EJBContext.lookup() vs InitialContext.lookup() @ EJB3.0

current design of my application uses a helper class for retrieving the system resources. according the new ejb3.0 specs i can't use dependency injection in helper classes. so, if we wanna stay with the current design, we should continue with the 2.1 style initialcontext.lookup.
my question is if brand new EJBContext.lookup()is just a wrapper that hides direct jndi usage, or there is any deep reason not to use InitialContext.lookup() anymore even if this means that the whole design should be rebuilt from scratch?
thanks folks!

Good question :-) EJBContext.lookup() is just a convenience method to make the lookup a bit easier to understand.
-- It removes the JNDI API from the developer's view.
-- It removes the notion of "java:comp/env" from the developer's view.
-- It removes the need to catch a checked exception during the lookup.
However, the object returned from EJBContext.lookup("abc") is exactly the same as the equivalent new InitialContext().lookup("java:comp/env/abc"). So, you can feel free to use the older more explicit JNDI component environment lookup wherever you'd like, but for code located directly within the bean class, it's easiest to use injection or EJBContext.lookup.
--ken

Similar Messages

NamingException initialContext.lookup

I have imported a working eclipse J2EE project into JDeveloper 10.1.2.1.0 and fixed some compilation errors by adding the proper libraries. When I run the dcount-ejb project it serves up the login.jsp, but when I attempt to login to the app I get this error:
05/11/02 13:18:16 $$$getRemoteHome() NamingException:
initialContext.lookup of java:comp/env/ejb/UsersSBManager
Error instantiating web-app JNDI-context: No location specified and no suitable instance of the type 'com.cde.decemberCount.ejbJar.users.ejb.SessionBean.UsersSBRemote' found for the ejb-ref ejb/UsersSBManager
I ran "Verify Enterprise Bean" on UsersSBManager and received:
Error : Referenced home interface com.cde.decemberCount.ejbJar.users.ejb.EntityBean.UsersEBRemoteHome not found in project.
Error : Referenced remote interface
com.cde.decemberCount.ejbJar.users.ejb.EntityBean.UsersEBRemote
not found in project.
I have added the dcount-ejb.jar (containing all the beans including the UsersEBRemote) to my libraries and classpath.
What else am I forgetting

In continuing to troubleshoot this problem, I increased the size of the Embedded OC4J Server log and caught this error at the top of the file. I believe it is the core of the problem with the EJB's. I'm hoping its a simple configuration issue:
05/11/03 08:59:30 Error instantiating application 'current-workspace-app' at file:/H:/DCount/dCount-oc4j-app.xml: Error initializing ejb-module; Exception Unable to find/read assembly info for H:\DCount\ProjectNotes\config\build\lib\classes\com\cde\decemberCount (META-INF/ejb-jar.xml)

RFC lookup Vs Database lookup

Dear experts,
When it comes to PI mappings which one is resource or time intensive between RFC lookup Vs Database lookup where the table is hosted in PI itself. I really looking for this comparision in PI 7.1. If anybody looked to compare the pros & cons of using RFC lookup Vs datbase lookup where the table is hosted in PI itself.
Thanks,
-Bharath

I have no data to prove my allegation, but I would expect JDBC to be slower, because you need a separate driver to make the connection to the database, perform the query. The RFC adapter doesn't need that, plus it already has the connection to the SAP system, so you only need to perform the query.
Steps for JDBC:
1. call JDBC driver
2. logon to DB
3. perform query
4. return result
Steps for RFC:
1. perform query
2. return result

Howto: portable bind, lookup remote + local beans (EJB3) to JNDI with WL 10

I struggle writing portable ejb3 applications with weblogic.
I have a JEE5 (EJB3) application using the standard and it works fine on JBoss and WebSphere.
Porting it to Weblogic (10.0) there are quite a lot of problems, mainly
in the area of JNDI lookups of EJBs.
So the issue is how to configure the JNDI-Names of local and remote beans (EJB3!)
in a portable way with Weblogic without making the application unportable.
The biggest problem is caused by the fact, that WL 10 does not bind the EJBs
to the JNDI-tree by default. The bound name would not really be important (JBoss and WebSphere
come up with a value that can easily be used for a generic lookup). Just any binding
would be fine.
So there has to be some extra step in order to get WL 10 binding a remote or local
bean into the JNDI tree. Unfortunately none of them works/is acceptable in an portable JEE app.
Idea1: Using the "mappedName" attribute (of @Stateless, @Stateful)
Well, this one is not portable as the documentation states:
[http://edocs.bea.com/wls/docs100/ejb30/annotations.html#wp1417411|http://edocs.bea.com/wls/docs100/ejb30/annotations.html#wp1417411]
Idea2: Using @weblogic.ejbgen.JndiName
Using weblogic's annotation JndiName also doesn't bind the bean to the given names:
@Stateless
@weblogic.ejbgen.JndiName(local="myapp/XBean/local",remote="myapp/XBean/remote")
public class XBean implements XLocal {
Besides this, that code would not be portable (as nobody wants an weblogic.jar within another application server).
Idea3: Using weblogic-ejb-jar.xml
Mapping a bean using the weblogic-ejb-jar.xml below, also doesn't work.
The bean is not bound to the given name "myapp/XBean/remote":
Exception in thread "main" javax.naming.NameNotFoundException:
Unable to resolve 'myapp.XBean.remote'. Resolved 'myapp.XBean' [Root exception is javax.naming.NameNotFoundException: Unable to resolve 'myapp.XBean.remote'. Resolved 'myapp.XBean']; remaining name 'remote'
Looking at the JNDI tree, there is "something" bound to "myapp.XBean", but it is not the bean.
Doing a lookup with this name ("myapp/XBean" or "myapp.XBean"), gives this:
Exception in thread "main" java.lang.ClassCastException: weblogic.jndi.internal.WLContextImpl cannot be cast to test.server.XRemote
at test.client.Client.main(Client.java:14)
Question 1) What is the preferred way to bind and lookup a remote bean in Weblogic 10?
Again: the goal is to use pure EJB3.
Question 2) What is the preferred way to bind and lookup a local bean in WL 10?
This lookup has to be done on the server in some "unmanaged" classes
where injection doesn't work.
If there is an unportable way to bind a local bean to jndi, it might do for the time.
The important thing would be, that there is no weblogic specific code within the application.
So, solving it using weblogic-xml files would be best.
This declaration
<local-jndi-name>myapp/XBean/local</local-jndi-name>
will come up with the same problems, I guess.
Here's the code. We want the local interface extending the remote interface,
so that the local beans can do all what the remote beans can do too.
XRemote.java
package test.server;
import javax.ejb.Remote;
@Remote
public interface XRemote {
String JNDI_NAME = "myapp/XBean/remote";
void doXRemote();
XLocal.java
package test.server;
import javax.ejb.Local;
@Local
public interface XLocal extends XRemote {
String JNDI_NAME = "myapp/XBean/local";
void doXLocal();
XBean.java
package test.server;
import javax.ejb.Stateless;
@Stateless
public class XBean implements XLocal {
@Override
public void doXLocal() {
System.out.println("doXLocal() called");
@Override
public void doXRemote() {
System.out.println("doXRemote() called");
weblogic-ejb-jar.xml
<weblogic-ejb-jar
xmlns="http://www.bea.com/ns/weblogic/10.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://www.bea.com/ns/weblogic/10.0 http://www.bea.com/ns/weblogic/10.0/weblogic-ejb-jar.xsd">
<weblogic-enterprise-bean>
<ejb-name>XBean</ejb-name>
<jndi-name>myapp/XBean/remote</jndi-name>
</weblogic-enterprise-bean>
</weblogic-ejb-jar>
Client.java
package test.client;
import java.util.Properties;
import javax.naming.Context;
import javax.naming.InitialContext;
import javax.naming.NamingException;
import test.server.XRemote;
public class Client {
public static void main(String[] args) throws NamingException {
InitialContext context = getInitialContext();
XRemote xRemote = (XRemote) context.lookup(XRemote.JNDI_NAME);
xRemote.doXRemote();
private static InitialContext getInitialContext() throws NamingException {
Properties properties = new Properties();
properties.put(Context.PROVIDER_URL, "t3://localhost:7001");
properties.put(Context.INITIAL_CONTEXT_FACTORY,
"weblogic.jndi.WLInitialContextFactory");
properties.put(Context.SECURITY_PRINCIPAL, "weblogic");
properties.put(Context.SECURITY_CREDENTIALS, "weblogic");
return new InitialContext(properties);
}

Hi,
about how to specify jndi name of remote ejb, please refer to this link: http://forums.oracle.com/forums/thread.jspa?threadID=800314&tstart=1
please note @JNDIName isn't an EJBGen annotation. its package name is weblogic.javaee.
So far, based on the EJB spec, local ejb will not be bound to global JNDI. EJB3.1 specification is trying to specify the portable JNDI names including local EJBs, but it isn't published yet.
local EJB is used within one application and ejb reference is used to lookup the local EJB. please refer to EJB specification about how to declare an EJB local reference. once it is declared, it can be looked up from the java:comp\env JNDI namespace.
But, if the local client of the EJB isn't a JavaEE component, there won't be any declaration of ejb-local-ref. You may want to try remote access to the EJB from the non-JavaEE client.

Understand InitialContext lookup - Understanding concepts

I am new to EJB, I am trying to understand the concepts. I am trying to understand example posted on https://glassfish.dev.java.net/javaee5/ejb/examples/Sful.html.
What I don't understand is how Standalone Java Client is doing the lookup and against what is it doing the lookup. Is it just getting the reference from the class that's packaged in jar file ? If so, then how is it maintaining state of the object - isn't that just as using any other class in the package. I would have thought that there would be some trading service like CORBA where server registers the object and client gets the reference to the object. But, in Client examples that I see it isn't clear how it get's the reference to the object that could either be running on the same host or different host in different JVM i.e being deployed by App server.
I am trying to understand specifics and can't seem to get simple and good information about it. Could somebody help me in understanding this.
I want to start this discussion and probably if you guys help me I'll ask some more questions.

When you deploy an application containing an EJB that exposes a Remote interface, the EJB
container registers an object representing the Remote EJB in its naming service. The naming
service lives in the server JVM. The name of the Remote EJB object is vendor-specific. SUN's
approach for assigning global JNDI names (a.k.a. "mapped" names) is here :
https://glassfish.dev.java.net/javaee5/ejb/EJB_FAQ.html#SessionBeanGlobalJNDINameAssignment
The portable way to retrieve an EJB reference is to define an EJB dependency and access that
dependency via a java:comp/env lookup or through dependency injection. However, those options
are not available to stand-alone Java clients, since stand-alone Java clients are not Java EE
components. Take a look at the approach used by the Application Client in the Sful example to
see the difference. There's more on the difference between Application Clients and stand-alone
Java clients here :
https://glassfish.dev.java.net/javaee5/ejb/EJB_FAQ.html#AppClientDef
In the case of the stand-alone client, when the no-arg InitialContext is instantiated, the client
JVM bootstraps a naming provider that knows about the server's naming service. That
works because the appserv-rt.jar has a jndi.properties file that controls the bootstrapping.
Since the stand-alone java client can not use the Java EE component environment to access
the EJB dependency, it is forced to explicitly use a global JNDI name instead. In the example
you cited, the global JNDI name has defaulted to the fully qualified name of the remote business
interface.
--ken

LoadBalance in InitialContext Lookup

Hi,
I'm having following requirement.An Ejb has been deplyed in a weblogic cluster which has two Servers Server A & B.
In Server A the Ejb is in started mode,& in Server B it is in Stopped state.Now from the server B i'm using a standalone client to lookup the ejb using JNDI name & initial context.
+          Hashtable env = new Hashtable();+
+          env.put( javax.naming.Context.INITIAL_CONTEXT_FACTORY , "weblogic.jndi.WLInitialContextFactory" );+
+          env.put( javax.naming.Context.PROVIDER_URL , "t3://SERVERB:9001,SERVERA:9001" );+
+          ctx = new InitialContext( env );+
I had given two URL's [ both Server A & Server B] in the initialcontext. What I'm expecting is the initialcontext will lookup serverB where the EJB is in stopped state, then it will lookup in Server A where the EJB is in started stated & communicate with it.
But now i'm getting the exception, it is throw from ServerB where the EJB is in Stopped state.
+Exception in thread "main" javax.naming.NameNotFoundException: Unable to resolve 'XXXEngineBean'. Resolved '' [Root exception is javax.naming.NameNotFoundException: Unable to resolve 'XXXEngineBean'. Resolved '']; remaining name 'XXXEngineBean'+
+     at weblogic.rjvm.ResponseImpl.unmarshalReturn(ResponseImpl.java:234)+
Thanks in Advance

Hi,
this may be the same implementation problem of class loading inside SAP NetWeaver Java we stumled oger. I posted an OSS but they will not change it.
If you have to EJBs in different enterprise applications, you MUST set an application reference from one enterprise application "A" (typically the calling with the lookup) to the enterprise application "B" that contains the EJB that is looked up.
see [http://help.sap.com/saphelp_nw73/helpdata/en/4a/ee99c229370c9ee10000000a42189c/content.htm]
"To look up a remote object, you must have its remote interface in the clientu2019s class path (for external clients) or have a class loader reference to it (for components and applications running on the server process)."
This makes the class loader of "B" the parent class loader of "A". If you omit this reference, the object that is returned by the lookup is loaded by class loader of "B" and you try to cast it to the class of class loader "A" which of course fails.
Even calling
PortableRemoteObject.narrow(obj, MyEjbLocalHome .class);
before the class cast does not help (because it does nothing to obj at all).
How do you set an application reference?
In META-INF/application-j2ee-engine.xml in enterprise applicatoin "A" enter
<?xml version="1.0" encoding="UTF-8"?>
<application-j2ee-engine xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="application-j2ee-engine.xsd">
<reference reference-type="hard">
    <reference-target provider-name="VENDOR_NAME_OF_EAR_B" target-type="application">APP_NAME_OF_EAR_B</reference-target>
</reference>
</application-j2ee-engine>
If you are working with Development Components, just add a dependency with "Runtime" and this is done automatically.
BTW if you redeploy "B", "A" will be restarted because its parent class loader changed.
Regards
Rolf

Ramifications of caching results of InitialContext(().lookup?

One of the thing we discovered during our early efforts to port a 5.1 app to
7.0 was that in 7.0 the JNDI lookups were simply taking FOREVER. It was
really horrible.
So, the question is, what are the ramifications of caching the results of
this:
Context ctx = new InitialContext();
SessionBeanHome = (SessionBeanHome) ctx.lookup("SessionBean")
We're guessing that this will fail horrible in a clustered environment, but
what about a stand alone environment?
Thanx!
Will Hartung
([email protected])

Can you provide some statistics, how much time it used to take and how much
is it taking now etc.
In 70, We know that the first InitialContext() call will take some time, as
it needs to initialize kernel and generate the hot-codegened initial context
stub. But once you have this call done, next initialContext call should be
pretty fast.
If you want to avoid the hot-codegen cost of stub, use this work around.
From the browser, try
http://server:port/bea_wls_internal/classes/weblogic/jndi/internal/RootNamin
gNode_WLStub.class
Save this class in your client package. This may give some performance
benefit.
This needs that, your classpath servlet should be turned on. See docs for
more info on this.
But I don't recommend this. This may become an issue later and may generate
version incompatibilities, if you upgrade server and forgot to re-pack the
client etc. I am not sure though.
Hope this helps.
Cheers,
..maruthi
"Will Hartung" <[email protected]> wrote in message
news:3d6a8d58$[email protected]..
One of the thing we discovered during our early efforts to port a 5.1 appto
7.0 was that in 7.0 the JNDI lookups were simply taking FOREVER. It was
really horrible.
So, the question is, what are the ramifications of caching the results of
this:
Context ctx = new InitialContext();
SessionBeanHome = (SessionBeanHome) ctx.lookup("SessionBean")
We're guessing that this will fail horrible in a clustered environment,but
what about a stand alone environment?
Thanx!
Will Hartung
([email protected])

InitialContext lookup over HTTP using a full PROVIDER_URL (including path)

Hi,
We have a standalone Java JMS client application that is sending messages to a queue on a WebLogic (WLS 9.2 MP3) server. It uses the WLInitialContextFactory (WebLogic implementation of InitialContextFactory) class to look up the initial context over HTTP when making the connection to send messages to the queue.
This all works fine when the client connects directly to the server - i.e. supplies the address of the server as the PROVIDER_URL. However, in our infrastructure, we cannot connect directly from this client to the server; we need to go via a HTTP proxy.
In a basic configuration HTTP proxy we set the provider URL to be the address of the proxy server, with the proxy is configured to route all requests straight through to the WebLogic server. This configuration also works fine - the JMS client can send messages successfully.
Unfortunately, this basic proxy configuration is still not sufficient for our Production environment. Our production proxy "fronts" multiple servers, each one identified by the first portion of the URL. So, the proxy is be configured such that requests to "http://proxy/server1/test_request" will be routed on to "http://server1/test_request", whereas a request to "http://proxy/server2/test_request" will be routed to "http://server2/test_request". This configuration works fine for basic HTTP requests, to webpages etc.
However, when we try we use this mechanism with our JMS client over HTTP, it is not working. We set the provider URL on the client to "http://proxy/server1" in order to send a message to server1. However, we find that the WLInitialContextFactory class seems to be ignoring the final part of the URL ("server1") and still sending requests directly to the root proxy address of "http://proxy", so it doesn't work.
I appreciate that this can be viewed as a WebLogic specific implementation issue for their WLInitialContextFactory, and I've raised a query with their support people on this. From my own research however, it seems the behaviour of their factory is not different to most other implementations - i.e. the major implementations would all have this issue.
I'm wondering if anyone is aware of any implementations that would handle the situation I've described. If not (or otherwise!) is there something crazy about our approach that noone has catered for it?! It seems a reasonable situation to me, but maybe I'm missing something!
Any comments, advice, etc. would be greatly appreciated, thanks.
Regards,
Adrian
Edited by: AdrianFitz on 08-Apr-2010 12:21

Hi,
Thanks for the quick response. Would you mind providing a bit more detail on your suggestion, as am not 100% sure I understand it fully.
If I see what you are getting at, are you talking about configuring proxy settings on my client side, similar to how I have to configure my internet browser with proxy settings when accessing the internet from within my companys LAN?
If I'm interpreting that right, then I guess my use of the word proxy might be a bit inaccurate or lead to some confusion. What my orginal post described as a proxy should have been more accurately describes as a reverse proxy (acting primarily as a load balancer).
Ideally, other than changing the URL slightly, the client should be 'unaware' that it is going through a proxy at all.
Hope that clarifies somewhat, any further comments would be great! Thanks.
Regards,
Adrian

How can I use a Lookup task to lookup from my SQL Result set and have a join

So in my Control Flow, I have an Execute SQL Task which gets my Table result set. I then have a Foreach Loop Container that iterates through the result set and a Data Flow. The first task in the Data Flow is an OLE DB Source SQL Command that retrieves data
columns associated with my result set. I then do a Derived Column so I can SUBSTRING from one of my data columns and now I want to perform a Lookup to my Application Database.
How do I code my Lookup task to utilize my SQL Result set variable and match on it? I cannot use the GUI for the Lookup task as my Lookup has to have some JOINS in it.
Thanks for your review and am hopeful for a reply.

Can you expand on that? I'm sorry but I am new and a novice to the SSIS world and I want to do this as best I can and as efficiently as I can. Are you saying that Rajen's way suggested above is the way to go?
A little background....external data from a 3rd party client. I'v staged that external data to a SQL Server staging table. I have to try and match that data up to our database using SSN, DOB, and Gender...and if I can't match that way then I have to try
and match by Name. I need to make sure that there is only one and only one account for that match. If I cannot match and match one and only one, then I'll create rows on a DataAnomaly Table. If I do match, then I have to check and make sure that there is only
one and only one Member span for that match. Similarly handle the data anomaly and then check and make sure there is a "Diabetes" claim and similarly handle the DataAnomaly accordingly.
That's where I'm at. Sooooo are you saying to use Rajen's suggestion? I don't think I can do that because I need multiple SQL tasks and I cannot connect multiple OLE DB Source tasks.
Any help and suggestions are greatly appreciated.
Thanks.

How to avoid hardcoding in RFC Lookup and JDBC Lookup during transport...

Hi All,
We have RFC and JDBC Lookups in some of our scenarios(JDBC to RFC synch)....Now problem is during transport of design objects, we have to change the business system and channel name manually in UDF in target environment....as it wont change although we define the target group in SLD.....we want to avoid this hardcoding.....one idea is we can use a business service inspite of business system and assign the comm channel to it.....this business service will be generic for all environment(dev,qua and prod)...and hence we dnt hav to hardcode....pls let me knw how to achieve or its possible or not....
waiting for ur inputs....
Regds,
Pinangshuk.

Hi,
For RFC lookup you might be doing hardcoding for Business system and RFC channel name.
Options are
1) Create value mapping for Business system. So you can maintain it in Directory part.
2) Keep the RFC channel name same for all system (Development, Quality, Production)
The details of channel can be changed in the Directory part of the respective system

Difference between Naming.lookup and Registry.lookup in RMI

Hi Folks,
I have an RMI client which tries to look up a remote object using both
Naming.lookup("//host[:port]/name OR rmi://<host_name>[:<name_service_port>]/<service_name>") and LocateRegistry.getRegistry().lookup("rmi://<host_name>[:<name_service_port>]/<service_name>").
The first one fails whereas the second one succeeds. I am in a fix as to why it is so.Is there any difference between the lookup methods provided by Naming and Registry?
The remote object was registered on the server using
LocateRegistry.createRegistry().rebind("rmi://<host_name>[:<name_service_port>]/<service_name>").

I suggest you take a look at this site to understand how to ask questions.
And don't crosspost again

Is it possible to do lookup without using lookup transformation?

Could you please brief about data masking transformation..?

Hi Deepsikha, You can also use Data Masking Transformation (Substitution Masking) in place of lookup. This performs similar functionality as Lookup. Thanks,Ramesh.

Speed: Array lookup vs field lookup

I have particles in my particle simulator referred to by array lookups - would it be faster to 'chain' the particles by including a 'next' field in each particle instance that links to the next particle in the chain, and refer to that field when moving through the particles?

I have particles in my particle simulator referred to
by array lookups - would it be faster to 'chain' the
particles by including a 'next' field in each
particle instance that links to the next particle in
the chain, and refer to that field when moving
through the particles?so basically you're asking whether iterating a linked list is faster than iterating an array. I guess the array will be an unmeasurable amout of time faster since there'll be no additionalmethod call.
If you really want to know, do both and test.

Show itemcode on production order lookup and BOM lookup

When looking up a Production Order or Bill of Material, the ItemCode is not available in the lookup window. Even when you "Product No." in the case of the Production Order or "Parent Item" for BOMs you get the Item description rather than the ItemCode.
Many clients know their part numbers well and would like to be able to look by ItemCode. I know I could build in a Formatted Search/UDV to one of the fields but it seems like this should be a natural.
Am I missing something?
Alan

Hi Alan.......
Are you trying to say when you Press Tab on Item Code Field in Production order at parent level there you dont find the Item Code which you want to be produced?
If yes then It happens if you are doing Standard or Disassembly Production Order and Bill if Material of such product is not present in Database then it wont list up......
Please check the BOM.......
Regards,
Rahul

HP ALM Premium - Unable to customize filter lookup list, screen lookup list is working

We are using a Project Template to create a base project.
We have about 6 user fields where we define lookup lists that are different for each project.
Workflow code is used to reassign the list values from the DEFAULT list to a specific project list.
This is working for the screen; however the lookup list for the filter is still using the DEFAULT list. We are having a similar problem when using the upload facility.
Any suggestions outside of adding all the items from each of the project lists to the DEFAULT list?
Thanks

Hi Valdas,
Please check the following,
http://www.andrewconnell.com/blog/Applying-Filters-to-Lookup-Fields-with-the-SP2013-REST-API
Best Regards.
Kelly Chen
TechNet Community Support

EJBContext.lookup() vs InitialContext.lookup() @ EJB3.0

Similar Messages

Maybe you are looking for