SOAP to RFC Scenario Issue

[Former Member]

Hi PI Gurus,

Need your help here.

We have SOAP(Adapter) -> SAP PI -> RFC(CRM) synchronous scenario.

Earlier huge volumes of data is being flown in PI PROD system and we have encountered no issues.

Suddenly few random messages are failing in SAP PI with Synchronous Timed Out Error; No Response received.

I have observed post which all messages start failing in SAP PI , after some time messages start getting processed successfully again.

also note at CRM side, there is no processing time consumed, immediate response is sent back to PI

I have increased Max Connection to 20 in RFC receiver channel, post which encountered a JCO_ERROR_RESOURCE(JCO_ERROR_RESOURCE: Connection pool RfcRepository[RfcClient[RFC_Receiver]]-- is exhausted. The current pool size limit (max connections) is 1 connections) only in random 4 messages as below:

We have done the below to resolve this error:

1. Checked and found synchronous time out is 5 minutes
2. Increased number of threads for SOAP and RFC adapters from 5 to 12
3. Basis team increased server nodes to 5 for parallel processing

    4. Increased Max Connections to 50 in RFC receiver channel

Please let us know your views and comments on the same that if any other changes are required in SAP PI to resolve this issue?

Thanks,SJ

[iaki_vila]

Hi Sarvjeet,

In my opinion your five steps are enough, taking into account Hasan notes suggestion. Also you should consider if the RFC processing time is too long for a huge load requests and if the bottleneck will not go to ECC endpoint capacity.

Regards.

[hasan_celebi]

Hi Sarvjeet;

Check Notes;

2267584 - JCoDestination does not recover from a pool exhaustion

2005477 - Configurable SAP JCo pooling parameters

There was a program error in the SAP Java Connector internal connection pool management which caused that a waiting thread was not supplied with a free connection although it was available. This error situation could only have occurred if the respectively used destination was not configured with the property jco.destination.max_get_client_time set to 0 and furthermore regular pool exhaustion exceptions for this destination already occurred before. The first occurrence of such an exception was always correct for a given destination.

Regards

Hasan

[ashish_goel4]

Hi Sarvjeet,

Check the below document.