Has anyone ever been able to restart a failed cube build using the .qyj files, or any other method?  Some of our cube builds are very large and pull millions of rows to the temp directories, running for hours.  Then when I check on the build, I can see it crashed but all of the data is already pulled.  I hate having to rerun the whole build.

We've used Transformer for over 15 years and I've never been able to figure this out.  Any help is appreciated!

I think that the actual problem is the time it takes to build a PC cube ( hours) and pull so many fact rows (millions).

Can you modify Transformer Datasource queries  to reduce the number of rows ( apply group by)?

If you haven't already, I would suggest reviewing this document: https://www.ibm.com/docs/ja/SSEP7J_11.2.0/pdf/ug_cogtr.pdf.  If you look at Chapter 7 there is a section about checkpoints that can be used to restart a cube build without starting it from the beginning. In Chapter 8, there is a section that discusses Reducing Build Times for Large PowerCubes. This document overall has a great deal of information about designing and building PowerCubes. It is a deep dive on Transformer and all the various options and settings available to maximize your ability to design, deploy, and manage PowerCubes.

Hi,

Just for my understanding. It is a problem you encounter with the last version of transformer or it is happening independently of the version ?

I had some issues with existing models when migrating to 11.2.4 initials versions and IBM provided one version that worked fine.  (transformer-11.2.4-24022410-win32).

Have you tried transforming your .qyj to .mdl. ? just in order to open the model and reduce data scope.

Best regards