Flow Parallel Processing Fault handler

I need to implement a solution that uses a parallel processing part. If some of the parts generate a fault I need to compensate all the others. There are some way to do that automatically or I have to design a switch that decides based on the designation of the raised fault ?
Thanks 4 the help.

brian,
sounds to me that it will be the second approach.
clemens

Similar Messages

Parallel processing--share your suggestions

Hi Experts,
I have a program which is taking long time (min 10days)to run because of 4 millian data.Thing is i am planning to use parallel processing, i have gone through help documents.They have used one Syntax
"CALL FUNCTION func ...STARTING NEW TASK task name "
ans some additions.
What this function module name?.Do i need to create a
"RFC function module" and call my report from this function module.My problem is how to use my progarm and this SYNTAX.
Can any will explain and give some suggestion on it.
Note : I fine tunned this program Max.extend
Thanks in advance.
Murali

Hello Murali,
There are many considerations for starting the job in parallel.
1. u should be able to logically break the job for parallel process to handle.
2. When the jobs starts in parallel mode there is no sequence in which they will complete and so should be independant of each other.
3. The FM which is called in CALL FUNCTION func ...STARTING NEW TASK task name should be RFC enabled.
4. In order to process the jobs in parallel ur system should have atleast 3 dialog work process and should have one dialog process free when u start parallel processing.
I am assuming u don't have the FM which is why u are asking what is the fucntion module name. I would suggest u need to modify ur code in such a way that u are able to create a FM which is going to be called in the CALL FUNCTION func ...STARTING NEW TASK task name.
Abhijit

Fault Handling in BPEL process

hi
We have to develop an application which involves several BPEL processes and proxy service(OSB) and JMS queue etc.
What is the difference between Fault Management Framework in SOA11g and normal fault handling(using catch and throw activities)?
which one we should prefer for fault handling in our application n why?
Plz help thanks in advance.

Hi-
In Normal Fault handling process, you will be able to catch the faluts only but in Fault Management Framework if a fault occurs the framework catches the fault and performs a user-specified action defined in a fault policy file associated with the activity. You can also have a human intervention prescribed in it, where you perform recovery actions from Enterprise Manager.
Pls go throgh the below post for detailed description
http://download.oracle.com/docs/cd/E12839_01/integration.1111/e10224/bp_faults.htm
Fault Management Framework and non BPEL soap faults
Edited by: 333333 on Mar 23, 2011 5:01 PM

Fault handling policy for a process in the fault-binding.xml

Can I specify a fault handling policy for a specific process using fault handling framework 10.1.3.3?
<process faultPolicy="DefaultPolicy"/>
I cannot specify the name of the process according to the xsd :(
Can this be achieved?
I have a two different partner links in different processes having the same name. If I add the partner link to the fault-binding.xml what would be the result?
Note: I cannot use bpel.xml to specify the fault policies as of now.

The fault policy bindings file does not allow you to specify a specific fault policy for an individual processes. You can only do this for all BPEL processes (eg: process Level) or at finer grained levels such as partner link levels.
You might be able to use the partner link level for any services that call the process in which you wnat to error handle (eg: Process C needs to be error handled. Process A and B call C so for the partner link names for invoking C could have the fault policy defined for them). Therefore any faults returned from C could propagate back to process A / B and be handled within by the policy outside of Process C. Obviously if this apporach was used you would not want to use process level definitions as C would continue to use this.
If the partner links were of the same name, they should both be handled by the fault policy that is defined withiin the Fault policy (e.g Partner Link level definitions)
An approach that can be used is to specify the fault policy within the bpel.xml file. The information would be added as follows:
</partnerLinkBindings>

<faultPolicyBindings>
<process faultPolicy="AProcessFaultPolicy"/>
<partnerLink faultPolicy="APartnerLinnkPolicy">
<name>insertSSN_dbAdapter</name>
<name>Another_Adapter</name>
</partnerLink>
</faultPolicyBindings>

</BPELProcess>
</BPELSuitcase>
However, I noted you said you were unable to do this. I wasn't sure if this was for technical/governamce reasons or knowledge reasons (unsure where to place to the details.
Hope that helps and does not confuse matters
Dave

Parallel processing using flow activity

Hey, I have two input directories in the same process for two different partnerlinks and I want to use these two directories to input two flat files to two tables simultaneously.I tried using flow to do this parallel process. But the process when uploaded onto the console takes one input file and it fails. It changes the state of the process to 'off'. Can anyone let me know as to what could be the reason? thanks and regards

Hi can someone help me please

Parallel Processing : How to Handle Resource failure?

Hi,
I have implemented the parallel processing/ asynchronous rfc call in my system because we have to process millions of records and processing is important. My Program does work fine in Development and quality for small number of records but during SVT I am encountering RESOURCE_FAILURE exception. As of now I have tried to wait for more time and then process it again and also on failure I have tried to process sequential but nothing worked with second approach that is on resource_failure execute normal FM call it is resulting in terminating Parallel processing.
Any Pointer on how to handle it is appreciated.
Regards,
Deepak Bhalla

<b>Handling the RESOURCE_FAILURE exception:</b> As each parallel processing task is dispatched, the SAP system counts down the number of resources (dialog work processes) available for processing additional tasks. This count goes up again as each parallel processing task is completed and returns to your program.
Should your parallel processing tasks take a long time to complete, then the parallel processing resources may temporarily run out. In this case, CALL FUNCTION returns the exception RESOURCE_FAILURE. This means simply that all dialog work processes in the RFC group that your program is using are in use.
Your program must now wait until resources become available and then re-issue the CALL FUNCTION that failed. In the sample program, we use a simple, reasonably failsafe wait mechanism. The program waits for parallel processing tasks to return, freeing up resources. The WAIT also specifies a initial timeout of 1 second. If the CALL FUNCTION again fails, the WAIT is repeated with a longer time-out. You can increase the time-outs if you expect that your parallel tasks will take longer to complete. You should also add code to exit from the retry loop after a suitable number of iterations.
Use WAIT statement.
Hope this resolves u r issue.
- Raj

Fault handling in osb using publish

Hi all,
i am new to OSB, every thing is working fine. i will explain how we design the basically we have inbound and outbound. call inbound as 'aaaa' and outbound as "bbbb". we have created the proxy and BS which will communicate with EBS, BAM and B2B. i have created BS for all the three. from PS i used 2 assign and 1 publish activity. we are publishing directly to BS correspondingly. since i am not using any route activity i could not get the error back or any invocation error messages. can any one tell how to capture the error and place it into BAM report. we can transform data to BAM we just need to get the details about the fault handling in OSB.
thanks ----

Hi prabu,
thanks for your response.
As we have the discussion on this issue for quit a long time. i have configure the error handler in the proxy services when my EBS throws the error. error handler come active and some data is published to bam. here i got the problem is that, the PO which we are sending is one way or fire and forget when using Publish action. when the PO hit the EBS, i.e. when error message is passed through the test XML, error is occuring and and reporting to bam with some null values. i need to capture the data processed at that transaction. is it possible to capture the data in error handler which is errored out in message flow of PS.
can you please help me
thanks--
anil

Parallel processing of mass data : sy-subrc value is not changed

Hi,
I have used the Parallel processing of mass data using the "Start New Task" . In my function module I am handling the exceptions and finally raise the application specific old exception to be handled in my main report program. Somehow the sy-subrc is not getting changed and always returns 0 even if the expection is raised.
Can anyone help me about the same.
Thanks & Regards,
Nitin

Hi Silky,
I've build a block of code to explain this.
DATA: ls_edgar TYPE zedgar,
        l_task(40).
DELETE FROM zedgar.
COMMIT WORK.
l_task = 'task1'.
ls_edgar-matnr = '123'.
ls_edgar-text = 'qwe'.
CALL FUNCTION 'Z_EDGAR_COMMIT_ROLLBACK' STARTING NEW TASK l_task PERFORMING f_go ON END OF TASK
    EXPORTING
      line = ls_edgar.
l_task = 'task2'.
ls_edgar-matnr = 'abc'.
ls_edgar-text = 'def'.
CALL FUNCTION 'Z_EDGAR_COMMIT_ROLLBACK' STARTING NEW TASK l_task PERFORMING f_go ON END OF TASK
    EXPORTING
      line = ls_edgar.
l_task = 'task3'.
ls_edgar-matnr = '456'.
ls_edgar-text = 'xyz'.
CALL FUNCTION 'Z_EDGAR_COMMIT_ROLLBACK' STARTING NEW TASK l_task PERFORMING f_go ON END OF TASK
    EXPORTING
      line = ls_edgar.
*&      Form f_go
FORM f_go USING p_c TYPE ctype.
RECEIVE RESULTS FROM FUNCTION 'Z_EDGAR_COMMIT_ROLLBACK' EXCEPTIONS err = 2.
IF sy-subrc = 2.
*this won't affect the LUW of the received function
    ROLLBACK WORK.
ELSE.
*this won't affect the LUW of the received function
    COMMIT WORK.
ENDIF.
ENDFORM.                    "f_go
and the function is:
FUNCTION z_edgar_commit_rollback.
*"*"Interface local:
*" IMPORTING
*"     VALUE(LINE) TYPE ZEDGAR
*" EXCEPTIONS
*"      ERR
MODIFY zedgar FROM line.
IF line-matnr CP 'a*'.
*comment raise or rollback/commit to test
*    RAISE err.
    ROLLBACK WORK.
ELSE.
    COMMIT WORK.
ENDIF.
ENDFUNCTION.
ok.
In your main program you have a Logical Unit of Work (LUW), witch consists of an application transaction and is associated with a database transaction. Once you start a new task, your creating an independent LUW, with it's own database transaction.
So if you do a commit or rollback in your function the effect is only on the records your processing in the function.
There is a way to capture the event when this LUW concludes in the main LUW. That is the PERFORMING whatever ON END OF TASK. In there you can get the result of the function but you cannot commit or rollback the LUW from the function since it already have implicitly happened at the conclusion of the funtion. You can test it by correctly comment the code I've supplied.
So, if you want to rollback the LUW of the function you better do it inside it.
I don't think it matches exactly your question, maybe it lead you on the right track. Give me more details if it doesn't.
Hope it helps,
Edgar

How to achieve parallel processing in a single request?

Hi all,
I have a method in a Session EJB that will perform some business logic before it returns an answer to the client. The logic it will perform is to collect data from the applications database and two external systems, before sending all data to a third external system to get a response and send it back to the client. Each external system is quite slow so I would like to do all the collecting of data concurrent, parallel processing. How should I handle this? I'm not allowed to create my own threads in EJB's. Can I use MDB in some way? To the calling client this should be a synchronous call...
Greatfull for any suggestions
Cheers
Anders =)

Usually, the request is received by a component located in the web container, such as by an HTTP request (including Web Services). This component is able to start threads to allow parallel processing. Now, if for some reason the request arrives directly at EJB level and that you cannot move its receiver to web component, I think JMS is not a viable solution because you will switch to asynchronous processing and you have no way to make your EJB wait for the responses while preserving the client request (waiting implies programmatic life cycle management, which is forbidden in EJB container). Maybe a resource adapter (JCA) can bring a solution. A resource adapter acts as a datasource (a datasource is a specialization of a resource adapter) and thus it is a logical way to implement an adapter to an external, eventually non-J2EE, resource, as the name implies :) But I don't have enough knowledge in JCA to be sure of this.
Hope it helps.
Bruno Collet
http://www.practicalsoftwarearchitect.com

FORK is Not happening Parallel processing- It's working sequential

Hi,
   we are into PI 7.O and SP 13.
   I am trying to test Parallel processing using Fork step. (With Two branches)
My problem is sxm_moni both branches are not executed simultenously and it's executing one after the other.
Did any body done in XI parallel processing using BPM...both calls has to finish at the same time. I mean first call 10 min and second call aslo has to finish first 10 min ..not other 10 min.
I heard this problem from XI 3.0 and PI 7.O. But PI 7.1 did any body test the Parallel processing using Fork step.
   Pls help me is this issue will resolve if I go to PI 7.1.
Regards,
Venu.

Hi Henrique,
they would not necessarily start at the same time but shouldnt also be queued - Customer expecting the response within a 17 sec or 20 Sec but coming response 34 sec will not ok for the customer..tomorrow need add some more target again 17 sec will take...How PI can handle the Multi threading they are checking...I am not sure this problem fixed in PI 7.1 or not.
there're # of connection restrictions in your system? Check that - Where can I check connections restrictions...If you know pls through some light on this.
Also, how's your BPM transactional behavior (did you flag the create new transaction steps)?
- I did not checked the flag for create new transaction step..once my server is up I can check the flag and I can test.
Regards,
Venu.

Clarification needed in the Fault Handling Framework(Error Hospital)

Hi,
I need some clarification on Fault Handling Framework...
Scenario:
My BPEL Process is Asynchronous.(10.1.3.3.1) and I tryed to invoke the another BPEL Process( which is used to select the Fname from the table using DB Adapter) . In the Main BPEL Process I have the Fault handling Framework and I done necessary changes in the bpel.xml. This is successfully working for Invoke Activity when ever an Remote or Binding Fault occurs in the Sub-BPEL Process and this Fault is send to Fault-Policy.xml in the server(bpel/domains/default/config/fault-Policy) and works accordingly to the action mentioned in the policy xml....( I have also made changes in the fault-bindings.xml ).
How ever this Framework is working fine for every actions.
Need Clarification:
1) Only the Binding and Remote faults will work for the Fault handling Framework.
2) can we use User defined faults (Buissness Faults,,,) for this........because I used the User defined faults to invoke the fault-policy.xml and it won't works....
3)Only Invoke Activity can be handled for this Framework or any Other Activity........
4)How to handle other Activity in the Fault-policy.xml...
5)Any difference for Asynchronous or Synchronous in this Fault Handling Framework.......
I also referred the Oracle SOA Suite New Feauters PDF .......
Can any one send me the related documents and try to help me on this.
Please help me on this.......
Thanks,
Ashok.

Hi Chintan,
You are saying that Custom faults are also supported for fault-Policy.xml.
I can't understand that 2nd answer you said. can you please explain how to use the Custom faults in the FaultHandlingFramework..
My Scenario:
I have 2 BPEL Process, BPEL1 will invoke the BPEL2 and if any remote fault or binding faults occurs in the BPEL 2 its returning the same fault to BPEL1 and here I am using FaulthandlingFramework , this fault send to fault-policy.xml and necessary action is took place succesfuly...(like retry, human intervention.....)
But when I throw some CustomFaults(userdefinedFaults)in the BPEL2 , its returning as a remotefault to BPEL1. I don't know how to use the customFaults in the Fault handling framework.......
Can you explain me in detail how to use custom faults in faulthandling framework...............
Send me some related documents for this.....

Integration Process exception handling & alerts, Java

Hello,
I would like to enhance my basic scenarios fool proof and with as much Java as possible.
I have set up several asynchronous and synchronous File to SOAP transfers between systems A and B
using PI70 including my own Java mapping classes and PI's AF_Modules beans. They work just fine.
I have not designed my own Integration Processes yet, so execution is based on channel settings and availability timing.
I have learned this aproach may be exposed to general faults resulting manual monitoring and repairing.
For example server problems at receiver side can result data losses as sender channel just keeps removing source
files like normally, as the process would be better to just stop right there and alert or something.
My question is, is the "Enterprise services Builders" Integration Process & Graphical definition screen the one and only tool to
customize whole process exception handling and alerts in PI, and Java is not an option?
Appreciate your advice on this.
Kind regards m

Hi m,
Strange Name
>>I have learned this aproach may be exposed to general faults resulting manual monitoring and repairing. For example server problems at receiver side can result data losses as sender channel just keeps removing source files like normally, as the process would be better to just stop right there and alert or something.
If you are looking to handle this particular scenario, then we have the alert mechanism and CCMS monitoring. There you will come to know whether the end system down or not.
Also in addition to this you can write your own java (in message mapping, as modules, java mapping) /abap code (as abap mapping, user exit in standard functions etc) for providing more details in error scenarios. But you need to validate whether the maintenance/development cost for the code is justified in your scenarios
Regards
Suraj

"Parallel" process

I don't think I ever got this answered..I'm running a functional test (LV2014..Win 7..DAQ6229) In the functional test.vi is a flat seq structure. In each of these seq's is a subvi... subvi1 then subvi2 etc. Each of these subvis I use the DAQ to measure DIO/voltages...etc. to do my UUT testing. Is there a way in ethier the functional test.vi or the subvis to monintor a boolean which stops the functional vi? Just looking for a way to have the DAQ running its thing...running the test ..and monitor a door switch ( the boolean) so that if the user opens the door to the test set it stops the test and turns on a Software LED that I've put on the front panel GUI of functional test. I know how to implement the boolean..I know how to implement the functional tests. I just don't know how to combined them so that they are both running simutaneously.
Thanks..
Solved!
Go to Solution.

There are two questions here. One is "How do I get an immediate response to a change, either to a control or to an instrument reading?", and the other is "How do I synchronize the activities of two parallel processes?"
The first can be handled by an Event structure residing inside a While loop. This is something that patiently "waits" for one or more of the Events it is programmed to recognize. These are typically changes in the value of a Front Panel control, or "signalling changes" to a LabVIEW control or indicator (this means that you change the value in LabVIEW, but "signal" that the value changed).
There are numerous ways to get two parallel processes to exchange information. If you want to control the second process (such as to stop it), you need one or more "points of intervention". Unfortunately, if you have your process-to-be-controlled structured as a long, single sequence, figuring out an intuitive way of "interrupting" it can be tricky.
One thing you can do (given your existing architecture) is to look at each of the sub-processes (which I gather reside in sub-VIs). Is there a point within them that, if you had a "Stop" VI that returned a True value, would cause the sub-process to immediately exit, and that if Stop were true when it started, would again cause it to exit? [This is sometimes called a VI Global or a Functional Global Variable].
If this is the case, write yourself a Stop VI that contains a While loop with an uninitialize Shift Register holding a Boolean value (and initially False, by default). Wire the output to a Boolean Indicator called Stop, with a Boolean input called "Set Stop". Inside, put a Case statement such that if Set Stop is ever True, it sets the Shift Register (and thus Stop) to True.
In your Event structure, if a Stop condition appears, call the Stop VI with True wired to Set Stop. In each of your sub-VIs, call Stop in such a way that if it ever becomes True, the sub-VI exits (doing a safe shutdown, if necessary).
Note that you can have multiple Stop VIs sprinkled around your code, and even within a sub-VI -- you need to find the "vulnerable" points that are accessed sufficiently frequently to allow Stop to have an effect.
I'll leave it to others to discuss restructuring the entire top-level program (a single sequence is not a good way to do this, if for no other reason than it is absolutely inflexible ...).
Bob Schor

Parallel Process Model Entry

Hello,
First, using LabVIEW 2011 and TestStand 2010 SP1.
We have a custum GUI setup to use the sequential model. At the heart of it, when we hit start, it accesses the Single Pass entry point and runs it with some modifications we've made. Now, we'd like to have a setup that can test multiple DUTs and use the Parallel Process Model. I'd like to have our GUI basically perform the same functions as the built in popup when running the TestUUTs entry point. I've seen the PreUUT callback where you can disable the built in popup. But what I can't seem to figure out is how to initiate a test socket test because it doesn't seem to line up with an entry point necessarily.
When they push the start button next to DUT 0, I want to start socket 0. And if they hit the start button next to DUT 1, I want to start socket 1. What entry point does the GUI use to make this happen? Does the GUI need to start an entry point (IE Test UUTs) then send more information later to start a test? if so, where does TestStand wait for that information within the parallel process model?
Thanks.

One way you could do it is as follows:
Still override PreUUT like in the example and add code which posts a UIMessage synchronously to the UI and then waits for a persocket notification (e.g. you can use the socket index as part of the name to make a per socket notification). Then in your UI, handle the UIMessageEvent on the ApplicationMgr. When you get the UIMessage from your PreUUT you will know it's ready and you can then update your UI to enable the start button, and when the user presses it, you can then Set() the notification to tell the testsocket thread to continue.
for example, in preuut:
Create Notification "MyUIStartNotification Socket 1"
Thread.PostUIMessage(UIMsg_UserMessageBase + 1, RunState.TestSockets.MyIndex, "MyUIStartNotification Socket 1", null, true)
Wait on Notification "MyUIStartNotification Socket 1"
In your UI
Handle UIMsg_UserMessageBase + 1
Enable Start Button
When start button pushed, Set Notification that corresponds to that socket. Pass any data needed with the notification, for example, create a container that contains a serialnumber property and a continuetesting property and pass that with the Set operation. In your PreUUT code get that data from the notification and update your parameters.
Not simple, but doable.
-Doug

Parallel process run independently, but do not stop when vi completes

So I posted a problem yesterday about getting 'elapsed time' express vi to provide updates from a sub-vi to the calling vi. It was suggested that I create a parallel process in the sub-vi that will run the elapsed time function at the same time the other processes are running. I tried to implement this idea, but ran into a problem. The elapsed time process is a While loop paralleled with the main While loop in the sub-vi, It updates every second, based on my time delay, and when I view the running sub-vi, the elapsed time indicator is updating as it should. The calling vi, however, is not seeing these updates. I have wired an indicator to the sub-vi icon, but does not change until the sub-vi finishes.
Also, the other problem with the parallel process is that it runs forever, regardless of the other loop finishing. I have tried to wire an or'd bool to the stop inside the While loop, but when I do that, the elapsed time process does not start.
I have tried data binding a shared variable in the project, and then dragging that to my calling vi, but again I get no updates on the elapsed time.
Any ideas????

A VI must finish looping before it has an available output (the terminal on the sub-vi icon). Research how to communicate between loops. What you are doing can be accomplished with Notifiers (that's what I'd use), Queues, or Global/Shared Variables.
Your issues appear to be due to a lack of understanding of the LabVIEW data-flow paradigm. Check out the Producer / Consumer example. Post code here and see if one of us can give more guidance.
Richard

Flow Parallel Processing Fault handler

Similar Messages

Maybe you are looking for