July 1, 2006 at 8:38 am
SQL 2005 sp1
When i process my cude it says the attribute key cannot be found. This happens for many rows.
My fact table and dimensions are all views. So the surrogate keys are built in the fact tables definition by joining to the dimension views on the original keys and effective dates.
This approach is a little unusual but it does allow the ods to trickle feed the cubes (held in rolap) in real time.
I dont believe that it is this approach that is causing the error but i mention it just in case you have experience wih a similair approach. this only happens with some dimension and they are all stored and populated in the same way.
The attribute key does exist in the dim tables and i can join them fine with SQL.
Why cant it find the keys when the cube processes??
Many Thanks,
Jules
July 4, 2006 at 8:00 am
This was removed by the editor as SPAM
July 4, 2006 at 12:13 pm
i also got this error msg. all u have to do is to process the database in analysis server and then u can process the cube.
hope this can help.
jomin.
April 16, 2007 at 1:27 pm
I also get this msg and tried to process SSAS DB but I have the same error in SSAS
Anybody can help me
April 17, 2007 at 4:36 am
You may find that processing the dimensions before processing the cubes will solve this.
I found that when automating the processing using an SSIS job I had to make sure all dimensions were processed before the cubes otherwise I got the same errors
April 18, 2007 at 6:03 am
ok, I found what was my problem, maybe it can help somebody else...
The problem was that I reached the limit of 64k of dimension member so some row in the fact table could not find the related dimention
hope this can help
April 18, 2007 at 8:07 am
I made more tests and discover that the problem wasn't the limit of 64k
I verified the number of the row that cause the error and it's seem to be random !!!!
Here what I did to test it...
I create a new project, add a datasource and a data source view
I add a new dimension from the only table in the data source view when I process the dimension I got the attribute key not found error on line 320 but when I try to process it again the line throwing the error isn't the same !!! but it's often between 300 and 400 don't know why.... so if I change the dsv and add TOP 200 to my query, it's always working but if I add TOP 500 it fail most of the time !!!
I really need some help on that one !
April 18, 2007 at 8:47 am
April 21, 2007 at 10:04 pm
BB,
In your last post, you daid you received the Missing Key error when processing only the dimension? Can we assume you meant you processed the cube as processing the dim by itself should never result in this error as it doesn't have the concept of a missing key (ie processing the dim is defining the keys, so none can be missing yet, this requires the processing of a partition to indicate that the partition has a key not present in the dim).
One thing you may want to check is the use of 'top' anything in processing a dimension. Unless you are using the same filtered set of dimension keys over the top of your fact table load, there's no guarantee that loading the first 500 keys in to the dimension will mean that the first (say) 10,000 rows loaded from the fact will only use keys from those first 500 dim records.
Have you tried loading the dim *without* a TOP on the view and then processing the partition? Unless you have RI issues (which is a different story) you should not have a (foreign) key loaded in the fact load that doesn't exist (as a PK) in the dimension.
Steve.
January 18, 2008 at 5:15 am
Hi guys
I am having the same error as describled above. I am getting the 'Attribute key cannot be found' error when processing the dimension only, which I understand it shouldn't be possible. It always occurs on a different row, too.
Has anyone encountered this error before, and how did you resolve it?
Thanks in advance,
Matt
Viewing 10 posts - 1 through 9 (of 9 total)
You must be logged in to reply to this topic. Login to reply