on 03-16-2007 1:20 PM
Hi All,
We have client requirement where if posting of data into SAP system by XI is not successful there should be some mechanism of resending it.
What should be approach for this? Is there resending facility in XI? Or should we handle resending mechanism in SAP system?
HI,
we can do it by creating the package of messages. and schedule the error messages only at certain period of time i mean for example sending the alert from message mapping to the respective user if any data error occurs /sending the wrong data, thse can be rectified before that perticular time.
At that time we have rectified messages and the system error messages ,so before executing the restart messages program , we would have to stop those errors . So now only we have recified errors.these errors can be restarted by the RSXMB_RESTART_MESSAGES program scheduling .
This is the correct way i hope.
Regards
Chilla
<i>reward points if it is useful..</i>
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
One Point I wanted to add :
There will huge amount of data flowing through the interface... in more than 20 millions per day....very challenging situation...
keeping this point in mind ... should we go for 'resending from XI' approach.
or should be do error handling mechanish in sap r3 side?
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
>><i>There will huge amount of data flowing through the interface... in more than 20 millions per day....very challenging situation...
keeping this point in mind ... should we go for 'resending from XI' approach.
See 20 millions per day right but not throguh one message ??
so i would suggest restart one message not all.
>><i>or should be do error handling mechanish in sap r3 side?</i></i>
If errors are realted to application functional data errors then better to handle in R3 side.
else if it is related to strcuture transformation errors then try to rectify in XI.
I always suggest try to send correct data.
Regards
Chilla
Are 20 million messages in error?
This solution is applicable for errors because of temporary issues like connection failure etc.
If your messages in error because of incorrect payload/functional error etc then I recommend to resend from source system (whatever it is).
Before scheduling this Job you have to change IS_RETRY_LIMIT = 10 parameter in SXMB_ADM->configure integration engine->specific configuration
Tuning.
Regards,
Gourav
Hi Ranjit,
see your idoc is not getting posted in R/3.....so do error handling at R/3 side...no need of doing anything at XI side as on resending the same msg from XI to R/3 will again result in the same error in R/3 as your source msg has not changed....
So do your error handing of idoc not posting in R/3 side only......you can have an ABAP code to post an error in your applicaton log of R/3 for every idoc that is not posting in R/3.....
Thanks,
Rajeev Gupta
you can change retry count as per your requirement and business use.
for example if you are resending message in 5 minute then retry count can be more and if resending time is higher then retry count can be less.
Retry count make sure that no message restarted for too long (by default 999 times).
Regards,
Gourav
hi Gaurave,
as far possibility of error is concerned:
there could three locations where errors can come:
1. In XI : error due to mappings--- this will be handled in XI
2. During posting (we are calling BAPI through a RFC adapter) into SAP system---
this could be primarly three causes for this kind of error
-
object locks
-
static data not configured (like some data that need to configured in r3 system prior to posting but not done)..
-
Connection failure.
3. In SAP system: Posting by XI is successful but functinality of BAPI is not achieved somehow due to incorrect data sent from source/others reason------this will come in functional error i guess... handling this errpr will not be in our scope
Basically we are concerned about handling point 2 kind of situation.
As per client,source data will always be correct as it is coming after all the processing by different proven legacy systems...
hi
SEE
The parameter IS_RETRY_LIMIT contains the total number of repeat attempts if a retry-enabled error occurs during asynchronous processing and the error is set to retry-status in the queue.
This parameter is only relevant if the qRFC queue concerned has specified a larger retry parameter. In other words, the smaller of the two values is always used.
Possible Values
0 - 100
donot set more . because it will be perormance issue. that much number of time IE will try to repeat .this is not relevent.
because if error occurs we need to resend after correcting the error .
Also scheduling the restart message sis also the same issue.
My concern is why you want to restart the messages when? after rectfying the message or without rectification it should be restarted?
see the below link only for your information.
Reconciliation of Messages in BPM - /people/krishna.moorthyp/blog/2006/04/08/reconciliation-of-messages-in-bpm
Regards
Chilla.
For point 2 this solution is perfect, but also schedule queue processing program. I'll give you detail afterward in detail.
functional error related to payload can be corrected in SP19 but I dont recommed editing payload see /people/gourav.khare2/blog/2007/02/28/sapxi--message-editing-150-responsibility-and-liability-150-sp19
All functional error should be handled by functional owner of interface and this is out of scope of any XI support.
Regards,
Gourav
--
<i>Please reward points if it helps you</i>
Hi Ranjit,
see if the idoc is not getting posted because of some functional error in R/3, then there is no use of resending the same idoc to XI......in this case ask functional people why idoc is not getting posted - if they tell that some field should be mapped differently, then change mapping in XI and retrigger that idoc.....
Thanks,
Rajeev Gupta
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Yes it is possible, Schedule Job in SM36/37
RSXMB_RESTART_MESSAGES for every 10-15 minute.
This is ABAP report and you can find it in SE38 on XI, if you schedule this report XI will automatically try to resend all reprocessable error messages.
Message was edited by:
Gourav Khare
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
It is ABAP report (ABAP program) and you can see it using transaction SE38 on your XI system.
Following steps need to schedule it as Job:
1. use transaction SM36
2. give any name to this Job say "RESTRAT_MESSAGE"
3. Execution Target -- your XI system
4. On Menu Bar Click on "Step" button type ABAP program "RSXMB_RESTART_MESSAGES" leave other column blank.
5. click on "Save" Floppy icon
6. goback and then click on "Start Condition"
7. Click on Immediate then click on "Periodic Job" checkbox
8. Click on Periodic Value button and specify periodic value "Other value"
9. Add 15 in minute box and 0 in all other.
Regards,
Gourav
--
<i>Please reward point if it is helpful for you</i>
HI,
We have option to resend in Runtime Workbench, Message monitoring
and also we can resend/restart it fromSXMB_MONI also..
Also we can do it by scheduling the restart program
see the below links
message monitor and troubleshoot -
Monitoring for Processed XML messages - /people/krishna.moorthyp/blog/2005/12/23/monitoring-for-processed-xml-messages-in-abap-proxy
monitoring BPm https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/uuid/e7bc3a5a-0501-0010-1095-eb4...
Regards
Chilla
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Its possible. Check RWB.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
88 | |
10 | |
10 | |
9 | |
7 | |
7 | |
6 | |
5 | |
4 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.