Cube Deployment Status Confusion!

[jv_oz]

Hi all,

I have a new client with and existing environment that includes Dev, Test, Staging and Prod.  Transformer is installed in all environments. When I go to create a cube in Dev I get a message saying "Cube created successfully" and then a deployment message saying "Powercube has been deployed successfully".  Then if I click on details, I get "powercube 'test cube' cannot be deployed because directory 'D:\Powerplay Cubes\Public cubes\test_cube_v2' already exists".

Next, under the Data Source tab in Transformer there is a UNC path to the same location.

If I then go the deployment location ('D:\Powerplay Cubes\Public cubes\test_cube_v2'), the cube does not have today's date, however in the cube build location (d:\bi data\powercubes\) it is there with the correct date and time.

So I can see that the cube has been created successfully, but I am confused by the message "Powercube has been deployed successfully". Has it really when the cube at the deployment location is out of date?

Thanks

John

[bdbits]

When Transformer builds the cube, it will create the output cube file using what you have as the "Powercube file name" on the Output tab. I am guessing that in your case, this is "d:\bi data\powercubes".

The deployment step is separate. Based on what you have specified on the Deployment tab, it will copy the file created above to the specified deployment locations and activate the new cube files. You will also want to check "Enable automatic Powercube deletion" and specify how many versions you want to keep (in case you need to roll back). It looks like in your case, the deployment location is "D:\Powerplay Cubes\Public cubes\test_cube_v2". You might expect to see an .mdc file in this location. But, if you have versions enabled on the deployment tab, the contents in the deployment location are a little different. You will see folders like "my_cube_name_yyyymmddhhmmss" where the last portion is a datetimestamp string. You will see the normal "my_cube_name.mdc" file but it is actually nearly empty. There will also be a "my_cube_name_yyyymmddhhmmss.ver" file. What happens is that Cognos will see the "stub" .mdc file and know it is not the cube data. Then it finds the .ver file, and matches that up with the folder having the same name as the base of the .ver file. Inside there are the actual mdc files making up the current version of the cube. (Note that if you ever want to roll back, you rename the .ver file to match the folder you want to be used.)

The Data Source tab is used to set the data source properties you would see in administration. When you "publish" the cube, it will set the data source properties to what you have specified. The location here is the file the cube data source will use. Normally, you are going to want this to be the same as the deployment file location. Actually, I cannot imagine a scenario where you would not have it that way.

I hope this long-winded explanation has helped clarify how this stuff works.

If the deployment said it was successful, it is possible that the message about the directory is informational. Using the above information, maybe the file at the deployment location really was updated but you were looking in the wrong place. I hope so.  :)

[jv_oz]

Great explanation @bdbits, thank you!