Binding to RMI registry on remote host

Is it possible to bing to a RMI registry on remote host. I get a Access exception when I try it. Is there is any work around for this?
Any help will be greatly appreciated.
Thanks
R

This is the same problem I ran into. Unfortunately there is no way around this. I had to abondon using RMI registry and instead I changed my program to use the tnameserv/orbd stand-alone naming services in a typical JNDI fashion. This tutorial has some examples, good luck:
http://java.sun.com/j2se/1.4.1/docs/guide/rmi-iiop/tutorial.html

Similar Messages

Bind to RMI registry on remote host

Is it possible to bing to a RMI registry on remote host. I get a Access exception when I try it. Is there is any work around for this?
Any help will be greatly appreciated.
Thanks
R

:Like I say, JINI uses a different model to RMI, and you can probably have a look at the forum for that here.
but a way around it is to create a server which listens for requests, accepts a remote object and a remote object name, and binds those into the registry on which its running - remember that the restriction only applies to registering objects, not to looking them up, so, once the object is registered on the remote host, it can be retrieved quite easily.
This of course leaves the problem of exception ahandling aside, but I trust the problems there are obvious

BIND appends my domain to remote host names when querying

I'm running BIND v9.3.0 on Solaris 8.
All the zone files, named.conf, resolv.conf etc seem to be properly
configured.
I get normal name resolution for hosts located inside my v-lan.
Sendmail works inside my v-lan.
However, when I try to hit an internet site outside of my v-lan it
won't resolv.
So, setting nslookup to debug mode, I did a lookup of a remote host.
The result is that, when my local dns is queried, the host name alone
is used, like its supposed to
i.e.
;;res_nmkquery(QUERY, hostname, IN A)
This is a remote host so, obviously, my DNS has no record of it, so it
tries the remote server. This is where the problem comes in. When
the remote server is queried, my domain gets appended to the host
name:
i.e.
;;res_nmkquery(QUERY, hostname.MYDOMAIN, IN A)
Since the host does not reside in my domain, obviously this fully
qualified domain name will never resolve because it isn't correct.
How do I make it stop????!!!!!

I notice the following error logs in server :
EXCH.xxxx.org.xx in the personal store on the local computer. Therefore, it is unable to support the STARTTLS SMTP verb for the connector Default Frontend EXCH with a FQDN parameter of EXCH.xxxx.org.xx. If the connector's FQDN is not specified, the computer's
FQDN is used. Verify the connector configuration and the installed certificates to make sure that there is a certificate with a domain name for that FQDN. If this certificate exists, run Enable-ExchangeCertificate -Services SMTP to make sure that the Microsoft
Exchange Transport service has access to the certificate key.
What this issues,

RMI Registry fail to initiate

Hi Guruz,
Let me describe the problem..
I have 2 apps running on tomcat server. Both the apps initiates RMI server and then regiter the object on RMI registry. If I deploy both the app on a single web server ( Tomcat ) and start the server, it successfully create and register the rmi object but the subsequent app when attempts to register their object, it crashes.
I am pasting the snippet of code that initiate rmi regitry.I am using the same code to initiate rmi registry with the same port.
try{
//Start the RMI registry
registry = LocateRegistry.createRegistry(ServiceProperties.rmiServerBindPort);
catch(RemoteException re){
try {
registry = LocateRegistry.getRegistry(ServiceProperties.rmiAppHost, ServiceProperties.rmiServerBindPort);
} catch (RemoteException remE) {
log.error("could not initialize RMI registry", remE);
throw remE;
Here's the error that is being thrown if subscequent app try to initiate rmi registry. Both the apps have same security policy present. Please give you input why this is happening.
ERROR : remote exception.....Failed to bind to RMI registry : RemoteException occurred in server thread; nested exception is:
java.rmi.UnmarshalException: error unmarshalling arguments; nested exception is:
java.lang.ClassNotFoundException: com.abc.efg.hij.RMIListenerImpl_Stub (no security manager: RMI class loader disabled)
Regards,
Kashif

You can only run one rmi on a single port. You are going to have to use different machines or different ports. How would java be able to distinguish which registry to use if this were possible?

Binding RMI-IIOP Remote Object in RMI Registry through JNDI

hi friends
I am writing RMI-IIOP Remote Object, both server program, and client program
are java programs. Through JNDI (with cosnaming name service), my program is working.
But what i want is, I want to use JNDI (with rmi registry name service) for my RMI-IIOP Remote Object ( and not RMI -JRMP Remote Object). Both my server
program and client programs are java(and not corba)
I am not getting this, while starting server its showing some error
Is it not possible to bind rmi-iiop remote object in rmi registry through jndi, why

because you are supposed to use the COSNaming service with IIOP. Even if you could bind an IIOP remote object into an RMIRegistry the clients wouldn't be able to use it because the RMI Registry doesn't do the extra processing that the COSNaming service does with IIOP references.

Binding an object to a remote registry. Can it be done?

I am having problems vending an object to a remote registry. Most of the examples that I have seen vend the object to a local registry which is not really what I need to do.
I was trying to do:
<PRE>
Remote myObject = ...
Registry r = LocateRegistry.getRegistry("someMachineOnTheNetwork");
r.bind("rmi://mymachine/objectName", myObject);
</PRE>
This fails every time with an AccessException. It seems that the RMI registry only allows binding of objects that are local to it (I don't know of a way of enabling permissions for this. Seems reasonable to expect that not everyone should have the ability to register objects on any registry, but there must be a workaround.
Any ideas?

considering the security reasons the rules of the game is that registry must be in the same host as the probram doing the binding.
So what is LocateRegistry for in that case? Basically it is to be noted that a getRegistry call does not actually make a connection to the remote host. It simply creates a local reference to the remote registry and will succeed even if no registry is running on the remote host. Therefore, a subsequent method invocation to a remote registry returned as a result of this method may fail.
I hope I answered your question.

Some basic questions about rmi registry context "bind" and "lookup"

We have more processing to do than can be accomplished with a single computer. To solve the problem I've implemented a distributed computing solution using RMI. (The first time I saw RMI was about 2 weeks ago, so please bear with me!)
The implementation is a proof of concept not a fully fleshed out system. I have one "Workunit Distributor" computer and any number of "Data Processor" computers all on the same lan segment. "Workunit Distributor" and "Data Processor" computers are both RMI client and server to each other. The "Data Processor" computers are given the ip address and name of the "Data Distributor" on the commandline when they start. They communicate their willingness to receive and process a workunit to the ""Workunit Distributor" via a RMI call. Work units are sent to available "DataProcessors" and results are eventually returned to the "WorkunitDistributor" (minutes or hours later). The model program works quite well, and appears to be capable of doing the processing we need to get done.
But now that it seems viable, I've been asked to make it a little more scalable, flexible and self configuring. In particular, instead of one "Workunit Distributor", any number of "Workunit Distributors" should be allowed to show up or disappear from the lan at any time and the system should continue to function. I've worked out a good scheme for how this can be done, but I have a couple of questions about the RMI registry (registries?). I'm trying to keep from implementing some functionality that may already be available as a library or subsystem.
With my current model design, each computer binds to its own registry with a unique name. For instance:
CRDataProcessorImpl crdpi = new CRDataProcessorImpl(svr);
Context crDataProcessingContext = new InitialContext();
crDataProcessingContext.bind("rmi:"+hostName, crdpi);
Currently the "Data Processors" get the info they need for a Context lookup() of the one and only "Workunit Distributor" from the commandline. And the info the "Workunit Distributor" needs to do a Context lookup() of a "DataProcessor" is passed to it from each "DataProcessor" via a RMI call.
But in the newer (yet to be implemented) scheme where any and all "Workunit Distributors" show up and disappear whenever they feel like, the naming bootstrapping scheme described above won't work.
I can imagine a few ways of solving this problem. For instance, having "Workunit Distributors" multicast their contact information on the lan and have a worker thread on each "Data Processor" keep track of the naming information that was multicast. Another alternative (more organized, but more complex) might be to have a dedicated host with a "well known" address and port that "Workunit Distributors" and "Data Processors" could all go to, to register or look up at an application level. Sort of a "domain name service" for RMI. But both these schemes look like a lot of work to implement , debug and maintain.
The BEST thing would be if there was one plain vanilla RMI registry that was usable by all RMI enabled computers instead of having each computer have its own local name registry. In volume 2 of the Core Java2 book it says that every registry must be local. I'm only hoping there's been progress since the book was published and now a central rmi registry is available.
If you have any ideas about this I'd like to hear what you know.
Thanks in advance for any advice.
Lenny Wintfeld
ps - I don't believe web services, as full featured as it is, is a useful alternative. I'm moving 100's (in the future possibly 1000's) of megabytes back an forth for processing.

The local bind/rebind/unbind restriction is still there and it will always be there.
I would look at
(a) RMI/IIOP, where you use COSNaming as a registry, which doesn't have that registriction, and which also has location-independent object identifiers
(b) Jini.

Simple question about java.rmi.registry.Registry

Hi,
where the object java.rmi.registry.Registry takes a URL to rmiregistry when perform bind method.
thanks.

Ok, will try to explain.
I have a server that I register in the RMI registry. Client from a remote machine get the server (as a remote object) and calls his method for the generation of another object on the parameters passed. The server creates this object and tries to register it in the RMI registry, and this raises an exception Registry.Registry.rebind disallowed; origin /11.0.10.31 is non-local host.
11.0.10.31 is a IP - address of the client

RMI registry on multiple net interfaces

I'm interested in creating several RMI registry servers, each one listening in one different net interface of the same machine (two ethernet cards). Is there any way to do this without using custom RMI socket factories? From the documentation I have read, it looks like there's no possibility to choose a specific network interface. Instead, RMI registry chooses the one associated to localhost.
Thanks in advance.
Zealon

You could use
LocaeRegistry.createRegistry(port,csf,ssf) with an
RMIServerSocketFactory that binds the ServerSocket to
the appropriate network address, but you will then
discover that you can only have one Registry per JVM
because it has a fixed objectID, so don't bother.
But actually, why do you want to do this? Your stated
reason is not correct. By default any ServerSocket,
including those created by the Registry and RMI, is
bound to 'any', which means it should accept from all
network interfaces. Perhaps you have the classic
'hosts' hosts/DNS setup problem where your servers
embed 127.0.0.1 in the stub? Use
-Djava.rmi.server.hostname at the server JVM, or
better still fix your 'hosts' file or DNS setup.
EJPThanks for your reply. Maybe I didn't explain myself enough.
I'm working in medical electrocardiography, and have a PC with two net interfaces. The first one is bound to a LAN IP address that belongs to a medical network. Each one of the hosts in this medical network is an electrocardiography monitor, attached to a patient..
The second net interface is an external IP address (not LAN address).
I created a remote class called MonitorFactory, that should provide information about the electrocardiography monitors existing into the medical network. It works OK, and I cand bind MonitorFactory objects to a RMI registry with no problem at all.
There are some reasons I would like to use several RMI registries:
- First one, I need to control access from external network to my medical network. Medical patient data must be handled very carefully. I thought I could do so by enabling/disabling an RMI registry bound to the external IP address. This way, I could enable access to medical data from outside the medical network, by registering my MonitorFactory objects to this external RMI registry, but only when I decide to do that.
- Second one, I want an RMI registry always running bound to the medical LAN IP address, so I could use the applications that use MonitorFactory like a local application, not only like a remote one.
This scheme represents the system I have to build:
PC with two network interfaces
External PC <----------->|213.xxx.xxx.xxx 192.xxx.xxx.xxx|<----------> Medical network
MonitorFactory - Listens to medical data into 192.xxx.xxx.xxx network only. Bound to
external RMI registry when available and always bound to internal RMI registry.
External RMI registry - bound to 213.xxx.xxx.xxx. Enabled when I decide to do so.
Internal RMI registry - bound to 192.xxx.xxx.xxx. Always enabled.
If I create an RMI registry bound to "any" of the interfaces, I always risk my medical data to the Internet, because there is always a port open at 213.xxx.xxx.xxx accepting connections.
One possible solution could be using SSL sockets for RMI. This way, only allowed clients could connect to RMI registry. But what worries me most is the fact that a port could be open to the Internet - what about attacks to this open port? If those attacks could make RMI registry fall, medical data flow to another information systems could be interrupted, including medical alarms like critical patient status.
On the other way, I would like to maintain RMI default sockets for compatibility and future releases of my application.
Any suggestions?
Thanks in advance.
Zealon.

RMI Registry (bug?)

I am developing an RMI-based program. I want the program to terminate when I unbind all the remote objects from the registry. I think I may have stumbled on a bug.
import java.io.Serializable;
import java.rmi.RemoteException;
import java.rmi.registry.LocateRegistry;
import java.rmi.registry.Registry;
import java.rmi.server.UnicastRemoteObject;
public class Host implements TestInt, Serializable {
     private static Registry r;
     public static void main(String[] args) throws Exception {
          LocateRegistry.createRegistry(1099);
          r = LocateRegistry.getRegistry(1099);
          r.bind("T", UnicastRemoteObject.exportObject(new Host(),1099));
     public void foo() throws RemoteException {
          System.out.println("do smuffin");
}The previous code quits immediately once it gets started because the garbage collector cleans up the exported object [new Host()] once it gets initialized.
To combat this, I keep a local copy of the exported object
import java.io.Serializable;
import java.rmi.RemoteException;
import java.rmi.registry.LocateRegistry;
import java.rmi.registry.Registry;
import java.rmi.server.UnicastRemoteObject;
public class Host implements TestInt, Serializable {
     private static Registry r;
     private static Host h; //keep copy of exported obj so it won't get garbage collected
     public static void main(String[] args) throws Exception {
          h = new Host();
          LocateRegistry.createRegistry(1099);
          r = LocateRegistry.getRegistry(1099);
          r.bind("T", UnicastRemoteObject.exportObject(h,1099));
     public void foo() throws RemoteException {
          System.out.println("do smuffin");
}However, the new code will never exit. I want the prog to exit once I unbind the "T" object in the registry.
import java.io.Serializable;
import java.rmi.RemoteException;
import java.rmi.registry.LocateRegistry;
import java.rmi.registry.Registry;
import java.rmi.server.UnicastRemoteObject;
public class Host implements TestInt, Serializable {
     private static Registry r;
     private static Host h; //keep copy of exported obj so it won't get garbage collected
     public static void main(String[] args) throws Exception {
          h = new Host();
          LocateRegistry.createRegistry(1099);
          r = LocateRegistry.getRegistry(1099);
          r.bind("T", UnicastRemoteObject.exportObject(h,1099));
          r.unbind("T"); //I want the program to quit!
     public void foo() throws RemoteException {
          System.out.println("do smuffin");
}For some reason, this code doesn't terminate either. Is this a bug? Logically speaking, the program should terminate once all objects are either garbage collected or unbound. However it does not.

private static Host h; //keep copy of exported objj so it won't get garbage collected
Logically speaking, the program should
terminate once all objects are either garbage
collected or unbound.Logically speaking you are keeping a static reference to the Host object so it will never be garbage-collected. You either have to release the reference and wait for it to be GCd, which will unexport it automatically, or you can unexport it manually as genady suggested.

Can anybody help me to understand this RMI registry service code?

RMI is so difficult to me. please let me know following code what it means.
If this class is instantiated with 50000 port number,
Does RMI Registry Service get started with 50000 port number in place of rmiregistry commands?
And I don't have to execute rmiregistry command?
thanks in advance.
import java.rmi.AccessException;
import java.rmi.AlreadyBoundException;
import java.rmi.NotBoundException;
import java.rmi.Remote;
import java.rmi.RemoteException;
import java.rmi.registry.Registry;
import java.rmi.server.ObjID;
import java.rmi.server.RemoteServer;
import java.util.Enumeration;
import java.util.Hashtable;
import sun.rmi.server.UnicastServerRef;
import sun.rmi.transport.LiveRef;
* Registry implementation.
* @version 1.0
public class RegistryImpl extends RemoteServer implements Registry {
     private static final long serialVersionUID = -7162736590129110751;
     private Hashtable bindings;
    private static ObjID id = new ObjID(0);
     * @param i port number
     * @throws RemoteException
    public RegistryImpl(int i) throws RemoteException {
        bindings = new Hashtable(101);
        LiveRef liveref = new LiveRef(id, i);
        setup(new UnicastServerRef(liveref));
    private void setup(UnicastServerRef unicastserverref) throws RemoteException {
        super.ref = unicastserverref;
        unicastserverref.exportObject(this, null, true);
     * Lookup a remote object.
     * @param name Name of the remote object.
     * @return Remote object of the specified name.
    public Remote lookup(String name) throws RemoteException, NotBoundException {
//        Logger.debug("NamingService", "lookup: " + name);
        Remote obj;
        synchronized(bindings) {
            obj = (Remote)bindings.get(name);
            if(obj == null)
                throw new NotBoundException(name);
        return obj;
     * Binds the specified name to a remote object.
     * @param name A URL-formatted name for the remote object.
     * @param obj A reference for the remote object (usually a stub).
    public void bind(String name, Remote obj) throws RemoteException, AlreadyBoundException, AccessException {
//        Logger.debug("NamingService", "bind: " + name);
        checkAccess("Registry.bind");
        synchronized(bindings) {
            if(bindings.containsKey(name))
                throw new AlreadyBoundException(name);
            bindings.put(name, obj);
     * Destroys the binding for the specified name that is associated with a remote object.
     * @param name A URL-formatted name for the remote object.
    public void unbind(String name) throws RemoteException, NotBoundException, AccessException {
//        Logger.debug("NamingService", "unbind: " + name);
        checkAccess("Registry.unbind");
        synchronized(bindings) {
            if(!bindings.containsKey(name))
                throw new NotBoundException(name);
            bindings.remove(name);
     * Rebinds the specified name to a new remote object. Any existing binding
     * for the name is replaced.
     * @param name A URL-formatted name for the remote object.
     * @param obj A reference for the remote object (usually a stub).
    public void rebind(String name, Remote obj) throws RemoteException, AccessException {
//        Logger.debug("NamingService", "rebind: " + name);
        checkAccess("Registry.rebind");
        bindings.put(name, obj);
     * Returns an array of the names bound in the registry.
     * @return An array of the names bound in the registry.
    public String[] list() throws RemoteException {
        String[] names;
        synchronized(bindings) {
            int i = bindings.size();
            names = new String;
Enumeration e = bindings.keys();
while(--i >= 0)
names[i] = (String)e.nextElement();
return names;
* TO-DO
public static void checkAccess(String s) throws AccessException {
// check if s is in the list of agents.
if(true)
else
throw new AccessException("");
public static ObjID getID() {
return id;

You don't have any need to understand this code. The complication is due to the registry needing a fixed objectID.
You just need to know the methods of java.rmi.registry.LocateRegistry class. This allows you to start an RMI Registry inside your JVM and give it any TCP port number you like.

Is the RMI registry "process-atomic"?

Hi,
Sorry if this is a really stupid question, and I have really tried to find this out on my own.
When I say "process" I mean Process, as opposed to Thread.
What I mean is, if one Process calls Registry.bind( String, Remote ), can we be sure that another Process which does the same thing one nanosecond later will get an AlreadyBoundException if it calls with the same String?
The context is that up to now I have been starting the RMI Server (only for logging at the mo, these are my first steps) explicitly in a Process. But now I want to move to "lazy" startup of the RMI registry and of the Server... so I would go LoggerInterface.log( bindingName, message ), and if the name "bindingName" is not presently in the registry it will create a new Server, and log the message using it.
More specifically, this is about me trying to automate OpenOffice apps using Java... and it occurs to me that 2 events could indeed happen almost simultaneously, so in fact even explicitly starting up the Server in each handler module might encounter "Process concurrency" problems ... so I need to have an answer to this.
I hope the answer is yes... otherwise I'm going to be a bit flummoxed about how to ensure one doesn't start up one Server Process from one calling Process, and another Server Process from another calling Process.
Thanks

I mean to use rebind() instead of the lookup()/bind() pair. It is atomic.... I understand it is atomic, but my point is that between checking whether you need to rebind (because the bound stub is invalid) and actually doing the rebind there will be an interval. If you systematically use rebind, without a prior check to make sure your Remote is actually attached to an existing Process, every call will cause a new Server Process to be created and then rebound, one after another. Which would be a mess.
I'm going to spend a bit of time looking at your two references... thanks for them.
I have thought of a possible solution involving a sort of lock, but I'm not sure if it works.
1. create a "DummyInterface" extending Remote:
public interface DummyInterface extends Remote {
}2. in the Process + Thread which actually sets up the Registry, bind a DummyInterface
Registry reg = null;
try {
reg = LocateRegistry.createRegistry( 1099 );
} catch (RemoteException e) {
return; // is this right? see below...
DummyInterface dummyStub = new DummyInterface(){};
try {
dummyStub = (DummyInterface) UnicastRemoteObject.exportObject( dummyStub, 0);
} catch (RemoteException e) {
// TODO sthg
e.printStackTrace();
try {
reg.bind( "DummyLockForLoggerServer", dummyStub );
} catch (AlreadyBoundException e) {
// every Thread in every Process (except the first such) to call "bind" will get here
return;
} catch (Exception e) {
// TODO sthg
e.printStackTrace();
}3.     the first Process + Thread which wants to set up the LoggerServer will have first to unbind the DummyLock.
try {
   reg.unbind( "DummyLockForLoggerServer" );
} catch (AccessException e) {
   // TODO sthg
   e.printStackTrace();
} catch (RemoteException e) {
   // TODO sthg
   e.printStackTrace();
} catch (NotBoundException e) {
// every Thread in every Process (except the first such) to call "unbind" will get here
return;
// only one Thread in one Process will ever succeed in getting here
// start the LoggerServer
String qualifiedClassName = ".... rmi.LoggerServer";
String[] a_commandArgs = { "java", qualifiedClassName, "logName" };
Process process = null;
try {
process = Utils.runSubprocess(a_commandArgs, null, ProcessSettings.getJavaRootDir());
} catch (IOException e) {
   // TODO sthg
   e.printStackTrace();
}... however there seem to be one or two questions about this:
- does reg = LocateRegistry.createRegistry( 1099 ); throw a RemoteException if the Registry is already bound? I don't know and intend to do a few experiments to find this out for myself. But in any event only one Process+Thread can ever bind the dummy lock
- there might be a race condition nonetheless to do with the interval between creating the Registry and binding the dummy lock... during this interval the Registry is established, and the dummylock is not yet set up... but I think this does no harm, as a Process trying to start the LoggerServer during this time will get the "NotBoundException" and simply return (and the calling thread will then sleep for a bit before trying again).

Error in re-creating RMI registry when reloading Tomcat server.

Hi,
I use LocateRegistry.createRegistry() in a servlet which is load-on-startup. I've unexport the registered remote object in the HttpServlet.destroy().
But when I reload the tomcat server, such an exception ocurrs:
java.rmi.server.ExportException: internal error: ObjID already in use
        at sun.rmi.transport.ObjectTable.putTarget(ObjectTable.java:168)
        at sun.rmi.transport.Transport.exportObject(Transport.java:69)
        at sun.rmi.transport.tcp.TCPTransport.exportObject(TCPTransport.java:190)
        at sun.rmi.transport.tcp.TCPEndpoint.exportObject(TCPEndpoint.java:382)
        at sun.rmi.transport.LiveRef.exportObject(LiveRef.java:116)
        at sun.rmi.server.UnicastServerRef.exportObject(UnicastServerRef.java:145)
        at sun.rmi.registry.RegistryImpl.setup(RegistryImpl.java:92)
        at sun.rmi.registry.RegistryImpl.<init>(RegistryImpl.java:78)
        at java.rmi.registry.LocateRegistry.createRegistry(LocateRegistry.java:164)
        at RmiUtils.rebindLocal(RmiUtils.java:86)Here is the binding code in RmiUtils.rebindLocal:
        try {
            Naming.rebind(url, rmiImpl);
        } catch (RemoteException ce) {
            if (!ru.isLocalhost()) {
                ce.printStackTrace();
                // cannot cache
                return;
            } else {
                // try to create the registry in local machine if not created
                try {
                    LocateRegistry.createRegistry(ru.getPort());
                    Naming.rebind(url, rmiImpl);
                } catch (RemoteException e) {
                    System.err.println("Failure in create Registry on port "
                            + ru.getPort() + ", maybe it's been created already!");
                    e.printStackTrace();// handle exception
        }Can anybody help?
Thanks.

Hi
I havent code for quite a while.
I would think that this wont work.
The registry is created on startup (possibly init method in ur servlet) but it is never destroyed.
You are better off starting the registry externally to ur servlet engine, and then use do a bind/rebind on startup, unbind on destroy.
Hope this helps.

RMI Registry Workaround

Hi,
I don't want to start RMI registry for my server in different window. How can i bind my remote object without starting RMI Registry externally. I think it can be done , but how i don't know.
Please help
Ritesh Alagh

Read the APIs...
http://java.sun.com/j2se/1.4.2/docs/api/java/rmi/registry/LocateRegistry.html

RMI registry weirdness, API versioning, and more

Hi folks,
     I have some RMI questions. I'd like to start by giving background on the software and our intentions with RMI.
     Our software is a molecular biology analysis and simulation tool. Since the technology is very proprietary, we've used RMI to split the Java software into a front-end GUI client that that's distributed to users and a back-end server that makes use of our in-house C++ tools.
     At the moment our user base is small, mostly a handful of academic researchers, but even with these few releases are getting difficult to manage. The problem, of course, is that when the server back-end is upgraded, if any serializable or remotable APIs are changed, the user must also upgrade their client. Our users have made it clear that this is an unacceptable inconvenience, and since we frequently push new feature-filled releases, its difficult to keep the APIs stable. Additionally, 40% of the original code base is absolute tripe, the inflexible, ill-thought, and monolithic work of a contractor who was at at deadline and had to 'just get the job done'. With only two developers (one of them part-time), we are forced to rewrite portions of this old code in small spurts as most of our time is taken up either adding features or working on the C++ tools. In short, there's simply no way we can stabilize our APIs at this time.
     My first (perhaps naive) solution was simple: the client/server pair are built with a common API version number that is incremented each time the API of a serializable or remotable API must be changed for a new release. This version number is appended to the RMI URI for the server. For example, when using API version 13, if the back-end service is told to listen on the following URI:
     //remote.mycompany.com/myobject
...the back-end would implicitly bind to...
     //remote.mycompany.com/myobject-13
     Since the client and server are built with the same API version, the client knows to append -13 to whatever URI the user specifies to connect to a back-end with a compatible API. This lets the user think they're connecting to the 'usual' URI, but allows us to run servers with varying APIs concurrently on the same physical machine, thus our users can upgrade at their leisure and we can push new releases whenever we please.
     I prefer this solution because its very non-intrusive code-wise and allows us developers complete freedom, but it seems prone to subtle problems. We have three releases (hence three servers) and clients are able to function initially, but eventually they start reporting not-bound and class UID/checksum errors - exactly what this system is supposed to prevent. The three services appear to work when started initially but fail sporadically after some period of time. It seems to me that the RMI registry is getting confused when deciding which classes from which service should be sent to which client.
     I should note that rmiregistry requires us to pass in the path to our classes to its JVM (with -J) or it throws NoClassDefFoundError errors when a client connects to the back-end service (if memory serves, our logger is the offending class, so perhaps the registry is executing a static initializer that prints log messages or something). We only pass in the path to one set of classes right now (the oldest of our API revisions). I'm not sure if passing all three would help, but I don't see how it would, and I really don't have the time to experiment. Interestingly, this class path requirement only started happening a few weeks ago - previously rmiregistry didn't seem to care where our logging classes were. I'm not sure if this is contributing to the concurrency problem stated above, but if there's reasonable suspicion I'm more than happy to figure out why this is happening. Also, I remember reading on some web site that setting the rmiregistry class path with -J could cause other strange problems relating to RMI stubs, so could this be a likely suspect? I wish I was more familiar with the guts of RMI as to know what's going on in the registry. Should I be seeing such problems when binding multiple versions of the same class instance?
     I suppose I'd really like, aside from the informed opinions and suggestions of experienced Java programmers, is an architectural overview of the RMI implementation so I can answer these questions myself, but I haven't found anything in that vein. Ideally I'd like to look at the code for the RMI implementation. Is this possible? I'm not 'hip' to Sun's code sharing policies, what portions of their code base they open up if any. Can anyone offer any hints?
     We did come up with some other possible solutions for the API versioning problem. Dynamic class loading via HTTP in particular is my next best hope. Tentatively speaking, it seems that this would help minimize API breakage so long as modifications are limited to adding elements to the API and not removing them. It seems to me that allowing users to download classes at runtime would allow us developers more flexibility, even though users would be forced to upgrade every once in a while, albeit less frequently than they otherwise would.
     Also considered was a separate tool for automated management of the client software, i.e., an auto-update utility. I'd rather not go down this path; I think it would annoy our users, and our two-man development team is busy enough without worrying about more cruft.
     Any additional suggestions are welcome, as is any insight into the Sun's RMI implementation. Advice, links to documentation, "hey stupid, google for <somthing obvious>" would be appreciated.
Cheers :-)
-Nick

or you could buy mine (http://www.rmiproxy.com/javarmi), but I'll give you three tips:
(1) If possible don't change existing interfaces, just extend them, i.e. version them by extending them:
public interface MyRemoteVersion1 extends Remote {}
public inteface MyRemoteVersion2 extends MyRemoteVersion1 {}This way old clients can still use the old services even if they receive a newer stub.
(2) Do the same with the implementations, and load & register both, with different names as ou are doing. Don't use the same class name for a new version otherwise you will confuse the class loader. Altneratively you could use different class loaders for teh different versions to keep them separate, but that's a large can of worms.
(3) Ignore your feelings about the code base. It's installed out there and it's a product now, you just have to live with it unless you can instaneously upgrade all your clients, which you can't. You will probably always have a client running the original version for reasons entirely beyond your control.
(4) Do use the codebase feature if you possibly can. Don't install anything at a client that it could get via a codebase, e.g. stubs. The only thing that clients should need is the remote interfaces and the the true client code.
(5) Make sure every version of every marshalled class has the same setversionuid and make sure that this is there from the beginning. Make sure you only evolve these classes in ways which will be compatible under serialization - see the spec.
Good luck
EJP

Binding to RMI registry on remote host

Similar Messages

Maybe you are looking for