()) {
                    doc.appendChild(doc.importNode(node, true));
                }
fixed the problem.

I agree about the JDK XML API. It is horrible. I wrote a converter for DOM4J and always convert
the body to a DOM4J Document before any DOM manipulation.


> Multithreading bug: getBody sporadically returns null
> -----------------------------------------------------
>
>                 Key: CAMEL-2692
>                 URL: https://issues.apache.org/activemq/browse/CAMEL-2692
>             Project: Apache Camel
>          Issue Type: Bug
>          Components: camel-core
>    Affects Versions: 1.6.2, 2.3.0
>         Environment: Windows 7 64 bit, JDK 1.6.0_20
>            Reporter: Roland Knight
>            Assignee: Claus Ibsen
>             Fix For: 1.6.3, 2.3.0
>
>         Attachments: news_20100502000001.zip
>
>
> Note that the only workaround for this bug is to remove the parallelProcessing() call
in the builder.
> I have a simple route that processes a file by splitting on a tag and processing the
DOM of each split message. The problem is that getBody is randomly returning null but ONLY
when using the parallelProcessing feature of split. For some runs of the same XML file, the
error does not occur at all (the file is about 2MB of data), others it will happen once or
twice. I am currently using the latest 2.3-SNAPSHOT. 
> Also note, that after detecting the null, I tried calling getBody(String.class) - this
also may return null. Sometimes it does return the proper XML. 
> Route configuration that reproduces the problem (my input XML is about 2MB with about
500 article tags): 
>         public void configure() throws Exception { 
>           from("file:D:/inbox") 
>             .split(new XPathBuilder("//article")) 
>               .parallelProcessing()                     // remove this line getBody below
never returns null 
>               .process(new Processor() { 
>                 public void process(Exchange exchange) throws Exception { 
>                   Message inMessage = exchange.getIn(); 
>                   org.w3c.dom.Document domDocument = inMessage.getBody(org.w3c.dom.Document.class);

>                   if (domDocument == null) { 
>                     log("Null body"); 
>                   } else { 
>                     // process DOM here 
>                   } 
>                 } 
>               }) 
>             .end() 
>         } 
>       }); 

-- 
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