Unreliable behavior of MulticastSocket's receive method

I have a program that uses one MulticastSocket to receive packets from a multicast address on a given port (running on a dedicated thread), and another MulticastSocket to send both multicast- and unicast-adressed packets from the same port (running on a separate thread). The program runs on a Windows system that communicates with a separate program (written in C++) running on a Linux system (more specifically, a Gumstix processor running Ubuntu). Both applications are written to send some messages by multicast, and others by direct unicast, with all sending and receiving being performed using the same port number.
The problem I'm having is this... The MulticastSocket SOMETIMES receives packets sent by the remote system by BOTH multicast addressing AND unicast addressing, but other times, ONLY receives the packets sent by multicast addressing. By "sometimes" and "other times," I mean "distinct executions of the program." That is, if I run it one time, the MulticastSocket ALWAYS behaves one way, and if I run the same program again, it ALWAYS behaves the other way.
I know that this smacks of thread-interaction issues, but I have done due diligence to rule this out as best I can.
All of the documentation about MulticastSocket says that it is perfectly reasonable to create multiple MulticastSockets on the same socket address (i.e., the same port), by virtue of the fact that the class sets SO_REUSEADDR to true by default. But if the code never instantiates the sending MulticastSocket, the receiver seems to receive unicast-addressed packets quite reliably.
How can I ensure that the receiving MulticastSocket behaves in a predictable manner? Are there known issues with this?

I don't understand why you're using two MulticastSockets for this. You can send and receive from the same MulticastSocket simultaneously.

Similar Messages

MulticastSocket's receive() method acting strangely

In the Java 2 API documentation, the DatagramSocket.receive( DatagramPacket dp ) method is said to block until a packet is available to read. But i wrote an app that reads a 1.3 mbps multicast (mpeg-1 video) which sends out 1024 byte packets and the java app, after reading a few 1024 byte packets, starts to read smaller and smaller packets until the packets are just 34 bytes, and don't contain any data. I have one DatagramPacket object that's being read in this manner:
The loop, after reading the packet, sends a datagrampacket that uses the same data buffer out a different port, to a different address. (A multicast to unicast bridge)
while( true ) {
multicastsock.receive( dp );
Any ideas why its not blocking until its actually receiving a packet after a while, or why it sees the packet but not the data?
Thank you

So, you're sending multiple packages, right?
receive() blocks until <b>one</b> package is received.
I doubt you can't get data (if it's really in there).
Did you try to debug it (print it or smth.) with smth. like
DatagramPacket confirmationPacket = new DatagramPacket(buf, buf.length);
socket.receive(confirmationPacket);
String received = new String(confirmationPacket.getData());

Strange behavior of std::string find method in SS12

Hi
I faced with strange behavior of std::string.find method while compiling with Sunstudio12.
Compiler: CC: Sun C++ 5.9 SunOS_sparc Patch 124863-14 2009/06/23
Platform: SunOS xxxxx 5.10 Generic_141414-07 sun4u sparc SUNW,Sun-Fire-V250
Sometimes find method does not work, especially when content of string is larger than several Kb and it is needed to find pattern from some non-zero position in the string
For example, I have some demonstration program which tries to parse PDF file.
It loads PDF file completely to a string and then iterately searches all ocurrences of "obj" and "endobj".
If I compile it with GCC from Solaris - it works
If I compile it with Sunstudio12 and standard STL - does not work
If I compile it with Sunstudio12 and stlport - it works.
On win32 it always works fine (by VStudio or CodeBlocks)
Is there any limitation of standard string find method in Sunstudio12 STL ?
See below the code of tool.
Compilation: CC -o teststr teststr.cpp
You can use any PDF files larger than 2kb as input to reproduce the problem.
In this case std::string failes to find "endobj" from some position in the string,
while this pattern is located there for sure.
Example of output:
CC -o teststr teststr.cpp
teststr in.pdf Processing in.pdf
Found object:1
Broken PDF, endobj is not found from position1155
#include <string>
#include <iostream>
#include <fstream>
using namespace std;
bool parsePDF (string &content, size_t &position)
    position = content.find("obj",position);
    if( position == string::npos ) {
        cout<<"End of file"<<endl;
        return false;
    position += strlen("obj");
    size_t cur_pos = position;
    position = content.find("endobj",cur_pos);
    if( position == string::npos ){
        cerr<<"Broken PDF, endobj is not found from position"<<cur_pos<<endl;;
        return false;
    position += strlen("endobj");
    return true;
int main(int argc, char ** argv)
    if( argc < 2 ){
        cout<<"Usage:"<<argv[0]<<" [pdf files]\n";
        return -3;
    else {
        for(int i = 1;i<argc;i++) {
            ifstream pdfFile;
            pdfFile.open(argv,ios::binary);
if( pdfFile.fail()){
cerr<<"Error opening file:"<<argv[i]<<endl;
continue;
pdfFile.seekg(0,ios::end);
int length = pdfFile.tellg();
pdfFile.seekg(0,ios::beg);
char *buffer = new char [length];
if( !buffer ){
cerr<<"Cannot allocate\n";
continue;
pdfFile.read(buffer,length);
pdfFile.close();
string content;
content.insert(0,buffer,length);
delete buffer;
// the lets parse the file and find all endobj in the buffer
cout<<"Processing "<<argv[i]<<endl;
size_t start = 0;
int count = 0;
while( parsePDF(content,start) ){
cout<<"Found object:"<<++count<<"\n";
return 0;

Well, there is definitely some sort of problem here, maybe in string::find, but possibly elsewhere in the library.
Please file a bug report at [http://bugs.sun.com] and we'll have a look at it.

ICM_HTTP_CONNECTION_FAILED error on http receive method

Hi,
I am trying to read a XML via http. I have developed a code referencing this blog:
/people/rashid.javed/blog/2007/03/11/cricket-world-cup-http-client-and-simple-transformations
When http receive method is executed it gives me an error.
On executiong of: CALL METHOD client->receive it gives error:
code:    400 message: ICM_HTTP_CONNECTION_FAILED
I treid changing host and buf in code below both to http://www.google.com to rule out the possibility of bad url. Also tried setting time out to 500 in send method. Still no luck.
ICM trace shows following:
Thr 4864] *** WARNING => Connection request from (16/6462/1) to host: http://www.google.com/, service: 80 failed (NIEHOST_UNKNO
How do I solve this error, please give me some tips.
I was looking forum for this issue but was not able to find a solution.
Here is complete code:
DATA: client TYPE REF TO if_http_client.
DATA: host TYPE string.
* DATA: proxyh TYPE string VALUE 'IfYouHave.Proxy.com',
*      proxyp TYPE string VALUE '8080'.
DATA: buff TYPE string,
      respd TYPE string.
DATA: subrc TYPE sysubrc.
CALL METHOD cl_http_client=>create
EXPORTING
    host               = 'http://www.google.com/ig/api'
*    SERVICE            =
*    proxy_host         = proxyh
*    proxy_service      = proxyp
*    SCHEME             = SCHEMETYPE_HTTP
*    SSL_ID             =
*    SAP_USERNAME       =
*    SAP_CLIENT         =
IMPORTING
    client             = client
EXCEPTIONS
    argument_not_found = 1
    plugin_not_active = 2
    internal_error     = 3
    OTHERS             = 4
IF sy-subrc <> 0.
WRITE:/ ' cl_http_client=>create, subrc = ', sy-subrc.
EXIT.
* MESSAGE ID SY-MSGID TYPE SY-MSGTY NUMBER SY-MSGNO
*            WITH SY-MSGV1 SY-MSGV2 SY-MSGV3 SY-MSGV4.
ELSE.
buff = 'GET'.
CALL METHOD client->request->set_header_field
    EXPORTING
      name = '~request_method'
      value = buff.
buff = 'http://www.google.com/ig/api?weather=21218&hl=en'.
cl_http_utility=>set_request_uri( request = client->request
                                    uri     = buff ).
subrc = cl_http_utility=>get_last_error( ).
IF subrc <> 0.
    WRITE: / 'Wrong URI format'.
    EXIT.
ENDIF.
ENDIF.
CALL METHOD client->send
* EXPORTING
*    TIMEOUT                    = CO_TIMEOUT_DEFAULT
EXCEPTIONS
    http_communication_failure = 1
    http_invalid_state         = 2
    http_processing_failed     = 3
    http_invalid_timeout       = 4
    OTHERS                     = 5
IF sy-subrc <> 0.
CALL METHOD client->get_last_error
    IMPORTING
      code    = subrc
      MESSAGE = buff.
WRITE: / 'communication_error( send )',
         / 'code: ', subrc, 'message: ', buff.
EXIT.
ENDIF.
CALL METHOD client->receive
EXCEPTIONS
    http_communication_failure = 1
    http_invalid_state         = 2
    http_processing_failed     = 3
    OTHERS                     = 4.
IF sy-subrc <> 0.
CALL METHOD client->get_last_error
    IMPORTING
      code    = subrc
      MESSAGE = buff.
FORMAT COLOR COL_BACKGROUND.
WRITE: / 'communication_error( receive )',
         / 'code: ', subrc, 'message: ', buff.
WRITE: / 'communication_error'.
EXIT.
ENDIF.
respd = client->response->get_cdata( ).
WRITE:/ respd.
Thanks in advance,
CD
Edited by: CD on Feb 19, 2009 1:46 PM

Check this link..
[HTTP client code 400 reason ICM_HTTP_CONNECTION_FAILED;

(261205913) Q SYNCH-14 Is the client behavior still synchronous for buffered methods?

Q<SYNCH-14> Is the client behavior still synchronous for buffered methods?
Q<SYNCH-14> For buffered methods the client invocation is essentially asynchronous
since the return type for the method is void. Quote from BEA Workshop documentation:
If you add a message buffer to a method of your web service, incoming messages
(invocations of the method) are buffered on the local machine. The method invocation
returns to the client immediately. This prevents the client from waiting until
your web service processes the request. Note that since the buffering occurs on
your end of the wire, the client still must wait for the network roundtrip even
though it will return a void result. But the client does not have to wait for
your service to process the message

Futher information about the possibility of callback:
It may be possible for a synchronous only web service (i.e. MS .net) to even paticipant
in the callback functionality of asynchronous web services. If the client implements
the appropriate methods for the callback but listens for them on a different port
or binding than the SOAP request, then web service may be able to build a response
if the client's "callback URL" is submitted as the beginning part of a conversation.
Watch the BEA developer forum (http://dev2dev.bea.com) for more information about
this approach and other tips and techniques for building web services.
"Adam FitzGerald" <[email protected]> wrote:
>
Q<SYNCH-03> I heard that MS .net only implements synchrnonus method? If
this is true.
Does it means my async methods will only work with J2EE clients?
A<SYNCH-03> I do not know the limitations of .net but let me point out that
is very
difficult to provide asynchronous web service method invocation (this is
different
from an asynchronous web service). HTTP as a general communication protocol
is based
on a request and response paradigm so your client libraries will mostly
likely be
expecting a response even if it is empty (check the asynchronous example
from today
to see that the start method still returns an empty response). You must
distinguish
this from the notion of an asynchronous web service which is a business
operation
that occurs on the server whose return value/result is not directly associated
with
building response to the client. An asynchronous web service can (and generally
will)
be started and stopped with web service operations that are invoked synchronously.
Thus MS .net clients can still be client to WLS hosted web services.

Need help with my file receiving method ??????

Hi, I have a method that can recieve files from other peers. Currently, it can recieve any file name such as d1.txt, d2.txt ...d5.txt. However, my method doesn't know the file name of the file it is recieving and assigns it to c:\\files\d1.txt. How can I change my method so that It knows what the file name is that it is receiving and assign it properly? thanks
public void receive() throws IOException {
                try {
                        ServerSocket ss = new ServerSocket(1234);
                        while (true) {
                                Socket s = ss.accept();
                                BufferedInputStream bis =
                                        new BufferedInputStream(
                                        s.getInputStream());
                                FileOutputStream fis =
                                        new FileOutputStream(c:\\files\d1.txt);
                                BufferedOutputStream bos =
                                        new BufferedOutputStream(fis);
                                int c, i = 0;
                                c = bis.read();
                                while (c != -1) {
                                        bos.write(c);
                                        c = bis.read();
                                bos.flush();
                                bos.close();
                                fis.close();
                                bis.close();
                                s.close();
                } catch (Exception e) {
                        System.out.println(e);
        }

I see two options:
1. send (and thus receive) the file name first
2. use a static variable that you increment for each new file
Kind regards,
  Levi
PS. your writing algorithm is very slow, right?

Status code value after calling the method cl_http_client= receive

Hi,
I am sending a query string to 3rd party via outbound http. Now one of the methods needed in this case is CL_HTTP_CLIENT=>RECEIVE . This method has got 3 exceptions such as communication failure etc. If this method has sy-subrc 0 what does this indicate. If a sy-subrc value of 0 indicates success then why are we again quering the status code value between 200 and 299 . Can it happen that the value of sy-subrc = 0 but the status code value is not between 200 and 299 .
In which cases this situation will occur?
Can anyone elaborate on this.....

Hi i could not solve the problem with abap. We developed another application by using another programming language for this issue.
But i think you should create an instance for authantication. After authantication you should get the cookies .
Then create a new instance for other link and before send and receive methods set cookies ( which you got from authantication session ) for other link in ABAP.
We did not have enough time to solve this in ABAP. Hope you will solve. If you solve please let me know how you have solved.
Here is the code(another programming language), may be it will help you.
req = (HttpWebRequest)WebRequest.Create(@"https://login.xxx.com/authenticate.aspx?userid=xxx&password=xxx");
      req.CookieContainer = cc;
      HttpWebResponse httpResponse = (HttpWebResponse)req.GetResponse();
      HttpWebRequest httpRequest = (HttpWebRequest)WebRequest.Create(@"https://yyy.com/postxml.aspx");
      httpRequest.CookieContainer = cc;
      httpRequest.Method = "POST";
      httpRequest.ContentType = "text/xml";
Edited by: MusaCEBE on Sep 23, 2010 4:34 PM

QueueReceiver.receive(long timeout) and large messages

Hi ...
Our JMS application is using receive(milliseconds). I've been trying to research the exact behavior of the receive() method with a millisecond timeout, but haven't yet found out what happens if:
I specify receive (25000), i.e., timeout interval of 25 seconds;
At the 21-second mark, the response begins to arrive from the application server;
The response payload is quite large, and it takes 10 seconds for it to be completely transmitted down the network.
What happens at second number 25? Either MQ will see that the message is arriving and will wait the extra few seconds to assemble the whole thing and provide it as the return value from my receive(t) method, OR MQ will time me out, regardless.
Anybody know what's going on under the hood?
Thanks in advance!
John Taylor

This of course depends on the provider, but my guess is that most implementations don't "see" the message until it has arrived fully. In your case, this means that the message won't be available from the receive method until after 31 seconds. Rather than trying to time this, you may want to use the asynchronous model with a message listener.
- Bjarne.

What is happening in InputStream.Skip method?

Hi,
I need to read inputstream on Nokia 9500 under special circumstances, but the read method crashes because of a nokia implementation bug*. Interrestingly the skip method works correctly. According to MIDP2 API:
The skip method of InputStream creates a byte array and then repeatedly reads into it until n bytes have been read or the end of the stream has been reached. Subclasses are encouraged to provide a more efficient implementation of this method.
If this is true the skip method manages to read the stream that I don't manage to. My question is: is it possible to check somehow what is happening exactly when the skip method is called? Is it a usual java code? Is it public?
If yes I could copy the code from the skip method and avoid the error above.
I would greatly appreciate any pointers in this!
Thanks!
�rs
*For details please see:
http://discussion.forum.nokia.com/fo...d.php?t=102222

The section, which you are reffering to, describes the behaviour of the skip method from InputStream, but it's quite rare case, where the pure InputStream class is used. More likely implementation overides this method in one of the subclasses because, it's obvious, that more effective techinics can be used fir skip(). So the behavior of the read() & skip() methods is quite different.
In any case there is no way to touch internals of skip() realization.

How to call a static method in a class if I have just the object?

Hello. I have an abstract class A where I have a static method blah(). I have 2 classes that extend class A called B and C. In both classes I override method blah(). I have an array with objects of type B and C.
For every instance object of the array, I'm trying to call the static method in the corresponding class. For objects of type B I want to call blah() method in B class and for objects of type C I want to call blah() method in C class. I know it's possible to call a static method with the name of the object, too, but for some reason (?) it calls blah() method in class A if I try this.
So my question is: how do I code this? I guess I need to cast to the class name and then call the method with the class name, but I couldn't do it. I tried to use getClass() method to get the class name and it works, but I didn't know what to do from here...
So any help would be appreciated. Thank you.

As somebody already said, to get the behavior you
want, make the methods non-static.You all asked me why I need that method to be
static... I'm not surprised to hear this question
because I asked all my friends before posting here,
and all of them asked me this... It's because some
complicated reasons, I doubt it.
the application I'm writing is
quite big...Irrelevant.
Umm... So what you're saying is there is no way to do
this with that method being static? The behavior you describe cannot be obtained with only static methods in Java. You'd have to explicitly determine the class and then explicitly call the correct class' method.

Purchase Order Row UDF for Received Qty

Hi experts,
Can someone please recommend a way to create a row UDF for purchase orders to show the qty received? We cannot use a simple calculation like Qty-Open Qty because if the row is closed manually, the open qty becomes zero. Another problem is how to prompt the PO to update this value if additional GR PO's or AP Invoices are created. Thanks.

I should explain the circumstances more clearly. There is no problem with the standard GR PO or AP invoice receiving method. What I want is a reference field on the original PO showing the total qty received to date. For example: total of 1,000 pcs ordered -> 900 pcs received against several GR PO's -> want to see that 900 has been received back on the PO. Then let's say we want to close the PO line 100 pcs short.
I can think of 2 ways to go about this:
Method 1: Change the ordered qty from 1,000 to 900 -> row is closed, but lose the reference of the original order qty being 1,000.
Method 2: Close row manually and maintain the original order qty as reference, but lose b/o qty and it is difficult to tell how many pcs where received/closed short.
Is there another way to keep all the reference information?

Http_Communication_Failure in IF_HTTP_CLIENT-- receive : BASIS config

Hi,
I am facing "Http_Communication_Failure" error in ABAP code using IF_HTTP_CLIENT->receive method.
I came to know from several SDN threads that it needs some DNS or Network related configuration on SAP R/3 system I am using.
One such thread (2nd last reply)-->
HTTP_COMMUNICATION_FAILURE during IF_HTTP_CLIENT>receive
What configuration to be done here?
Thanks and regards,
Amey

Hi Pranay,
The trace file is huge one this is the part of that,
[Thr 05] Fri Jan 30 00:03:45 2015
[Thr 05] *** WARNING => Connection request from (1/2/0) to host: <Host ID>, service: 1090 failed (NIECONN_REFUSED)
[Thr 05] {00033ca0} [icxxconn_mt.c 2271]
[Thr 13] *** WARNING => Connection request from (1/2/0) to host: <PWD>, service: 1090 failed (NIEHOST_UNKNOWN)
[Thr 13] {00033ca2} [icxxconn_mt.c 2271]

Received message without correlationID

Hi,
I am facing a strange problem.
Whenever I execute any JPD in my weblogic workshop using the Workshop test browser, I am getting following error.
<Error> <WLW> <000000> <Received message without
correlationID. MessageID=ID:P<294710.1177353184648.0>>
1. I created a JPD which has a Client Request Node and uses some WLI JMS controls to post messages to a queue.
When I run this JPD from test browser, I am getting the above error.
2. Then I created another simple JPD with just one single perform node. Even that gives the same error.
3. Then I thought it may be a problem with my domain and created a fresh wli domain and tested the JPD. Still the same error.
I do not understand what could be the problem.
Can someone help me with this.
Thanks

This is a problem with the configuration of your JMS control. For me it worked when I stripped the @jc:jms annotations from the send/receive methods of my JMS control. Following is the code:
package processes;
import com.bea.control.*;
import java.io.Serializable;
* @jc:jms receive-type="queue" receive-jndi-name="TestQ" send-type="queue" send-jndi-name="TestQ"
* connection-factory-jndi-name="weblogic.jws.jms.QueueConnectionFactory"
public interface MyJms extends JMSControl,com.bea.control.ControlExtension
* this method will send a javax.jms.TextMessage to send-jndi-name
public void sendTextMessage(String payload);
static final long serialVersionUID = 1L;
interface Callback extends JMSControl.Callback
void pickMessage(String message);
Message was edited by sodhi at Jan 29, 2005 2:56 PM

Multiple guest users plus wierd behavior

Hello,
I'm having strange behavior in my afp logins - home folder house icon disappears, some users launch a file - it bounces and then it never launches, some share-points come up without mounting (only alias shows), sometimes a user takes 10 minutes to log-in, very long beachballs when opening a folder, etc. These behaviors are intermittent...
I see that in my AFP connections the following:
Name Status Address Connected Idle For
<Guest> Normal machine.name.private 04:57 00:15
<Guest> Normal machine.name.private 03:21 00:15
username Normal machine.name.private 00:15 00.00
The number of <Guest> users from the same machine varies. when I log out of the machine, the <Guest> users are still in the AFP list on the server as being logged in from that machine.
I don't know whether the behavior is related to the <Guest> issue or not, so I have the following questions: is the above status for APF normal? If not, how do I correct? If not, what else may be causing my unreliable behavior?
The server is 10.4.7 and the database is fully kerberized. Services running include: AFP DHCP, DNS, FTP, NFS, Open Directory, and Web.
Thanks,
Dave

I just noticed the same behavior on a client's new dual 2ghz G5 Xserve.
The AFP process spikes the CPU near 80% or so from time to time, though the server's not swapping out too much at all. It has 2 gb of RAM.
AFP looks like it hangs out at a reasonable load most of the time, but it does show a high number of threads in use. I tried setting the maxThreads key from 200 to 300 in /Library/Preferences/com.apple.AppleFileServer.plist. My client is going to restart AFP to see if that helps at all.
Regardless, what might cause the multiple authenticated/guest logins per IP? Has anyone else seen this?
Thanks for any tips you may have.

Why can't I access my button click method??

I have had this problem a couple of times now and I do not understand what is happening. Here is what is happening.
1) I have a login menu screen - user logs in OK.
2) User directed to another screen where user selects option
3) User directed to a third screen where the user selects a name from a list.
(this part below used to work, now it does not)
4) User double clicks on name in list
5) Javascript presses "select button" and is directed to the selectBut_action()
method in the page Java code.
For some reason, I cannot access the selectBut_action() method now and I do not understand why.
I also have a "select" button that accesses the "selectBut_action() method.
Here is the relevant portion of the page JSP code.
<ui:button action="#{EnrollStaffMain.selectBut_action}" binding="#{EnrollStaffMain.selectBut}" id="selectBut"
       style="height: 24px; left: 960px; top: 120px; position: absolute; width: 48px" text="Select"/>
<h:selectOneListbox binding="#{EnrollStaffMain.userListBox}" id="userListBox" ondblclick="getValue()" size="5" style="height: 400px; left: 1055px; top: 250px; position: absolute; width: 150px">
         <f:selectItems binding="#{EnrollStaffMain.listboxSelectItems}" id="listboxSelectItems" value="#{EnrollSessionBean.userListItems}"/>
</h:selectOneListbox>
<script type="text/javascript">
            function getValue()
                     alert( "Double click" );
                     form1["form1:selectBut"].click();
</script>I placed the alert("Double click"); in the javascript just to make sure it is being executed. It is, but the next line no longer accesses selectBut_action().
What is really irritating is that this exact code used to work. I started binding some other variables to session bean properties and it stopped working.
Here is the "selectBut_action()" method from the Java code for the page
    public void selectBut_action()
        String val = userListBox.getValue().toString();
        textField1.setText(val);
        esb.getGeneralStaffInfo( 3 );
    }I have run this in debug mode with a break at the first and second lines of selectBut_action(), but it does not break at that point.
It looks to me like it is just ignoring the Java for the page. I suspect this has to do with some state that it is in, but to not know how it got in that state or how to get out of it.
Any help is appreciated.
Thanks..

The IDs are correct. I did not try any of the other suggestions.
I did find out what caused it, but do not know why it behaved as it did.
I resolved the problem by removing the parameter I had bound to the 'selected' field of a drop down menu.
I was hoping to be able to use that to preset the dropdown to the desired item.
So it appears that the cause was an illegal or invalid binding.
It seems a rather peculiar behavior to ignore the action methods instead of posting some meaningless error message with traceback calls through the bowels of jsf like they normally do.

Unreliable behavior of MulticastSocket's receive method

Similar Messages

Maybe you are looking for