Page 1 of 1
Corruption of Aliases? :(
Posted: Fri Mar 22, 2013 6:09 am
by fleaster
Ok, so I inserted a new alias into a dimension (which already contains an existing alias), and updated the new alias to a unique name...
...now I did this for 4 dimensions - the new alias appears ok in the Cube Viewer, but when referencing the new alias via DBRW formula in XL, it returns a Key Error... anyone have any ideas how to resolve? maybe restart the service? somehow reindex the dimension...?
cheers!
Matt
Re: Corruption of Aliases? :(
Posted: Fri Mar 22, 2013 7:36 am
by lotsaram
There are some posts on this, it has certainly come up before and seems to affect many versions. To solve you force a "dimension recompile", simplest way is to add a new member. You can then go and delete the newly added member and you're back to the original dimension but with aliases working (when you have this issue you should also notice in subset editor when you have the properties window showing that all attribute values show an #ERR message). The adding and removing of a dummy element can all be automated by TI if the issue is affecting multiple dimensions, just note that you can't add and remove an element within the same process as this won't change the dimension.
Or if no too time consuming or service effecting you can restart the server which should also fix it.
Re: Corruption of Aliases? :(
Posted: Fri Mar 22, 2013 8:30 am
by Duncan P
Have you tried adding the element in the prolog and deleting it in the epilog? Crossing the metadata/data boundary normally commits the changes.
Re: Corruption of Aliases? :(
Posted: Sun Mar 24, 2013 10:05 am
by fleaster
thanks all for the tips - I found that changing the dimension sort order (auto/manual etc) also seemed to magically do something... + the add/delete element also works... =)
cheers,
Matt