Unable to Run Two Jobs At Once To Load SAP BW Target Tables

dataservices sap-erp-bw
1

When trying to start two or more jobs with SAP BW as the target, only one job will start on the Data Services Management Console.

Is the RFC server only single threaded? If so is there a work around?

DS Version: 12.2.2.1.

SAP BW version: 7.00

Thanks for your help :hb:

jsull2 (BOB member since 2009-06-04)

2

Yes, the RFC Server is single threaded. :reallymad:

It completely slipped my attention to check for that and development forgot about it as well. An emergency bug fix is currently in the makings, DS 12.2.2.3 will have a multithreaded RFC Server.

Sorry for that

Werner Daehn :de: (BOB member since 2004-12-17)

3

Thanks Werner for the information.

Do you know when the fix will be available?

jsull2 (BOB member since 2009-06-04)

4

is the answer I got

Werner Daehn :de: (BOB member since 2004-12-17)

5

Thanks again Werner for your help

jsull2 (BOB member since 2009-06-04)

6

Hi

Has 12.2.2.3 been released with the fix?

Cheers

jsull2 (BOB member since 2009-06-04)

7

yes, its avilable from download from SAP Service Market Place

manoj_d (BOB member since 2009-01-02)

8

Hello all,

I upgraded to version 12.2.2.3 to solve this issue. However, I can’t test because now I have another problem:

The Jobs can’t be started from BW. I receive the following error message:

ORA-01400: cannot insert NULL into (“BODS_ADM”."AL

I saw in RFC Log (Management Console) and the error is:

12/02/2010 10:39:38 [ SEVERE ] Could not execute the following SQL command: insert into AL_BW_REQUEST (REQUESTID,TYPE,VALUE,DF_NAME,CREATION_TIME) values(?,?,?,?,?)
12/02/2010 10:39:38 [ SEVERE ] ORA-01400: cannot insert NULL into (“BODS_ADM”.“AL_BW_REQUEST”.“VALUE”)

If I run the Jobs from Management Console or Designer, it’s fine.

Is this a bug in the version 12.2.2.3?

Regards,
Vinícius F.

ViniciusFossaluza :brazil: (BOB member since 2009-09-02)

9

This did resonate somehow, so I did a search on my emails. On 20.7.2010 I got an email saying “Before launching a job, the RFC Server will insert a record with value=NULL in the AL_BW_REQUEST table […] and when a request is received from BW, this record is updated”.

I believe, when you look at the repo scripts in your \DI\admin\repo… directory you will find that the script for this table allows NULLs on the value column. So when the repo was created, we obviously shipped an wrong version of the create script and later repo upgrades did not deal with that change.

Werner Daehn :de: (BOB member since 2004-12-17)

10

Hi Werner,

Thanks for your answer.

I looked in the \DI\admin\repo\oracle directory and two scripts have the string ‘AL_BW_REQUEST’: ora.sql and ora_upgrade1220.sql.
Both have the same creation table statement:

CREATE TABLE AL_BW_REQUEST
(
REQUESTID NVARCHAR2(64) NOT NULL,
TYPE INTEGER NOT NULL,
VALUE NVARCHAR2(255) NOT NULL,
DF_NAME NVARCHAR2(255),
CREATION_TIME DATE NOT NULL,
PRIMARY KEY (REQUESTID)
);

The column VALUE is a column NOT NULL.

So, if the statement (“Before launching a job, the RFC Server will insert a record with value=NULL in the AL_BW_REQUEST table […] and when a request is received from BW, this record is updated”.) really happens, something is wrong: or this statement or the creation table script.

Any ideas? Maybe change this column to NULL able?

Regards,
Vinícius F.

ViniciusFossaluza :brazil: (BOB member since 2009-09-02)

11

Hi Werner,

I changed the column to NULL able and my problem is solved.

Really the statement is correct. When the Job is executing, the value in the column is null. When the execution finishes, this value is updated.

I think that is a bug. The script should be changed to allow null values in this column.

Thank you,
Vinícius F.

ViniciusFossaluza :brazil: (BOB member since 2009-09-02)

12

We have a similar issue after upgrade to 12.2.2.3. The symptom in our case is when we run the infopackage, we see that the job starts, but then hangs for several minutes or sometimes several hours. When we run the InfoPackage with full trace switched on (-T-1 in advanced params), I see in the trace file:

(12.2) 01-28-11 16:13:45 (7472:7760) REPO: Saving object in the repository. The SQL generated is: .

The next line in the trace comes nearly two hours later.

(12.2) 01-28-11 18:05:01 (7472:7760) JOB: Job <JOB_XYZ> of runid <2011012816134574727760> is initiated by user .

I was thinking this maybe a locking issue specific to SQL Server, so I am thinking to change the repo to an Oracle db.

Any thoughts ?

Cheers,
Kevin.

KCALLIN (BOB member since 2010-04-27)

13

I doubt it is a locking thing in the database, we commit each update right away. Let me ask development for help…

Werner Daehn :de: (BOB member since 2004-12-17)

14

Hi Werner,

We have updated from 12.2.2.1 to 12.2.2.3 to fix this issue, but the issue remains - Jobs still queue rather than run in parallel. Does additional configuration need to happen to force multithreading?

The update was performed by uninstalling 12.2.2.1 but leaving the directory structure and config files in place, then running a full install.

Thanks,
Tom

TomA2011 :australia: (BOB member since 2011-02-07)

15

In the management console go the to RFC Server config. There you have a setting regarding the number of threads supported. I hope it is that.

Werner Daehn :de: (BOB member since 2004-12-17)

16

That option doesn’t seem to be available (see attached screenshot).

Thanks,
Tom
rfc_admin.JPG

TomA2011 :australia: (BOB member since 2011-02-07)

17

I should clarify, the jobs run in parallel when initiated from the DS console. It is only when initiated from the BW end that they queue.

There is also an intermittent error which occurs, when 1 job is initiated from DS and another concurrently from BW. In this case, the BW job succeeds but the DS one fails with this error:

Load data to BW failed: RequestID = <ZPAK_4JEG7UY6B41YY602RWVP3KAJZ>, Packet Number = <-1>.

The RFC log states that the error occurred in the source system, in Extractor.

TomA2011 :australia: (BOB member since 2011-02-07)

18

I think your DS repo is oracle. If this is the case, then download 12.2.3.0 from SAP market place. That will solve the issue.

adash (BOB member since 2009-01-11)

19

Thanks Ajit, always nice if development is answering questions which are above me level even.

Werner Daehn :de: (BOB member since 2004-12-17)

20

Adash, yes the repo is Oracle. Thanks for your help folks. I’ll try 12.2.3.0…

TomA2011 :australia: (BOB member since 2011-02-07)