Openhub DTP overall status set to red by BODS

dataservices sap-erp-bw
1

Hello All,

We are currently on:
BODS 4.0
BW 7.01 SP6

Jobs that extract data from BW Openhubs are failing in BW “sporadically”.
For a failed job the DTP for the openhub loads data successfully into the openhub table and BW sets the technical status of the DTP to green but at the last step BODS is setting it to Red status.
The last step in the DTP says “(OHSP) Data Services RFC Server xxxx at host JS3 set the user status to R.”

OSS Note "1497204 - P25: DTP: Open hub confirmation does not always work " explains this behavior. We implemented this note but the issue still persists.

I read the article in the below link :
http://wiki.sdn.sap.com/wiki/display/EIM/Openhub+Common+Questions

And that makes me think we have a configuration issue.

Here is our conifg:

We have 1 repository and 3 job servers are connected to it. The job servers are clustered. We also defined a server group in the management console with JS2 and JS3 in it.

We defined an rfc server interface with the same rfc program id that the BW openhubs are using. We did this from the management console for JS3.
However the same rfc server interface appears in the management consoles for JS1 and JS2 also.
(Same thing with server group. we defined the server group from the admin console for 1 JS and it appears in the admin consoles for all JS’s).
I guess this is because all these job servers are conected to the same repository?

In the “Select default RFC Server” we chose JS2.

Also we are using the aforementioned server group in the BODS schedules for the openhub jobs. So the openhub jobs run on either JS2 or JS3.
The jobs fail occasionally and there is no pattern. The failures are very sporadic.

Is this happening because we have multiple job servers connected to 1 repository? Please let me know what you think.

Another thing I found interesting was, in the repository table AL_RUNTIME_PARAM i see that, for all failed openhub jobs, either the entry in column DI_WS_REQID or the entry in column REQUESTID is empty.
On the other hand, for the successful jobs they are both populated.
Does that mean BW is trying to send the confirmation to BODS on one JS and BODS is looking for it on a different JS?

Any suggestions are greatly appreciated.

Sorry for the long post.

Thank you

hakuna_matata (BOB member since 2012-04-27)

2

I just got this email saying you seem to have setup one RFC Server in each DS installation but you should use only one.

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

3

Thank you so much for your reply, Werner!

It looks like currently our solution is to take one of the job servers out of the cluster and use a dedicated RFC program ID and run all the openhub jobs on that server.

is that correct?

also do you know when DI 4.1 SP1 will be available?

Thank you

hakuna_matata (BOB member since 2012-04-27)

4

No, there is no relationship between jobserver and rfc server. You can have three jobservers but just one RFC server. The entire synchronization between job and rfc server is happening via the repo, hence no problem.

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

5

Thanks, Werner!

We opened a ticket with SAP and they gave the following solution:

We implemented it and everything appears to work fine.

hakuna_matata (BOB member since 2012-04-27)