cube processing error - perhaps someone can explain

I have one fact table that is also a dimension table + one other dimension table.

When I deployed the cube I was getting an key attribute missing error.

I went back and did a "Process Full" on the fact diminsion and then the cube would

deploy without the error.

Should a fact dimension always be set to process full?

thanks

What could have happen here is:

You've processed your dimension earlier, and then you got some more data inserted into your table. Processing of cube will figure out that dimension is already processed and will only start processing of partitions. The partition processing will see new members that came with new data and will complain during processing.

You dont have to use ProcessFull for your dimensions to keep them in sync. You can run ProcessUpdate for dimensions. But if processing of your dimensions and the cube doesnt take long, I would say do the ProcessFull.

Here is whitepaper that gives bit more details about processing: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnsql90/html/sql2k5_asprocarch.asp

Hope that helps.

Edward.
--
This posting is provided "AS IS" with no warranties, and confers no rights.

|||

I went in and changed the process options for the dimension to 'process full'

however it always defaults back to 'process update'

is there a way to lock it on process full?

|||

Click on the "script" button in your processing dialog to get XMLA processing command generated.

You can send same XMLA command to process your dimension and not to use ProcessingDialog.

Edward.
--
This posting is provided "AS IS" with no warranties, and confers no rights.

csvde from Execute Process Task not returning groupType

When I run csvde from within an SSIS package's Execute Process Task, it does not return the groupType attribute, but when I run it directly from a command prompt, it does return that attribute. The csvde swiches are set the same in both cases:

-t 3268 -u -f allGroups.csv -d "DC=corp,DC=microsoft,DC=com" -r "(objectClass=group)" -l "groupType,mail,member"

The header line in the output file produced by the package is:

DN,member,mail,member;range=0-1499

while the header line in the file produced when csvde is run directly is:

DN,member,groupType,mail,member;range=0-1499

Has anyone else encountered this behavior?

Thanks,

Ron Rice

How do you have the Execute Process task setup?

Also, you may want to specify a full path to the -f flag.|||

Phil,

Unfortunately, making the -f switch a full path did not change the results.

I also tried changing the order of the attributes in the -l switch to "mail,groupType,member". At first I thought this had worked, because it did return the groupType attribute, but then I noticed that the "mail" attribute was not returned after making this change!

So for whatever reason, when I execute csvde from a package, the first attribute in the -l switch is not returned.

Thanks,

Ron

|||

Drop the double quotes around the -l flag parameters.

-l mail,groupType,member

|||

Phil,

Removing the double quotes from around the -l switch list of attributes did fix the problem. Thanks!

I wonder why csvde would behave differently when run in an Execute Process task versus being run directly from a command prompt. Also, I tried running it using T-SQL and xp_cmdshell, and that had the same problem. And I seem to remember running into issues concerning double quotes when I ran the bcp command from an Execute Process task, as well.

Whatever the ultimate cause of the problem, now that I know the work-around I am a happy camper!

Ron

|||

Rice31416 wrote:

Whatever the ultimate cause of the problem, now that I know the work-around I am a happy camper!

Ron

It's not really a work around. According to the csvde page, double quotes are not required. (See the examples at the bottom.)

http://technet2.microsoft.com/WindowsServer/en/Library/1050686f-3464-41af-b7e4-016ab0c4db261033.mspx