SAP Gateway issue / Gateway closed unexpectedly and delay in RFC processing -BTC WP "on Hold"

[Former Member]

HI All,

Here I am again facing a issue in one of our system . (BW)

The issue is a big slowness in RFC processing and the BTC WP are in "On Hold" status for a very long time . And more importantly we have upgraded the Kernel 742/200 but the issue was not solved.

The dev_disp is showing this error repeatedly.

[isaias_freitas]

Hello,

Those dev_disp errors indicate that the Dispatcher is losing the network connection to the SAP GUI users. This would not be (at least directly) related to the "on hold" issue.

The status "on hold" with reason "RFC" means that this work process made and RFC call, and now it is waiting for the results.

What you need to investigate is why the target system is taking so long to send the results back.

Double clicking at the work process should show you the "conversation ID".

Try searching for it at SMGW or at SM51 -> select the instance -> goto -> server name -> information -> communication table.

This should point you to the target of the call.

[Former Member]

Hi Isaias,

I understand this . But the issue with no other system but only the system error log I have provided.

The issue is gateway is unable to communicate with the dispatcher.

The dispatcher is also unable to provide any WP to gateway RFC call. So I have checked the sARFC transaction and it is showing "too few WP has been found" , though I have set this as per SAP recommended but the problem with gateway/dispatcher and WP is still unsolved.

Thanks

Biswajit

[isaias_freitas]

The "too few WP has been found" at SARFC can be a consequence of the "on hold / RFC" issue.

If you believe it is not, you need to add more dialog work processes to the instance, or reduce the value of the parameter "rdisp/rfc_min_wait_dia_wp".

For the "on hold / RFC" issue, you need to investigate at the target system...

[Former Member]

Hi Isaias,

Thanks for your reply.

But the actual problem lies in somewhere else. I wanted to find out the actual issue like RFC processing is totally affected where from SM59 no RFC connection test.

Since that time none of the gateway request was not processing. The "loopback" RFC was not processing and as well as the error message I attached which also gives you an insight that the error is between , Gateway and dispatcher .

As the gateway comm. table may be corrupted eventually and no incoming request was not processed by the gateway as in SM58 also the error show , CPIC Error and Gateway closed.

So I hope the issue is something else as irrespective of many free DIA Wp , the dispatcher is unable to give WP to gateway so somewhere the issue belongs between Disp. and Gateway .

Thanks.

Biswajit.

[isaias_freitas]

Hello Biswajit,

I do not see any issue at the screenshot.

The gateway puts the requests at the Dispatcher queue. This is the normal and expected behavior.

Regards,

Isaías

[Former Member]

Hi Isaias,

Can you please look it. You told the thing is absolutely normal. But I Hope there is something which SAP can only answer.

And also in sarfc it is showing too few WP where in system there are many WP free.

[Former Member]

I addition ,

I have checked the jco traces in the same splitted java system ,

which is havinh many error like same ,

The Kernel patch is there - 742/28 and it is connecting with ABAP which is having Kernel of 742/200 ?

Will there be any issue with that? What can be done ot the necessary changes which can not flodded the SAP abap gateway with the jstart connection with disconnecting status?

Thanks

[isaias_freitas]

Hello,

I already sent explanation about the "on hold / RFC" status.

The status "on hold" with reason "RFC" means that this work process made and RFC call, and now it is waiting for the results.

What you need to investigate is why the target system is taking so long to send the results back.

Double clicking at the work process should show you the "conversation ID".
Try searching for it at SMGW or at SM51 -> select the instance -> goto -> server name -> information -> communication table.

This should point you to the target of the call.

About SARFC, if it shows "too few dialog WPs" but you still see several free, you might consider reducing the value of the parameter "rdisp/rfc_min_wait_dia_wp". You can read the documentation of this parameter at the transaction RZ11, or at the SAP note 74141.

Regards,

Isaías

[isaias_freitas]

About the "jstart" with "disconnecting" status, if this keeps for too long, I would say there is something wrong at the Java end.

[Former Member]

Hi Isaias,

Thanks for the reply,

But I want to investigate that what the exact error on java side.

As on the trace of dev_jco_jrfc.trc  are showing the error on connecting gateway ?

Is there any parameter or recommendation for Java side for Jco RFC ?

Like parameter from NWA or any JVM related parameter for Jco connections?

Do you feel this should be Kernel bug on 742/28 which is connecting the ABAP 742/200 , Throwing some kind of issue at gateway?

Thanks

Biswajit

[isaias_freitas]

Hello Biswajit,

I am not aware of any kernel issue that would cause this. If you can update the kernel to 742/200 (jsut in case) it would be good.

About the java part, I'm sorry. I don't have much knowledge on the Java-stack...

Regards,

Isaías