We are pulling data from BW 7.3
using BODS 4.0
We are using ABAP data flows to pull from BW as recommended
After all the development was finalized, we have given the codes to the BW folks and they activated it on their side. We have flipped the option to execute the ABAP pre-loaded and tried to run all the flows…
But we have observed a weird behaviour where the flow runs successfully some times and it fails to run sometimes
The reason why it fails is due to the column delimiter issue or the row delimiter issue …
All this is happening while the data from the open hub is written to the FILE on the BW server. How can an ABAP behave in this way ?
Can any one please provide a solution or alternate to overcome this strange behavior of ABAP generated by BODS…
That is not a ABAP problem, but a data problem. The row/column delimiters are present in the data. Since it is writing to a flat file and you read from it, there is the problem.
Lots of things to read on this forum about this problem.
Yes I understood what you said, But this is not the issue with data…
Lets say, I’m trying to pull an open hub ‘X’ with a volume of 5 million records…
The open hub does not have any special characters for sure I checked all the way from ECC and BW and the Open Hub Table…
Now when I run a flow to pull this particular data, It fails for the first time in most cases, for the second time it may load successfully with no warnings of coulumn / row delimiter , on the third run it may vary again. So I’m pretty sure that there is no Delimiter or return character in the data which would result in the same error for all the triggers. But here it loads successfully some times and it fails sometimes…cannot predict when it runs successfully or when it fails…
Observation :
–>when I looked into the file created on the BW server, it varies all the times
for example : It may strip of the first two or 3 fields for a records(mostly last record of that packet) which results in the column delimiter issue and eventually it skips one orfew records.
–> It may strip off a full record and also the 2 or 3 fields from the next record…
So the behavior is different at all the times…
Note : All these observations were made for the same load in the OPEN HUB ie the open hub was never refreshed while running flows in BODS multiple times.
Can you please help me out with a solution … Please let me know if you have any questions in particular …
I have read most of the topics related to this … But din’t find any solution
Hey …I want to confirm that this issue is resolved…the reason behind this was BW installed on UNIX and BODS installed on Windows so when we specified the shared directory approach…we need a common directory where BW(UNIX) Can write to and BODS (Windows can read from)…So for this to happen windows disc was NFS mounted on UNIX …So when I ran a BW flow the ABAP used to run on BW UNIX and write the file on the windows disc pretending it as UNIX. But there are some issues (writing speed as well as file corruption) with this kind of NFS mount. So no issues with BODS or BW its the NFS mount…Please let me know if any comments…
(BOB member since 2012-03-20)