Anyone had success with NTEXT in a Parallel job?

[chulett]

I'm curious if anyone has been successful reading an NTEXT column via the ODBC Enterprise stage on UNIX from a SQL Server database.

I can manage this in a Server job but it throws a "metadata mismatch" warning because I have to declare the column as LongVarChar rather than the LongNVarchar it really wants. I have to because, while I can View Data just find in the Server job with it as LongNVarchar, I can't compile the dang job that way since NLS is not installed here.

I was hoping to make this a PX job and use a message handler to demote the warning but I can't even View Data from the OE stage. All I get is:

Code: Select all

Error executing View Data command:

##E TDOD 000001 09:30:45(006) <main_program>  Datatype Not Supported 

It seems to be basing that on the actual source metadata rather than whatever I tell it in the job. Don't know that much about SQL Server but CAST or CONVERT seem out of the question as the XML chunks that are stored in this "ntext" field are well over 8k. Any thoughts / ideas / workarounds? I could always stick with the Server job but want to see what my PX options are.

[suryapkakani]

Hi ,

I have a similar kind of issue, I am trying to read a field on sql server though odbc connection and trying to write to a file. The field is defined as text in source. I get the following error

main_program: Datatype Not Supported

Any Ideas, of how it can be resolved..

[suryapkakani]

I could overcome the problem with the following.

SELECT cast(value_text AS Varchar(8000)) value_text from dbo.table

Thanks for some valuable notes from the blog.

Regards,
Sunny

[chulett]

As noted, I could have overcome the problem that way as well except for the fact that we were dealing with strings much larger than 8k. Honestly, though, I don't recall what we ended up doing as this was many months and a client ago. I believe "they" ended up doing something to deliver it to us rather than making us go get it... something like that.