Nested exception is java.net.SocketException: Broken pipe

[Former Member]

Hi All,

We have a webservice to RFC synchronous  scenario via PI 7.0, in which thirdparty request will have employee id and the corresponding employee details will be retrieved and sent back to the thirdparty.

The scenario was working fine for single employee details.But is getting failed for the record having 1300 employee ids.The thirdparty team is facing the below error.

Exception while retrieving SAP Ws : I/O error: Broken pipe; nested exception is java.net.SocketException: Broken pipe

The input does not even reaching XI server.The issue is in production environment.It works fine in development.Please help me in resolving the below issue.

Already check the below note 1394546.

Thanks,

--Ramya

[iaki_vila]

Hi Ramya,

Can you contact with the sender system administrators to know if they have any timeout set in their system or any firewall that could stop the communication?, the broken pipe exception should appear when PI can't find the system that call it synchronously. Check the audit log of the communication channel monitoring of your sender SOAP adapter, if you can observe that the response message has come and the problem is sending the response, you will can assure the problem is not in PI.

Regards.

[Former Member]

Hi Inaki,

We could not find any message log in SXMB_MONI and in the channel.we checked with the network team also , could not find any issue from their end .

Is there is any possibility to trace the issue?

Regards,

--Ramya

[former_member186851]

Hello Ramya,

Did you check the timeout parameter?

[iaki_vila]

Hi Ramya,

Just to add, i have seen sometimes the timeout raised for sender applications outside the PI. If the RFC is not doing any modification you can check with SOAPui in the production environment, to assure the timeout problem comes from PI or it is a problem in the sender system.

Regards.

[former_member186851]

ya valid point Inaki.

Also take the data in prod and test in developement and see how it reacts for 1300 records.

[Former Member]

Hi inaki,

The above exception was encountered in the third party system.How we can analyze that the issue is with the sender system or PI?

Thanks,

--Ramya

[former_member186851]

Ramya set those parameters then try once.

[Former Member]

Hi Raghu,

The scenario was successful in UAT with 2000 records.And the timeout value is same in QA and PROD.

Since issue occurs only in production we are unable to check the issue.More over the interface retrieves the details and updates in thirdparty application.Timeout value set in adapter is 900000ms.

Basis has raised sap message and it was suggested to check the SAP NOTE 1405895 - Large messages from SOAP receiver channel fail.

Please let me know how it can be resolved.

[former_member186851]

Hello Ramya,

The settings in PI seems to be fine.

And also check in HTTP trace whether any call is hitting the PI server.

Check with sender WS Team as well once Since the request is not reaching PI at all.

[Former Member]

Hi All,

Could you please help in resolving the above issue

[former_member186851]

Hello Ramya,

Can you check the RFC program internally in SAP ECC?

[Former Member]

Hi Raghuraman,

Thirdparty team before sending to PI they are splitting the records with 500 employee Id's.

For 1st day it got failed for first set and for other sets it  worked fine and on second day for first set they had faced read time out exception and all other records got failed with socket exception.Now it is failing for the all the records with socket exception.

How to analyze RFC program from ECC.

Is there any settings to resolve the issue.

[Former Member]

Hi Raghuraman,

Our version is sap pi 7.0.Please let me know the steps to check the http trace.

Thanks

[former_member186851]

Ramya,

Did you run and check the RFC internally?

[iaki_vila]

Hi Ramya,

The above exception was encountered in the third party system.How we can analyze that the issue is with the sender system or PI?

This is not an easy answer. Talk with your basis team to know what is your infrastructure, may be the problem is in one intermediate server, like an apache, in your environment or in the third-party tool environment that interrupt the communications if there is an overload situation. From own perspective, it will be diffcult, because it will be necessary to use an sniffer to know where the message is passing.

Mealtimes you can use XPI inspector tool and to raise a SAP assistance

Regards.

[Emre_tr]

Hi,

Check the response time, it can be a timeout issue.for 1300 employee it will take some time.

[former_member186851]

Hello Ramya.

Did you move the changes to Production if any?

And try increasing the threadcount

[Former Member]

Hi Raghuraman,

The interface was moved to production in december.And there were no recent changes.

Please let me know when we can encounter such error .

Thanks,

--Ramya