uima-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jaroslaw Cwiklik <uim...@gmail.com>
Subject Re: [jira] Commented: (UIMA-1288) Need a Better Approach For Specifying UIMA AS Broker URL in Deployment Descriptor
Date Fri, 27 Mar 2009 13:09:29 GMT
First, your notation is wrong. The syntax for a placeholder is: ${brokerURL}
not {@brokerURL}

I dont think we use vm://localhost anymore. This was used for internal
communication between colocated
services. We use java queues for this now.

Top level services use either http or tcp. Keep in mind that we most likely
will need to support a list of URLs in an near future, where Broker URL may
look like this:   tcp://host1:port1,host2:port2,...
The above will be needed to support network of brokers

Jerry C



On Thu, Mar 26, 2009 at 5:49 PM, Marshall Schor (JIRA) <
uima-dev@incubator.apache.org> wrote:

>
>    [
> https://issues.apache.org/jira/browse/UIMA-1288?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12689687#action_12689687]
>
> Marshall Schor commented on UIMA-1288:
> --------------------------------------
>
> dd2spring looks at all the brokers, and for those brokers other than
> "vm://localBroker...." it generates a factory for a specific external queue
> broker looking something like
> {noformat}
> <bean id="an id generated from the broker name"
>            class="org.apache.activemq.ActiveMQConnectionFactory">
>  <property name="brokerURL" value="{@brokerURL}"/>
>  <property name="prefetchPolicy" ref="prefetchPolicy"/>
> </bean>
>
> and a similar bean for the reply queue...
> {noformat}
>
> WIth this proposal, I could always generate this, even for brokers that
> later turn out to be "vm://localBroker... " kinds.  Is this OK?
>
> Dd2spring only generates beans for the localhost connection factory, and
> the broker deployer service if  vm://localhost is specified as the top-level
> broker.  If this was substituted, the logic would fail to generate those
> beans.  Should those beans always be generated just in case, if the ${...}
> syntax is used?
>
> The following error checks, now done by dd2spring, would not work (please
> confirm this is OK)
>  * vm: protocol was not being specified for a remote delegate
>  * top level input Queue broker must be a tcp:// style for C++ component
>
> > Need a Better Approach For Specifying UIMA AS Broker URL in Deployment
> Descriptor
> >
> ---------------------------------------------------------------------------------
> >
> >                 Key: UIMA-1288
> >                 URL: https://issues.apache.org/jira/browse/UIMA-1288
> >             Project: UIMA
> >          Issue Type: Improvement
> >          Components: Async Scaleout
> >            Reporter: Jerry Cwiklik
> >
> > Each UIMA AS deployment descriptor requires at least one Broker URL for
> the connection factory. Instead of hard coding the URL, perhaps a
> placeholder for it should be used. At runtime, the placeholder could be
> resolved to the actual URL by Spring resolver component. The value could
> come from either an external file or from an environment. The new approach
> should accommodate specifying more than one placeholder, since an aggregate
> may have delegates whose queues are managed by a different broker. The new
> approach would simplify deployment of UIMA AS services, especially during
> testing.
> > A placeholder syntax could be inherited from Spring, where it looks like
> ${placeholder-name}.
> > dd2spring would need to change to handle a placeholder notation in
> addition to supporting hard coded broker URL that we use now. A new bean
> must be added to the generated xml. It will be :
> > <bean
> class="org.springframework.beans.factory.config.PropertyPlaceholderConfigurer">
> >     <property name="systemPropertiesModeName"
>  value="SYSTEM_PROPERTIES_MODE_OVERRIDE"/>
> >    </bean>
> >
> > This bean is setup to resolve placeholders using system variables, like
> -Dplaceholder-name=placeholder-value
>
> --
> This message is automatically generated by JIRA.
> -
> You can reply to this email to add a comment to the issue online.
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message