Problems with new File Set (update 2016-12-14 - 16:30 PST)

  • [アーティクル]

In the latest push we enabled the new faster file set feature per default. Unfortunately that caused a regression that can manifest itself in one of the following ways:

  1. Your EXTRACT statement extracts 0 rows, even though there should be more data.
  2. You run into the following error message: E_RUNTIME_SYSTEM_CODEGENFAILURE -- Internal error! The SV_Extract_Combine vertex for File Set should take the if-branch above

As a short-term work-around until we release a fix, please disable the fast file set feature with the following statement in your scripts:

SET @@InternalDebug = "FileSetV2:off";.

Also, if you run into either of these issues or any other issue, feel free to reach out to us either via your Microsoft account representative or via usql at Microsoft with the job link.

I will update this blog post as we address the issue.

UPDATE 2016-12-14 - 16:30 PST: US EAST2 is now using FileSetV2 set to off as default until we can investigate and fix the issues. If you have a dependency on the improved scale and are not running into the issue above, you can explicitly set it to on with the above statement.