on 01-24-2005 7:40 PM
Hello ,
a ) I was wondering if anyone could give me the exact proc: in case a delta to my ODS failed . I would like to delete the failed delta request and reload the delta which will get me the new records as well as old records .However sometimes it loses the delta pointer and makes me reload the entire data starting with an init.I was wondering if anyone could let me know the exact proc of deleting a delta from ods w/o losing the delta pointer and wht measures I might need to take everytime a load fails. Point to note is I have the ods auto activate box off too.
Tx for ur responses
madhu
What I do is the following.
1. If there is data in the error request, then correct the wrong data and reload.
2. Reinitialise delta without data transfer.
3. If there is no data in the error request, then delete that request and do step 2.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
It also depends on the data source. In the old CO-PA extraction you were able to set the time stamp independently. There I often did a new initialization and set the time stamp correctly. If possible I always try a reload. But depending on the reason for the error it isn't always possible. So you have to look at the error code.
My personal rating:
1. If data is in PSA reload from PSA
2. If not, try to reload the data
3. If it still doesn't work reinitialize the delta without data and load the difference by full upload.
4. If you cannot do 3. do a complete reinit (if possible restricted to current month, current year or whatever the extractor has for options. Delete overlapping values from fact table before you do the upload.
5. If everything fails create a flat file from R/3 edit it until it fits your current data and load it up.
Everything after 2. might be a problem, so check carefully what you're doing.
Best regards
Dirk
User | Count |
---|---|
91 | |
10 | |
10 | |
9 | |
9 | |
7 | |
6 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.