Why I got DATA_BUFFER_EXCEED when calling RFC_READ

问题:

The requirement in my java/groovy program is like this: Users are allowed to input table names and table fields as a query input parameters and expect to get some the query result from SAP. From the user input, I get the table name [CDPOS] and [CDHDR]. After reading the SAP documentations and googling, I found these are tables storing change document logs. But I did not find any remote call functions that can be used in java to perform this kind of queries. Then I used the deprecated RFC, "RFC_READ_TABLE" and tried to build up customized queries only depending on this RFC. However, I found if the number of desired fields I passed to this RFC are more than 2, I always got the DATA_BUFFER_EXCEED error even if I limit the max rows.

I am not authorized to be an ABAP developer in the sap system and can not add any FM to existing systems, so I can only write code to accomplish this requirement in JAVA.

Am I doing something wrong? Could you give me some hints on that issue?

Many thanks in advance!

回答1:

DATA_BUFFER_EXCEEDED only happens if the total width of the fields you want to read exceeds the width of the DATA parameter, which may vary depending on the SAP release - 512 characters for current systems. It has nothing to do with the number of rows, but the size of a single dataset. So the question is: What are the contents of the FIELDS parameter? If it's empty, this means "read all fields." CDHDR is 192 characters in width, so I'd assume that the problem is CDPOS which is 774 characters wide. The main issue would be the fields VALUE_OLD and VALUE_NEW, both 245 Characters. Even if you don't get developer access, you should prod someone to get read-only dictionary access to be able to examine the structures in detail.

Shameless plug: RCER contains a wrapper class for RFC_READ_TABLE that takes care of field handling and ensures that the total width of the selected fields is below the limit imposed by the function module.

Also be aware that these tables can be HUGE in production environments - think billions of entries. You can easily bring your database to a grinding halt by performing excessive read operations on these tables.

回答2:

There is a way around the DATA_BUFFER_EXCEED error. Although this function is not released for customer use as per SAP OSS note 382318, you can get around this issue with changes to the way you pass parameters to this function. Its not a single field that is causing your error, but if the row of data exceeds 512 bytes this error will be raised. CDPOS will have this issue for sure!

The work around if you know how to call the function using Jco and pass table parameters is to specify the exact fields you want returned. You then can keep your returned results under the 512 byte limit.

Using your example of table CDPOS, specify something like this and you should be good to go...(be careful, CDPOS can get massive! You should specify and pass a where clause!)

FIELDS = 'OBJECTCLAS'.... FIELDS = 'OBJECTID'

In Java it can be expressed as..

listParams.setValue(this.getpObjectclas(), "OBJECTCLAS");

By limiting the fields you are returning you can avoid this error.