0
votes

SSAS Version: 14.0.226.1

Visual Studio Version: 4.7.02558

Issue: once model is delployed to the server, it is processed w/o any errors. But if the SSAS server is rebooted, one of the dimensions throws an error while processing. It just loses one of the column. Here is the error that I get (Failed to save modifications to the server. Error returned: 'The 'Global_Code_SKU' column does not exist in the rowset.):

Error screenshot

The column data sample looks like this:

Nvarchar column with string identifiers

The model contains 2 dimensions and a fact table with 632 million rows in it. May it be that the fact table size is an issue? Maybe dictionary's too big?

How I fix it: by deploying model again without partitions and roles, just metadata, and this fixes the issue, however sometimes servers can be rebooted without notification, so the processing job fails next day (it runs once a day).

Is there any suggestion I can consider to fix this? I searched for a while, haven't found any solution though.

2

2 Answers

0
votes

Some suggestions to try:

  • After reboot, connect to the SSAS server using SSMS and right click the database in question and choose Script -> Script database as. Is the column Global_Code_SKU still there? Is it hidden? Is it available in the source?
  • What datatype is the Global_Code_SKU? I've had problems with columns with similar values being auto-identified by SSAS as binary and therefore excluded from the load.
0
votes

There was a hidden sign in right before the first symbol in one of the names, so after comparing binaries of the two strings we wound that we just should recreate the table and that solved the problem